The literature

Note, September 2026.  The parent article is now an overview.  The full treatment of each topic, and the sources specific to it, moved to thirteen child articles, each with its own literature review:  Miller-Urey, Making Nucleotides, Homochirality, The RNA World, Metabolism First, Protocells, Is the Genetic Code Optimized?, The Ribosome, Biological Information, The Minimal Cell, LUCA, The Earliest Evidence for Life, Panspermia.  This page remains the union list;  a source marked here as used may now be cited only in a child.

This page exists so that nobody has to take the rest of the article on trust.  Every source it rests on is listed here with what it actually measured, who wrote it, whether anybody here read it, and which way it cuts.

A catalogue of the literature on the origin of life — the early atmosphere, monomer synthesis, nucleotides, homochirality, the RNA world, ribozymes, metabolism-first and hydrothermal vents, protocells and membranes, the genetic code and the ribosome, sequence space and functional information, the minimal cell, LUCA, the timing of the earliest life, extraterrestrial delivery, panspermia, and the philosophical objections.

It is deliberately not a reading list of things that agree with this site.  Sources are tagged by the direction they cut, and a reader who wants only the strongest material against the argument can filter for supports and read those.  Several of the most damaging entries were written by creationists and by the Intelligent Design movement's own authors, and several of the most useful concessions were written by committed naturalists.

How to read an entry

What it measured, versus what it claimed.  The most important distinction on this page.  A ribose decomposition rate measured at four temperatures is a number with error bars.  "The odds of life forming by chance are one in 10^40,000" is a rhetorical figure whose inputs nobody can check, and which in that particular case traces to a newspaper education supplement.  Both appear in this literature, often on the same page, and conflating them is how most bad arguments on this subject get made — on both sides.  Every entry states which kind of thing it is.

Whether anybody read it.  This is the page's whole value, and it is unusual.  Origin-of-life research is scattered across publishers that block automated access, so a great deal of what circulates as a citation is really a citation of somebody else's summary.  Every entry here says which of four things happened:  the source was opened in full; only its abstract was verified, verbatim, from a named index; only part of it was opened — supplementary information, a free preview, an author manuscript; or it was never opened at all and is known only second-hand.  One central source, Axe 2004, falls in the last category and its entry says so at length.

Direction.  Whether the source supports, undercuts, or complicates the case that the origin of life is a tractable naturalistic problem.  Assigned by what the source shows, not by what its author believes.  Leslie Orgel's demolition of metabolism-first is tagged undercuts and Orgel was a committed naturalist arguing for a different naturalistic route.  Answers in Genesis' own "Arguments to Avoid" index is tagged undercuts against the apologetics case, and it was written by young-earth creationists.  Sources that do both are tagged complicates.

Camp.  Disclosure, not a verdict.  A creationist organisation retiring one of its own arguments is worth more, not less, for having been written by creationists, and the tag exists so a reader can weigh it rather than being told what to think.

Post-cutoff sources.  Several of the most important results in this field are very recent — the QT45 self-copying ribozyme (February 2026), the Miyachi & Ichihashi error-minimisation experiment (2026), the Bennu sample-return chemistry (2025-2026), and the Chang'e-6 lunar chronology (2026).  These postdate most readers' sense of the field and postdate a good deal of the secondary literature that summarises it.  Every such entry is flagged post-cutoff.  Anything written about the origin of life before 2025 should be assumed not to know about them.

Field definitions

The sources that failed verification

Read this section first.  It is the most useful part of the page for anyone who intends to write about this subject, and it is the reason the rest of the page can be trusted.

Origin-of-life apologetics runs on quotations, and a large fraction of the quotations in circulation do not survive contact with the source they name.  Twelve of roughly twenty quotes in this project's own long-standing notes file failed.  The full list, with the defect named in each case, is on the companion page — what the researchers actually said, section C.  Four failures are worth summarising here because they are load-bearing citations elsewhere on the web.

Orgel's "near miracle" — misattributed, and it says the opposite of what it is used for

Universally cited to Orgel's 2004 Critical Reviews in Biochemistry and Molecular Biology review.  The 2004 paper was opened in full, all twenty-six pages, and searched: the word "miracle" occurs zero times.  The sentence is Joyce and Orgel's, from a chapter in The RNA World, and it describes what one of two groups of scientists believes rather than the authors' own verdict.  The sentence immediately after it is "Time will tell which is correct."

The underlying book chapter could not be opened — the Cold Spring Harbor Monograph Archive is purchase-walled and archive.org's search-inside API did not resolve from this environment — so the wording is verified only from its republication in Robertson & Joyce 2012, p. 15, which was opened.  Cite that, and never Orgel 2004.

Hoyle's 747 — the standard citation is to a news story about Albania

The tornado-in-a-junkyard line is nearly always cited to Fred Hoyle, "Hoyle on evolution," Nature 294 (12 November 1981), p. 105.  The table of contents for that issue was fetched and read item by item.  There is no article by Hoyle in it.  Page 105 of Nature 294 is a news story by Vera Rich titled "Albanian development: Hoxha looks ahead."

The companion figure — "a one followed by 40,000 noughts" — is worse.  Creation Ministries International's own page on it does not cite Hoyle; it cites Lee Elliot Major in a Guardian education supplement of 23 August 2001, the month Hoyle died.  A newspaper piece is being used as the source for a calculation supposedly made twenty years earlier.  Neither the citation nor the number may be used.  Hoyle did publish the argument himself, in plain prose, in his 1981 Caltech lecture, and that source is open, verified and catalogued below.

Stanley Miller, Discover, November 1992 — never reached

The source URL is dead and the article text was never obtained.  Every surviving rendering is on an apologetics site and they are word-for-word identical, which is the signature of a laundered quote.  The passage is also the journalist's narration with Miller's words embedded, so the quotation marks are misplaced in most versions.  The variant "the origin of life has turned out to be much more difficult than I, and most other people, envisioned" is cited to a Google Books search hit, which establishes that some book contains the string and nothing else.

Gutfreund & Toulouse, Biology and Computation, p. 75 — misattributed by construction

Biology and Computation: A Physicist's Choice is an anthology of reprinted papers.  A sentence on page 75 was written by whoever wrote the paper printed on page 75, not by the editors whose names are on the spine.  The sentence is also the standard set-up for arguing that protein evolution is not a random search — it states the combinatorial figure in order to reject the random-search model.  Quoting it against evolution reverses its function.

The rest

Also failed:  Harold Urey in the Christian Science Monitor, 4 January 1962 (never located, and Urey's own 1959 Science paper argues the opposite);  A. G. Cairns-Smith's "no obvious way at all" (no book, page or year given, and Cairns-Smith spent his career proposing a naturalistic answer to the gap he was describing);  Stuart Kauffman's "a fool or a knave" and his 10^40,000 (Google Books queries, and Kauffman's point is that the improbability is why he rejects naked replicating RNA);  Hubert Yockey 1981 (a real paper, but Yockey was not a creationist, argued the origin of life is axiomatically outside science's reach, was equally scathing about creationism, and wrote before ribozymes were discovered);  and Nick Lane on the RNA world (begins mid-passage;  the preceding sentences are "I don't think there can be much doubt that RNA was a major central player… So the RNA world is absolutely correct, as far as I'm concerned").

Two standing rules came out of this.  A Google Books search URL is a query, not a citation.  And a quotation whose only reachable source is an apologetics site does not run.

The early atmosphere, and the Miller-Urey experiment

The standard apologetics objection is that Miller used an atmosphere nobody believes in any more.  The first half of that is right and the second half is not.  The geology says the bulk Hadean atmosphere was not the methane-ammonia-hydrogen mixture Miller sparked.  The chemistry says it did not need to be.

A Production of Amino Acids Under Possible Primitive Earth Conditions

Miller, Stanley L.  Science 117 (1953): 528-529.

area: atmosphere camp: mainstream-chemistry type: experiment access: paywalled direction: supports

JSTOR scan on disk, opened in full.  The famous one, and smaller than its reputation.  A CH₄/NH₃/H₂/H₂O spark discharge produced three amino acids identified confidently — glycine, α-alanine, β-alanine — plus two tentative, in the "milligram range."  Only two of them are used in proteins.

Miller's own framing is worth having, because it answers a charge often made against him:  "an attempt was made to duplicate a primitive atmosphere… not to obtain optimum conditions."  He also noted colloidal silica dissolving out of the glass into solution, which turns out to matter sixty-eight years later.

Organic Compound Synthesis on the Primitive Earth

Miller, Stanley L., and Harold C. Urey.  Science 130 (1959): 245-251.

area: atmosphere camp: mainstream-chemistry type: experiment access: paywalled direction: complicates

Scan on disk, read in full including Table 2.  The yields, which are usually quoted only in the direction that suits the quoter.  From 710 mg of carbon as methane:  glycine 63 × 10^-5 mol, glycolic acid 56, alanine 34, lactic acid 31, aspartic acid 0.4, glutamic acid 0.6.

The number that matters is in their own summary:  the identified compounds "account for 15% of the carbon added as methane."  Glycine alone is 2.1%.  The other 85% went to polymeric material on the electrodes and walls, and as far as this research could establish, nobody has ever characterised it at the compound level in any Miller-type run.

The Miller Volcanic Spark Discharge Experiment

Johnson, Adam P., et al.  Science 322 (2008): 404.

area: atmosphere camp: mainstream-chemistry type: experiment access: paywalled direction: supports

Opened in full including the Supporting Online Material.  Miller's own archived 1953-54 vials from a second, volcanic-plume apparatus, re-analysed with modern instruments:  22 amino acids and 5 amines, at yields "comparable to, and in some cases exceed[ing]" the classic runs.  Products racemic within measurement precision.

Two honest limits.  It is a diversity result, not a yield result — the authors state "we were not able to calculate actual yields."  And the gas mixture is still reducing, so it relocates the chemistry to a volcanic plume rather than defending a globally reducing atmosphere.  Table S1, which carries the per-compound abundances, is an image in the SOM and did not survive text extraction;  it was not read.

Primordial synthesis of amines and amino acids in a 1958 Miller H₂S-rich spark discharge experiment

Parker, Eric T., et al.  PNAS 108 (2011): 5526-5531.

area: atmosphere camp: mainstream-chemistry type: experiment access: OA direction: supports

Opened in full.  More archived vials, this time with hydrogen sulfide in the mix:  23 amino acids and 4 amines including seven organosulfur compounds, and "the relative yield of some amino acids… [is the] highest ever found in a spark discharge experiment."

A detail worth keeping, because it shows how contamination is actually detected.  The abundant chiral amino acids came out racemic.  Three low-abundance ones — aspartic acid, glutamic acid, serine — showed an L-enrichment, and the authors flag it themselves as probable trace contamination.

A Reassessment of Prebiotic Organic Synthesis in Neutral Planetary Atmospheres

Cleaves, H. James, et al.  Origins of Life and Evolution of Biospheres 38 (2008): 105-115.

area: atmosphere camp: mainstream-chemistry type: experiment access: paywalled direction: supports

Opened in full, including Tables 1 and 2.  This is the paper that answers the objection, and apologetics almost never cites it.

A neutral CO₂/N₂ atmosphere gives essentially nothing — under 10^-6 to 0.013% — because nitrite and nitrate produced in the spark destroy the products during workup.  Buffer with calcium carbonate and add an antioxidant, and the amino acid yield rises to 2.5%.  Even an O₂-containing mixture gave 2 × 10^-4%, racemic:  not zero.  And unlike the reducing runs, the neutral runs produced no tar.

Three caveats, all from the paper.  The neutral and reducing yields are computed on different denominators — nitrogen versus carbon — so they cannot be directly ratioed against the ~5.5% usually quoted for reducing mixtures.  Ascorbate, which gave the roughly hundredfold rescue, is "an unlikely prebiotic anti-oxidant";  of the plausible natural alternatives tested only pyrite (10×) and ferrous sulfate (2×) worked.  And the paper names its target:  Jonathan Wells's Icons of Evolution is cited in the introduction as the objection being rebutted.

The role of borosilicate glass in Miller-Urey experiment

Criado-Reyes, Joaquín, et al.  Scientific Reports 11 (2021): 21009.

area: atmosphere camp: mainstream-chemistry type: experiment access: OA direction: complicates

CC BY, opened in full.  A genuinely surprising result that both sides should find interesting.  Running the same experiment in borosilicate versus Teflon:  48 compounds versus 31, and seventeen compounds including five amino acids and one dipeptide formed only when glass was present.  Total product yield 393.82 µg/mg crude in borosilicate against 118.28 in Teflon, with Teflon-plus-glass-chips intermediate at 239.41.

The authors turn the apparent artefact into a finding, and their sentence is the quotable one:  "Miller recreated… the atmosphere and waters of the primitive Earth.  The role of the rocks was hidden in the walls of the reactors."  Silicate surfaces are not an artefact on a planet made of silicate rock.

Chemical Analysis of a "Miller-Type" Complex Prebiotic Broth, Part I

Wollrab, Eva, et al.  Origins of Life and Evolution of Biospheres 46 (2016): 149-169.

area: monomers camp: mainstream-chemistry type: measurement access: paywalled direction: undercuts

Opened in full.  The strongest single citation on the problem side of this section, and it is a measurement rather than an argument.

FT-ICR mass spectrometry on a five-day CH₄:NH₃:H₂ discharge run found 668 distinct substances in one narrow mass window alone (m/z 150-300).  Their sentence:  "We cannot identify any preferred reaction product."  Average composition of the molecules found:  34.89 ± 6.68% C, 37.25 ± 9.82% H, 18.86 ± 9.77% N, 9.04 ± 7.20% O.

The authors read the diversity optimistically.  It is also, precisely, what Benner's asphalt paradox describes.  Compare Lazcano and Bada's 2003 fiftieth-anniversary review (opened in full), whose claim that "a relatively small number of compounds were produced in surprisingly high yields" rested on 1950s chromatography and does not survive better instruments — except for the amino acid subset specifically, which does show a skewed distribution.

The oxidation state of Hadean magmas · Oxidized Hadean magmas and Archean mobile-lid tectonics revealed by Jack Hills zircon

Trail, Dustin, E. Bruce Watson, and Nicholas D. Tailby.  Nature 480 (2011): 79-82.  ·  and  ·  Houchin, K., et al.  PNAS 123 (2026): e2525466123.

area: atmosphere camp: mainstream-geology type: measurement access: abstract-only direction: complicates

Both abstract only.  Trail's abstract came verbatim from Europe PMC;  Houchin's from the Crossref JATS record, because PNAS bot-blocks and it is not in PMC.  Neither body was read, and Trail's cerium-in-zircon calibration has a critical literature this research did not survey.

The geological premise the apologetics objection gets right.  Cerium-in-zircon oxybarometry on ~4,350 Ma Jack Hills zircons puts Hadean magmas near the fayalite-magnetite-quartz buffer, "similar to present-day conditions," so outgassing "would not have resulted in a reducing atmosphere."  Houchin et al. 2026 — post-cutoff — reach the same conclusion by a different method, uranium XANES, giving FMQ−1 to +1 and "near-modern redox states by 4.15 Ga."

Two independent methods agreeing on the same zircons is stronger evidence than either alone, and the article should concede the point plainly.

The Archean atmosphere

Catling, David C., and Kevin J. Zahnle.  Science Advances 6 (2020): eaax1420.

area: atmosphere camp: mainstream-geology type: review access: OA direction: complicates

CC BY, opened in full.  The best-constrained numbers in the field, and the ones that kill the "O₂-rich early atmosphere" claim outright.  Surface O₂ below 10^-6 of present — that is under 0.2 parts per million by volume — from mass-independent sulfur isotope fractionation.  N₂ similar to today or a few times lower;  CO₂ ten to 2,500 times modern;  CH₄ a hundred to ten thousand times modern.

Their definition of "weakly reducing" is the one to use:  minor CO, H₂ and CH₄ inside an anoxic bulk CO₂/N₂ atmosphere.

Creation and Evolution of Impact-generated Reduced Atmospheres of Early Earth · Origin-of-life Molecules in the Atmosphere after Big Impacts

Zahnle, Kevin J., et al.  Planetary Science Journal 1 (2020): 11.  ·  and  ·  Wogan, Nicholas F., et al.  Planetary Science Journal 4 (2023): 169.

area: atmosphere camp: mainstream-geology type: calculation access: OA direction: complicates

Both opened in full;  both CC BY or arXiv-open.  The current mainstream escape route, and it comes with a self-inflicted problem the authors state themselves.

Iron-rich late-veneer impactors reacting with ocean water make transient H₂-CH₄-NH₃ atmospheres lasting anywhere from under a million to a hundred million years.  Wogan et al. compute hydrogen cyanide rainout up to 10⁹ molecules cm^-2 s^-1 when CH₄/CO₂ exceeds 0.1, and put the minimum impactor at 4 × 10²⁰ to 5 × 10²¹ kg — 570 to 1,330 km across.  The Hadean saw four to seven of the smaller and zero to two of the larger.

The problem is heat.  The biggest impacts bring the most reducing power and leave the surface above 400 K under a hydrogen greenhouse.  Wogan et al. note that "nucleotide building blocks are fragile" above 360 K, and the proposed rescue — stockpiling HCN as ferrocyanide — needs 273-298 K, far colder than the regime that generates the nitriles.

A 2024 follow-up (abstract only, Liebert bot-blocks) puts median impact-driven biopoiesis at ~4.35 Ga and gives a 90% chance of favourable conditions on optimistic assumptions and ~20% on pessimistic ones.  Both figures are conditional on assuming prebiotic chemistry needed a post-impact reducing atmosphere;  they are impact statistics, not origin-of-life probabilities.

Spraying of water microdroplets forms luminescence and causes chemical reactions in surrounding gas

Meng, Yifan, Yu Xia, Jinheng Xu, and Richard N. Zare.  Science Advances 11 (2025): eadt8979.

area: atmosphere camp: mainstream-chemistry type: experiment access: OA direction: complicates

CC BY-NC, opened in full.  "Microlightning" — water microdroplet spray generating discharge without an external voltage, sufficient to make HCN, glycine, urea and uracil.

The press coverage blurred two different objections and the article should not.  This answers the claim that lightning was too rare to matter.  It does not answer the atmosphere-composition objection, because the gas mixture still contains methane and ammonia.  There is also a published commentary on this paper (Faubel, Natural Sciences, July 2025) which could not be opened — Wiley bot-blocks and Crossref carries no abstract — so do not present the microlightning result as uncontested.

Concomitant formation of protocells and prebiotic compounds under a plausible early Earth atmosphere

Jenewein, Christian, et al.  PNAS 122, no. 2 (2025): e2413816122.

area: atmosphere camp: mainstream-chemistry type: experiment access: OA direction: complicates

Opened in full via PMC.  Do not quote numbers from this paper.  It was formally corrected twice in 2025, in PNAS 122(9) and 122(14), and the corrections were not read.

The claim is that silica, and only with water present, induces a solid organic film of HCN polymers containing hollow "biomorphic protocells" — which are poly-HCN blobs, not membranes.  The authors' own warning is the useful sentence:  the coexistence of building blocks and biomorphic structures "does not necessarily mean evidence of life."

Why the Miller-Urey research argues against abiogenesis

Bergman, Jerry.  Journal of Creation 18, no. 2 (2004): 28-36.  Reprinted by Answers in Genesis.

area: apologetics camp: creationist type: web-page access: OA direction: undercuts

Opened in full including the reference list, and live at HTTP 200 on 2 August 2026.  The single worst item found in the apologetics corpus, and the article is stronger for naming it.

The paper tells its readers that the early atmosphere "was probably O₂-rich with some nitrogen."  Against Catling and Zahnle's O₂ ceiling of 0.2 ppmv, that is wrong by six orders of magnitude.  The footnote attached to the claim is a companion book to a 1998 PBS television series.

One honesty note about that finding.  The attribution depends on aligning Bergman's superscripts with his reference list, and his own internal cross-references are inconsistent — he cites "Lahav, ref. 21" when Lahav is item 26.  The specific attribution should be double-checked against the print Journal of Creation.  The conclusion does not depend on it:  every candidate source in that citation range is popular-level rather than geochemical.

Bergman's other charges — tar, racemic products, an incomplete amino acid set — are substantially true and are catalogued above under Miller & Urey 1959 and Wollrab 2016.

Also in the audit and not obtained:  creation.com's "The Miller-Urey experiment revisited" returned HTTP 403 to every tool tried on the atmosphere run, though creation.com served other pages normally to an ordinary desktop user-agent later the same day — the block is intermittent.  And Wells's Icons of Evolution was never obtained;  it appears here only as the work Cleaves et al. name.

Recent atmosphere work

Three 2025-2026 items round out the picture, and two of the three are post-cutoff.

Paschek, K., et al., arXiv:2503.15479 (2025), accepted in ApJ — preprint on disk, abstract and introduction read.  Serpentinization alone can reduce a 90% CO₂ atmosphere, giving warm-little-pond concentrations an order of magnitude below the impact case but still "sufficient for nucleotide synthesis."  A model, not a measurement, and it removes the impact dependency.

Moldogazieva, N. T., et al., Communications Chemistry 9 (2026) — downloaded, abstract read verbatim, body skimmed.  Post-cutoff.  The field's current consensus statement:  the Hadean atmosphere was "weakly oxidizing, but transiently reducing," life emerged across multiple local environments, and organics moved between them.

Ohmoto, H., and J. G. Ferry, Scientific Reports 16 (2026) — CC BY-NC-ND, opened in full.  Post-cutoff, and handle with care.  A minority dissent arguing from thermodynamic modelling that the atmosphere really was CH₄-NH₃-H₂-rich and that "the Oparin-Urey-Miller model… is valid;  the current paradigm… is invalid."  Ohmoto has a decades-long record of arguing against the standard oxygenation narrative, this is modelling rather than new proxy data, and the paper's own text concedes it contradicts current paradigms.  The article should not lean on it, and citing it while citing Catling and Zahnle would be incoherent.

Monomers, sugars and nucleotides

"Nobody has ever made a nucleotide" was true for fifty years and stopped being true in 2009.  What replaced it is a narrower and more interesting set of problems:  the routes that work bypass the classical difficulties rather than solving them, and they require a specific ordered sequence of conditions that somebody has to supply.

Rates of decomposition of ribose and other sugars

Larralde, Rosa, Michael P. Robertson, and Stanley L. Miller.  PNAS 92 (1995): 8158-8160.

area: nucleotides camp: mainstream-chemistry type: measurement access: OA direction: undercuts

Opened in full.  A measurement, not an argument, and the strongest thing on the problem side of this whole section — published by Stanley Miller's own laboratory.

Ribose half-life at pH 7:  73 minutes at 100 °C, about 300 days at 25 °C, 44 years at 0 °C.  Other sugars are within an order of magnitude.  Phosphorylation does not help at neutral pH:  ribose-5-phosphate has a half-life of 7 minutes, twelve times faster than ribose;  ribose-2,4-bisphosphate 31 minutes.

Their conclusion:  "stability considerations preclude the use of ribose and other sugars as prebiotic reagents except under very special conditions."  Real-world decomposition is probably faster still — ocean pH 8.2 raises the rate by half, and the Maillard and Kiliani reactions consume sugars further.

Borate Minerals Stabilize Ribose · Borate Minerals and Origin of the RNA World

Ricardo, A., M. A. Carrigan, A. N. Olcott, and S. A. Benner.  Science 303 (2004): 196.  ·  and  ·  Grew, Edward S., Jeffrey L. Bada, and Robert M. Hazen.  Origins of Life and Evolution of Biospheres 41 (2011): 307-316.

area: nucleotides camp: mainstream-chemistry type: experiment access: paywalled direction: complicates

The Brevia is a single page and was opened in full, as a page-1 image captured via the Wayback Machine and transcribed at magnification;  the Supporting Online Material, which carries the actual stability curves in Figures S1-S7, was not obtained.  Grew et al. is abstract only, verbatim from Europe PMC.

Borate takes ribose from minutes to days at room temperature, and a formose mixture stays clear for two months instead of browning within the hour.  Three caveats, all in the same one-page paper.  Borate protects arabinose, lyxose and xylose about equally, so it does nothing for selectivity.  The reaction ran in a lime slurry at about pH 12.  And borate had to be chemically removed before the products could be seen, which is an analytical step, not a prebiotic one.  The authors' own closing line is a double negative:  they "cannot exclude the availability of ribose formed prebiotically."

Grew, Bada and Hazen supply the geological objection, and it is the more useful citation:  concentrating boron enough to precipitate borate evaporites requires granitic continental crust, whether the Hadean had that crust is unresolved, and the oldest dated boron minerals are Isua tourmalines at ~3,650-3,600 Ma.  Benner himself conceded in 2020 that borate and sulfite stabilisation "is not absolute."

Rethinking early Earth phosphorus geochemistry

Pasek, Matthew A.  PNAS 105 (2008): 853-858.

area: nucleotides camp: mainstream-geology type: measurement access: OA direction: supports

Opened in full.  Concede this one cleanly.  Schreibersite, an iron-nickel meteorite mineral, oxidises in water on timescales of one to ten thousand years, producing phosphite;  more than half of the resulting aqueous phosphorus is in the reduced form, and the modelled steady-state ocean concentration is 10^-5 to 10^-2 M.  The "phosphate was locked up in insoluble minerals" objection does not survive this.

The concentration figure is a model estimate for a post-bombardment ocean, not a measurement of an ancient one.

Synthesis of activated pyrimidine ribonucleotides in prebiotically plausible conditions

Powner, Matthew W., Béatrice Gerland, and John D. Sutherland.  Nature 459 (14 May 2009): 239-242.

area: nucleotides camp: mainstream-chemistry type: experiment access: paywalled direction: supports

The paper that killed "nobody has ever made a nucleotide," and the largest single access gap in this project.

Access, stated exactly.  The main text was never obtained.  The Supplementary Information was opened in full and every experimental condition quoted below comes from it.  The abstract was verified verbatim twice, from two independent sources that agree word for word — the Europe PMC record for the DOI, and an archived 2009 nature.com article page.  Somebody with institutional Nature access should pull the main text, because its own yield summary and Figure 3 discussion have not been checked.

What it did.  Built β-ribocytidine-2′,3′-cyclic phosphate, and by further UV irradiation the uridine counterpart, from cyanamide, cyanoacetylene, glycolaldehyde, glyceraldehyde and inorganic phosphate.  It never passes through free ribose or a free nucleobase at all, which is why the classical formose, selectivity and glycosylation objections do not apply — they are bypassed rather than solved.

What the SI shows about the cost.  The one-pot aminooxazoline step gave eight products at 70% conversion, with arabinose-aminooxazoline — the productive one — at 15%.  Phosphorylation ran two days drying at 40 °C, then 24 hours at 100 °C on glass fibre discs, for 14-32% product.  UV irradiation at 254 nm ran up to 82 hours.

What it does not do.  No purines.  No handedness — Gerald Joyce, on the record in the New York Times the same week, said "Sutherland's reactions do not supply any such explanation."  And it rules out deep-sea vents, because it needs ultraviolet light.

Prebiotic chemistry: a new modus operandi

Powner, Matthew W., and John D. Sutherland.  Philosophical Transactions of the Royal Society B 366 (2011): 2870-2877.

area: nucleotides camp: mainstream-chemistry type: philosophy-paper access: paywalled direction: supports

Opened in full.  Sutherland's own answer to the "you designed the sequence" objection, and it deserves to be quoted rather than dismissed.

"One is not seeking to recreate exactly what happened on early Earth, but to show what is plausible."  On the ordering charge specifically:  the sequence of conditions "was not arbitrarily selected… [it was] empirically realized by the chemistry."  He proposes triangulation as a method, and names a specific geological setting — schreibersite phosphate plus Kölbel-Engelhardt alkanols after an impact.  And he concedes the corollary that his own chemistry imposes:  "deep-sea vents do not allow drying down, dry state heating or UV irradiation."

Note his verbs.  He uses "could have" three times in three sentences describing the scenario.  That is the correct modality and the article should match it.

Prebiotic synthesis of simple sugars by photoredox systems chemistry · Common origins of RNA, protein and lipid precursors in a cyanosulfidic protometabolism

Ritson, Dougal, and John D. Sutherland.  Nature Chemistry 4 (2012): 895-899.  ·  and  ·  Patel, Bhavesh H., Claudia Percivalle, Dougal J. Ritson, Colm D. Duffy, and John D. Sutherland.  Nature Chemistry 7 (2015): 301-307.

area: nucleotides camp: mainstream-chemistry type: experiment access: paywalled direction: supports

Both opened in full as author manuscripts.  Ritson and Sutherland make glycolaldehyde and glyceraldehyde from HCN under UV with copper cyanide complexes, bypassing the formose reaction entirely.  Patel et al. extend the same network to precursors of RNA, amino acids and lipids from one chemistry.

The price is environmental, and it rises with each paper.  The whole network needs ultraviolet light, hydrogen sulfide as reductant, and Cu(I)/Cu(II) photoredox cycling.  A separate caution:  the "lipid precursor" in Patel et al. is glycerol phosphate — the headgroup only.  The network does not make acyl chains, and it is often reported as though it solved the whole lipid problem.

Prebiotic stereoselective synthesis of purine and noncanonical pyrimidine nucleotide · Abiotic synthesis of purine and pyrimidine ribonucleosides in aqueous microdroplets

Kim, Hyo-Joong, and Steven A. Benner.  PNAS 114 (2017): 11315-11320.  ·  and  ·  Nam, Inho, Hong Gil Nam, and Richard N. Zare.  PNAS 115 (2018): 36-40.

area: nucleotides camp: mainstream-chemistry type: experiment access: OA direction: complicates

Both opened in full.  Kim and Benner couple cyclic carbohydrate phosphates directly to free nucleobases, stereoselectively for the β-anomer and regioselectively at N-9 for adenine.  The motive is an internal critique:  they wanted to avoid Sutherland's cyanamide and cyanoacetylene feedstocks.  The cost is that they must assume phosphorylated carbohydrates, and one product is zebularine, which is not an RNA base.

Nam, Nam and Zare get uridine at 2.5%, adenosine at 2.5%, cytidine at 0.7% and inosine at 1.7% in microdroplets under 1.3 µm, in about fifty microseconds of flight — a partial answer to the water paradox.  They start from supplied D-ribose, substitute inosine for guanosine because guanine is too insoluble, and electrospray at +5 kV into a mass spectrometer.  The authors argue the interface rather than the voltage does the chemistry.

Prebiotic chemistry and human intervention

Richert, Clemens.  Nature Communications 9 (2018): 5177.

area: nucleotides camp: mainstream-chemistry type: philosophy-paper access: OA direction: undercuts

CC BY, opened in full.  The "chemist's hand" objection, stated from inside a prebiotic chemistry laboratory.  Richert's proposal is concrete:  "just state the number of discontinuities in the experimental conditions or human interventions!"  He judges the required sequence of conditions more important than the source of the reagents, and sets a permissive standard — swings from −20 to 100 °C in water are fine;  arctic to volcanic within hours is not.  He concludes some intervention "may always be required" for endergonic multistep processes.

Two warnings about how this source is used.  It is a three-page Comment, not a study.  It does not tabulate anyone's interventions, despite naming Powner 2009 and Becker 2016 as examples;  citing it as a study that counted them is a misreport.  And Richert states in the same piece that "most of us are not comfortable with the idea of divine intervention in this context," so he is not available as a witness for a design conclusion.

Its deliberately paired companion, published four articles later in the same issue — Benner, Nature Communications 9 (2018): 5173, opened in full — argues that plausibility comes from interconnections across independently grounded models, and traces the word "plausible" back to Orgel and Lohrman in 1974.

Unified prebiotically plausible synthesis of pyrimidine and purine RNA ribonucleotides

Becker, Sidney, et al.  Science 366 (2019): 76-82.

area: nucleotides camp: mainstream-chemistry type: experiment access: paywalled direction: complicates

Author-accepted manuscript opened in full from the LMU repository.  (Note a title discrepancy:  the manuscript reads "purine RNA building blocks," the published version "ribonucleotides.")

Wet-dry cycling gives 85% combined nucleoside yield and 95% ribosylation with boric acid at 95 °C, with selective 5′-mono- and di-phosphates via lüneburgite and struvite.  All four Watson-Crick nucleosides were obtained — but only when the C, U, A and G reactions were run in separate vials.  The true one-pot gives C and A as main products, with G arriving indirectly through diaminopurine at lower yield.

Two concessions in the paper's own text.  It feeds in ribose as a reagent, and says why:  "It remains unclear… how ribose or any other carbohydrate… could have formed selectively."  And the geological scenario needs pH swings between 3 and 10 across "several basins" connected by streams, a landscape the authors say "we need to assume."

On the modern-reagent objection, the paper answers well:  the reductant was tested against propanedithiol, mercaptoethanol and cysteine, and all worked.

Selective prebiotic formation of RNA pyrimidine and DNA purine nucleosides

Xu, Jianfeng, et al.  Nature 582 (2020): 60-66.

area: nucleotides camp: mainstream-chemistry type: experiment access: paywalled direction: complicates

Author manuscript opened in full.  High-yielding, stereo- and regioselective synthesis of purine deoxyribonucleosides — deoxyadenosine and deoxyinosine, not deoxyguanosine — alongside the pyrimidine ribonucleosides, in one connected route.

This is not four canonical RNA ribonucleotides, and the article should not report it as such.  The authors reframe the mismatch as a positive hypothesis:  a mixed RNA/DNA first genetic system.

Extraterrestrial ribose and other sugars in primitive meteorites

Furukawa, Yoshihiro, et al.  PNAS 116 (2019): 24440-24445.

area: meteoritics camp: mainstream-chemistry type: measurement access: OA direction: supports

Opened in full from the Europe PMC XML.  All four aldopentoses — ribose, arabinose, xylose, lyxose — at 2.3-11 ppb in NWA 801 and 6.7-180 ppb in Murchison, below 0.5 ppb in NWA 7020.  Carbon isotope enrichment establishes the extraterrestrial origin rather than assuming it.

Two consequences.  It falsifies the assumption behind Larralde et al. 1995 that sugars are absent from meteorites.  And all four pentoses appear together, so the selectivity problem survives in a new form — space delivers ribose along with three other sugars, not ribose instead of them.

Timescales for Prebiotic Photochemistry Under Realistic Surface Ultraviolet Conditions

Rimmer, Paul B., et al.  Astrobiology 21 (2021): 1099-1120.

area: nucleotides camp: mainstream-chemistry type: measurement access: OA direction: complicates

CC BY, opened in full.  The most honest test in this section, because it reports what failed.  Replacing the mercury lamps with a solar simulator tuned to the faint young Sun:  hypophosphite oxidation, HCN reduction and nitroprusside photoproduction run "in days to weeks";  photoanomerisation of α-thiocytidine takes months;  hydroxyl-radical oxidation products take years;  and photoreduction of thioanhydrouridine with H₂S did not succeed even after long irradiation.

Three steps validated under realistic light, two very slow, one outright failure.  Their qualifier — "given a favorable atmosphere" — depends on an atmosphere model, which is itself inferred.

Paradoxes in the Origin of Life

Benner, Steven A.  Origins of Life and Evolution of Biospheres 44 (2014): 339-343.

area: philosophy camp: mainstream-chemistry type: philosophy-paper access: paywalled direction: undercuts

Opened in full.  The strongest framing available for the problem side, and the citation apologetics most often gets wrong.

It is not Astrobiology 2014.  It is Origins of Life and Evolution of Biospheres 44: 339-343.  A reviewer who checks one citation and finds it wrong will check no others.

The asphalt paradox:  "organic systems, given energy and left to themselves, devolve to give uselessly complex mixtures, 'asphalts'… the literature reports (to our knowledge) exactly zero confirmed observations where RIRI evolution emerged spontaneously from a devolving chemical system."  Benner applies it to metabolism-first and protocell-first as readily as to RNA-first — lipids, he writes, assemble "under the close supervision of a graduate student."  The water paradox:  "RNA requires water to function, but RNA cannot emerge in water, and does not persist in water without repair."

The last paragraph must be printed with them.  "We expect that (hope that?) most of these paradoxes will be resolved by experiments that show that the theories that generate them are incorrect, incomplete, or inapplicable."  Benner built the paradoxes as a research programme for breaking his own field's assumptions, and any presentation that omits this is a quote-mine.

The asphalt paradox originates in Neveu, Kim and Benner, Astrobiology 13 (2013): 391-403, which was not opened — Benner's restatement is verified, the original argument and any quantitative backing for "exactly zero confirmed observations" are not.  The paradox also has no quantitative form.  Nobody has counted how many prebiotic reaction mixtures were characterised for tar fraction, and that would be a genuinely original piece of work.

Interstep compatibility of a model for the prebiotic synthesis of RNA

Hirakawa, Yuki, et al.  PNAS 122 (2025): e2516418122.

area: nucleotides camp: mainstream-chemistry type: experiment access: OA direction: supports

CC BY-NC-ND, opened in full.  The strongest published rebuttal to the "chemist's hand does all the work" objection, and it is very recent.

A six-step "Discontinuous Synthesis Model" running from one- to three-carbon precursors to oligomeric RNA — predominantly 3′,5′-linked, at least six nucleotides, "possibly much longer" — claimed as a route with no human intervention.  Borate turned out to help other steps rather than poison them.

Three limits.  The product is a 6-mer, and the shortest known ribozymes are tens of nucleotides.  The paper's own claim of continuity carries asterisks — "continuous… except at points marked with *".  And it starts from D-ribose supplied as a laboratory reagent, inside a stacked geological scenario:  borate, basalt-constrained intermittently irrigated aquifers, a redox-neutral mantle, and a Vesta-sized impactor for transient atmospheric reduction.

Its methodological companion — White and Rimmer, "Do-Nothing Prebiotic Chemistry," Accounts of Chemical Research 58 (2025): 1-10, CC BY, opened in full — proposes measuring rate constants and building kinetic networks so that a model predicts what a real environment does "without direct intervention from a chemist."  That this is now an organised research programme is itself a concession that the objection is real.

Recent nucleotide work, and what could not be checked

Four post-cutoff items, all abstract only:

• Kan, H. W., et al., JACS Au 6 (2026): 922-931. Microkinetic simulation claiming to settle the formose autocatalysis debate; still describes the products as "complex" and mechanistically unresolved "after decades of study," with ribose "a particularly significant though minor product."
• Leyva, V., et al., Nature Communications 17 (2026): 2060. Multiple aldoses in the Orgueil meteorite, several near-racemic, with recovery experiments suggesting "sugar abundances are severely underestimated."
• Fang, Y., et al., National Science Review 13 (2026): nwag182. Ribose as the major sugar product from water and CO₂ by hydrodynamic cavitation. If that survives replication it materially weakens the ribose section. It ran in an engineered Venturi reactor, and the claim of natural occurrence is asserted in the abstract rather than demonstrated. Open this paper before relying on it; it is free at PMC.
• Vázquez-Salazar, A., Life 16 (2026): 240. The best current statement of where the problem now lies: "the emergence of functional sequence information and of control architectures that couple sequence to reproducible function remains a central open problem, and it sets clear limits on what chemistry alone can explain."

Never opened at all, and named here so nobody re-chases them:  Butlerow 1861 (metadata only);  Decker, Schweer & Pohlmann 1982, which is the universal citation for "the formose reaction makes more than 40 sugars" — no abstract was found and no count is asserted anywhere in this project;  Zubay 1998 and Zubay & Mui 2001, whose Pb²⁺-catalysed formose giving "more than 30% of the input formaldehyde… to a mixture of the aldopentoses" is the strongest published counter to "the formose reaction gives nothing" and is known only through Orgel's characterisation;  Mueller et al. 1990;  Sanchez & Orgel 1970;  Krishnamurthy, Guntha & Eschenmoser 2000;  and Becker et al. 2016, of which only an unusable 24 KB PDF fragment exists.  Szostak's 2009 Nature News & Views, source of the famous "one of the great advances in prebiotic chemistry," was not opened — only the free standfirst — and the sentence is relayed through Nicholas Wade's New York Times report, which was opened in full via the Wayback Machine and should be used only for on-the-record quotations, never for chemistry.

Homochirality

Life uses left-handed amino acids and right-handed sugars.  The apologetics claim that this cannot arise naturally is overstated, and several demonstrated mechanisms amplify a small handedness bias to near-purity in water, on real biological molecules.  What survives is smaller:  nobody knows where the initial bias came from, and the best empirical candidate — a chiral excess delivered from space — has just been undermined by the cleanest samples ever measured.

Chiral selection in poly(C)-directed synthesis of oligo(G)

Joyce, Gerald F., et al. (incl. Leslie Orgel).  Nature 310 (1984): 602-604.

area: homochirality camp: mainstream-chemistry type: experiment access: abstract-only direction: undercuts

Abstract only, verbatim from Europe PMC.  The load-bearing "this is a real problem" citation, and it comes from Orgel's own laboratory.  In template-directed oligomerisation of activated guanosine, wrong-handed monomers in a racemic mixture are incorporated as chain terminators.  "This inhibition raises an important problem for many theories of the origin of life."

Thirty years later Sczepanski and Joyce, Nature 515 (2014): 440-442 (abstract only), evolved an 83-nucleotide cross-chiral D-RNA polymerase that joins L-substrates on an L-template with a millionfold rate acceleration — showing a mirror-image replicating pair is chemically possible, and restating the 1984 problem as a "severe challenge" in the same breath.

On spontaneous asymmetric synthesis

Frank, F. C.  Biochimica et Biophysica Acta 11 (1953): 459-463.

area: homochirality camp: mainstream-chemistry type: calculation access: closed direction: supports

Never opened.  Crossref metadata only;  no abstract was obtained.  The theoretical foundation — an enantiomer that autocatalyses itself while suppressing its mirror amplifies a tiny fluctuation nonlinearly.  Frank's closing line, "A laboratory demonstration may not be impossible," is verified only as Blackmond's transcription of Frank, not against Frank's paper.

Asymmetric autocatalysis and amplification of enantiomeric excess of a chiral molecule

Soai, Kenso, et al.  Nature 378 (1995): 767-768.

area: homochirality camp: mainstream-chemistry type: experiment access: abstract-only direction: complicates

Abstract wording unverified.  The Nature page returned HTTP 200 but the fetch produced a paraphrase rather than the abstract, and Europe PMC has no record for the DOI.  Do not quote the 2% → 88% sequence.

The famous amplifier, and the one apologetics and popular science both over-rely on.  Blackmond 2010, which was opened in full, confirms the phenomenon and gives a different figure — 85% ee from a 0.1% ee circularly-polarised-light-derived initiator, citing Shibata et al. 1998 — and states that the Soai conditions "preclude it from being of direct prebiotic importance," because dialkylzinc chemistry is incompatible with water and air.  Blackmond put it more bluntly in 2026:  "completely prebiotically irrelevant, because it doesn't work in water."

Chiral symmetry breaking during crystallization · Emergence of a single solid chiral state from a nearly racemic amino acid derivative

Viedma, Cristóbal.  Physical Review Letters 94 (2005): 065504.  ·  and  ·  Noorduin, Wim L., et al.  JACS 130 (2008): 1158-1159.

area: homochirality camp: mainstream-chemistry type: experiment access: preprint direction: complicates

Viedma opened in full as the arXiv preprint (APS bot-blocks);  Noorduin opened in full as green OA from Groningen.

Viedma ripening:  grind a suspension of crystals with glass beads while they dissolve and regrow, and any small initial excess amplifies to complete chiral purity.  Viedma's system was sodium chlorate, which is achiral in solution — the handedness exists only in the crystal and is erased on dissolution, so the result says nothing directly about biological molecules.

Noorduin et al. closed that gap on a genuine chiral molecule — an imine of 2-methylbenzaldehyde and phenylglycinamide — driving an initial excess as low as 2-3% to a single chiral solid state.  The price:  an amino acid derivative rather than a free amino acid, a synthetic organic superbase, and mechanical attrition at 1,250 rpm.  The authors call it a proof of concept and claim no prebiotic scenario.  Blackmond adds the structural limit:  only about 10% of known chiral compounds form conglomerates, which is the prerequisite for this mechanism.

Amplification of enantiomeric concentrations under credible prebiotic conditions

Breslow, Ronald, and Mindy S. Levine.  PNAS 103 (2006): 12979-12980.

area: homochirality camp: mainstream-chemistry type: experiment access: OA direction: supports

Abstract verbatim from PubMed;  the free PMC full text was not read.  The strongest single result on the amplification side, and the one the article must concede by name.

1% enantiomeric excess of phenylalanine amplified to 90% by two successive evaporations that precipitate the racemate.  It runs in water, on a real protein amino acid, and needs nothing but a drying pond.

The mechanism was characterised the same year by Klussmann et al., Nature 441 (2006): 621-623 (abstract only):  the racemic crystalline compound is less soluble than the pure enantiomers, so partially dissolving a near-racemic solid leaves a strongly enriched solution.  Blackmond's eutectic table gives the ceiling for each amino acid:  serine above 99%, histidine 93%, phenylalanine 88%, leucine 87%, methionine 85%, alanine 60%.

A route to enantiopure RNA precursors from nearly racemic starting materials

Hein, Jason E., Eric Tse, and Donna G. Blackmond.  Nature Chemistry 3 (2011): 704-706.

area: homochirality camp: mainstream-chemistry type: experiment access: paywalled direction: supports

Opened in full via a mirror.  This is the paper the article must concede, because it undercuts the apologetics claim about as hard as it can be undercut.

Racemic glyceraldehyde plus 2-aminooxazole plus an amino acid additive gives over 90% ee D-glyceraldehyde, and enantiopure ribo-aminooxazoline crystals come out of a solution with an ee as low as 20% — starting from just 1% ee L-proline.  Their additive table:  proline 55%, tryptophan 10.2%, alanine 8.9%.

And it is an RNA precursor on the actual Sutherland route, not a model compound.

The same authors supply the honest cost of the photochemical route to the initial bias:  getting 20% ee by photoresolution destroys about 99% of both enantiomers.

The peptide-catalyzed stereospecific synthesis of tetroses

Weber, Arthur L., and Sandra Pizzarello.  PNAS 103 (2006): 12713-12717.

area: homochirality camp: mainstream-chemistry type: experiment access: OA direction: supports

Abstract verbatim from Europe PMC;  free at PMC.  L-dipeptide catalysts drive glycolaldehyde self-condensation to erythrose at more than 80% D-enantiomeric excess, in water — a demonstrated transfer of handedness from amino acids to sugars.

Use this and not the 2004 Science Brevia it follows up.  Pizzarello and Weber, Science 303 (2004): 1151, is a one-page Brevia with no abstract available anywhere;  it was never opened, and no figure from it may be quoted.

The Origin of Biological Homochirality

Blackmond, Donna G.  Cold Spring Harbor Perspectives in Biology 2 (2010): a002147.

area: homochirality camp: mainstream-chemistry type: review access: OA direction: complicates

Opened in full from the free PDF.  The standard review, by the field's leading experimentalist, and the right place to get the baseline statements and the caveats attached to every mechanism above.  Credits Morowitz 1969 for the eutectic idea.

The meteorite evidence, and why it just changed

area: homochirality camp: mainstream-chemistry type: measurement direction: complicates

This is the most important development in this section, and it goes against both sides.

The classical result.  Cronin and Pizzarello, Science 275 (1997) — abstract only — reported L-excesses of 7.0% and 9.1% in α-methyl amino acids from Murchison, with the α-hydrogen analogues racemic.  Pizzarello and Cronin, GCA 64 (2000) — abstract only — extended this to 1.0-9.2% across six α-methyl amino acids, and stated the crucial negative:  for the protein amino acids, excess was "not observed… or attributed to terrestrial contamination," and "substantial excess of L-alanine reported by others was not found."  Glavin and Dworkin, PNAS 106 (2009) — abstract only — got the largest figure ever reported, L-isovaline at 18.5 ± 2.6% in Murchison — and showed it correlates with aqueous alteration on the parent body, concluding that the "results are inconsistent with UV circularly polarized light as the primary mechanism."

The sample-return results, which are cleaner because nothing ever touched the ground.  Parker et al. on Ryugu, GCA 347 (2023), opened in full:  β-aminoisobutyric acid D/L = 0.95 ± 0.08 and 0.98 ± 0.12, alanine racemic within error, a large serine L-excess read as terrestrial contamination.  Chan, Watson and Sephton on Winchcombe, LPSC 54 abstract 1474 (2023), opened in full:  alanine 1.13 ± 0.16, norvaline 0.91, isovaline 1.06 — racemic, in the freshest CM2 fall available.  (The journal version, MAPS 59 (2024): 1101-1130, could not be opened — Wiley and the Open University repository both bot-block — so the numbers used here are the conference abstract's.)

And then Bennu.  Glavin et al., Nature Astronomy 9 (2025), opened in full including Extended Data Table 4:  every chiral non-protein amino acid that could be resolved came back racemic or near-racemic, isovaline included (3 ± 12% free, −3.9 ± 6.2% total).  Their own conclusion:  "The source of the meteoritic L-amino acid enrichments remains a mystery," and the Bennu result "challenges the hypothesis that the emergence of left-handed protein-based life on Earth was influenced by an early Solar System bias toward L-amino acids."

An unexplained discrepancy that somebody should chase.  The same 2025 paper measured Murchison isovaline alongside Bennu in a hot-water extract and got 0.6 ± 15% free and −0.8 ± 6.5% total — racemic, against the 18.5% of 2009.  The paper does not discuss it.  Stone-to-stone heterogeneity, a different extraction protocol, or something else;  this research could not tell which.  Do not print the 18.5% figure without this qualification.

Also unverified:  Glavin et al. 2012 on Tagish Lake, the only claimed large L-excess in genuine protein amino acids at around 60% in aspartic and glutamic acid, was never opened (Wiley bot-blocks, not in PubMed) and the figure is second-hand through Glavin 2025.

This cuts both ways and the article should say so.  It removes the naturalist's best empirical evidence for a delivered chiral bias.  It also removes the creationist's "the meteorite excesses are just contamination" reply, because the pristine samples establish an extraterrestrial origin by racemic composition, Chargaff-rule violation and non-protein amino acid diversity, and then report no bias.

A 2026 subtlety, post-cutoff:  Baczynski et al., PNAS 123 (2026): e2517723123, CC BY, opened in full, found δ¹⁵N of D-glutamic acid in Bennu 87‰ higher than L-glutamic acid even though the bulk is racemic — "even when bulk products are racemic, at the smallest scales, D- and L-enantiomers are not necessarily in equal proportions."  Carry that caveat rather than declaring the delivery hypothesis dead outright.

Circular polarisation and the astrophysical route

Bailey, Jeremy, et al.  Science 281 (1998): 672-674.  ·  Fukue, Tsubasa, et al.  Origins of Life and Evolution of Biospheres 40 (2010): 335-346.  ·  Bocková, Jana, et al.  Nature Communications 14 (2023): 3381.

area: homochirality camp: mainstream-chemistry type: measurement access: preprint direction: complicates

Fukue opened in full as the arXiv preprint;  Bailey and Bocková abstract only.

Bailey et al. measured strong infrared circular polarisation in Orion and wrote, in carefully modal language, that ultraviolet CPL "might have been important" and "could account for" the Murchison excess.  Fukue et al. quantified it:  up to +17% and −5% in the BN/KL region, no significant polarisation around the Trapezium, and none of 353 low-mass Sun-like young stars showing significant integrated polarisation above 1.5%.  Bocková et al. measured what CPL actually delivers on isovaline films:  up to 2% ee, and framed the low efficiency as an explanation for why no excess is seen in the most pristine chondrites — which is a rescue, not a prediction.

Modica et al., ApJ 788 (2014): 79, the source of the widely quoted "~1% ee" from CPL on interstellar ice analogues, was never opened;  the figure is relayed through Sharma, MNRAS 517 (2022), which was fetched.

The magnetic surface route

Ozturk, S. Furkan, Zhaobin Liu, John D. Sutherland, and Dimitar D. Sasselov.  Science Advances 9 (2023): eadg8274.  ·  and  ·  Devienne, J. A. P. M., et al.  arXiv:2605.19387 (2026).

area: homochirality camp: mainstream-chemistry type: experiment access: preprint direction: complicates

Ozturk abstract only (science.org bot-blocks);  Devienne opened in full.  The latter is post-cutoff.

The current best candidate for the initial bias.  Crystallising racemic ribo-aminooxazoline — a real Sutherland-route RNA precursor — on magnetised magnetite gives ~60% ee, and 100% on a second crystallisation, via the chiral-induced spin selectivity effect.  The magnetic field used was stronger than natural rock magnetisation, and the underlying CISS physics is not settled;  Matthew Fuchter, on the record in 2026, says there is "still a debate about how and why it works" and "I don't see it as a panacea."

Devienne et al. characterise magnetite made by two geochemically plausible routes and simulate the magnetic response, finding single- and multi-vortex domains that can store and reinforce a weak chiral bias.  It is characterisation plus simulation, not a crystallisation experiment on natural mineral surfaces.  No such wet-lab follow-up was found as of 2 August 2026.

Their framing of the field is the most useful sentence in the recent literature, and it is not a creationist writing:  proposed mechanisms "are generally considered insufficient… they lack both an effective mechanism to amplify an initial enantiomeric imbalance and a plausible chemical pathway to transmit homochirality."

Prebiotic chiral transfer from self-aminoacylating ribozymes may favor either handedness

Kenchel, Joseph, et al.  Nature Communications 15 (2024): 7980.

area: homochirality camp: mainstream-chemistry type: experiment access: OA direction: complicates

Abstract verbatim from Europe PMC.  Of fifteen self-aminoacylating D-ribozymes tested, "a substantial fraction react preferentially to produce the D-enantiomer of the product."  D-RNA has no intrinsic bias toward L-amino acids, so the specific pairing life uses — D-sugars with L-amino acids — is not forced by chemistry.  That is a frozen accident, and it is a fact both sides should have to sit with.

Origin of life: the chirality problem

Sarfati, Jonathan.  Journal of Creation 12, no. 3 (1998): 263-266.  Updated 24 September 2024.

area: apologetics camp: creationist type: web-page access: OA direction: undercuts

Opened at HTTP 200 in the apologetics audit run.  (The homochirality run itself could not reach creation.com;  the block is intermittent and user-agent-dependent.)

The claim quotes a correct organic-chemistry textbook rule — synthesis of chiral compounds from achiral reagents yields the racemic modification — and computes 2^-100 ≈ 10^-30 for a hundred-residue homochiral chain arising by chance.

The rule is right and the calculation is beside the point.  2^-100 assumes each residue is an independent draw.  Every mechanism catalogued above — Viedma ripening, eutectic enrichment, Breslow's evaporation, Hein and Blackmond's crystallisation, Ozturk's magnetite — is non-independent by construction, and Sarfati discusses autocatalysis elsewhere in the same piece.

Credit where it is due, and the article should give it:  CMI's current page does not claim homochirality is impossible.  It calls it a "huge problem" and a "complete mystery," and it engages Blackmond's work by name.  That is a defensible position and close to the one this article takes.

The RNA world and ribozymes

"No molecule has ever copied itself" stopped being true on 12 February 2026, and this section exists mostly to state precisely what is and is not true now.  The literature here moves faster than any other part of the field — five significant papers in eighteen months — so anything written about it before 2025 should be assumed out of date, and this section should be re-checked immediately before publication.

Selforganization of matter and the evolution of biological macromolecules

Eigen, Manfred.  Naturwissenschaften 58 (1971): 465-523.

area: rna-world camp: mainstream-chemistry type: calculation access: closed direction: undercuts

Never opened;  no free copy found anywhere and no abstract obtained.  The error threshold — a genome cannot exceed roughly ln s/(1−q) nucleotides at per-nucleotide fidelity q without melting under its own copying errors.  Cite Joyce & Szostak 2018 for the arithmetic and Eigen for priority, because that is the honest chain of custody here.

Kun, Santos and Szathmáry, Nature Genetics 37 (2005): 1008-1011 — abstract only, body not opened — relaxed the naive threshold by using real ribozyme fitness landscapes:  above 7,000 nucleotides at a copying fidelity of 0.999.  Their own caveat is the operative half:  that "requires an order of magnitude of improvement in the accuracy of in vitro-generated polymerase ribozymes."  The 7,000 figure carries a lot of weight and is verified only against the abstract.

Prebiotic Chemistry and the Origin of the RNA World

Orgel, Leslie E.  Critical Reviews in Biochemistry and Molecular Biology 39 (2004): 99-123.

area: rna-world camp: mainstream-chemistry type: review access: paywalled direction: undercuts

Opened in full, all twenty-six pages, and searched:  the word "miracle" appears zero times.  See the failed-verification section above.

The real content is better than the misquotation.  "The abiotic synthesis of RNA is so difficult that it is unclear that the RNA World could have evolved de novo on the primitive Earth."  "There is at present no convincing, prebiotic total synthesis of any of the nucleotides."  "The prebiotic synthesis of nucleotides in a sufficiently pure state to support RNA synthesis cannot be achieved using presently known chemistry."  And:  "One must recognize that, despite considerable progress, the problem of the origin of the RNA World is far from being solved."

Orgel was not conceding what he is quoted as conceding.  He was arguing against RNA-first and for a simpler pre-RNA genetic polymer, following Cairns-Smith.  His pessimism is a position inside naturalistic origin-of-life research, not a verdict against it.  Note also the date-stamp in his own sentence:  "presently known chemistry" means 2004, and Powner 2009 and Becker 2019 came after.

A Simpler Origin for Life · Small Molecule Interactions Were Central to the Origin of Life

Shapiro, Robert.  Scientific American 296, no. 6 (2007): 46-53.  ·  and  ·  Quarterly Review of Biology 81 (2006): 105-125.

area: rna-world camp: atheist-explicit type: popular access: OA direction: undercuts

The Scientific American print-view PDF was opened and read in full.  The QRB body was not opened — the University of Chicago Press bot-blocks, and the clay-mineral oligomerisation calculation that is its technical core has not been checked.  JSTOR may hold QRB;  worth a look before quoting any number from it.

The demolition of RNA-first from inside the field:  the "chemist's hand" charge, the golfer analogy — no physical law need be broken, but the odds are immense — and the gorilla at the keyboard.  "Many chemists… have fled the RNA-first hypothesis as if it were a building on fire."

Read it with what Shapiro then does.  He proposes a small-molecule metabolism-first alternative, concedes "relatively little experimental work" supports it, calls for a "smoking gun" experiment nobody has run, and closes by quoting Kauffman approvingly against Monod:  "we are far more likely to share it with unknown companions."  He is arguing for a different naturalistic route, not against natural origins.

The Origins of the RNA World

Robertson, Michael P., and Gerald F. Joyce.  Cold Spring Harbor Perspectives in Biology 4 (2012): a003608.

area: rna-world camp: mainstream-chemistry type: review access: OA direction: complicates

Opened in full;  the "near miracle" passage verified on rendered page 15.  This is the only citation the article may use for that phrase.

The passage describes what "the second group of scientists" believes, contrasted against a first and more optimistic group.  It is not a first-person verdict.  The next sentence is "Time will tell which is correct."

Two details flagged for anyone who wants to close the loop.  The 2012 text reads "on the abiotic Earth" where circulating versions read "on the primitive earth."  And the parenthetical "(The authors subscribe to this latter view.)" that some renderings carry does not appear in the verified 2012 text.  The underlying book chapter in The RNA World was never opened — the Cold Spring Harbor Monograph Archive is purchase-walled and archive.org's search-inside API did not resolve — so whether that parenthetical was ever printed is unknown.  If it is real, the apologetics use is considerably fairer than it currently looks, and somebody with the book should check chapter 1 of the 1993 edition, chapter 3 of the 1999, and chapter 2 of the 2006.

Their own balanced verdict is worth quoting alongside:  the constraints "are reasonably well understood… [but] details of this process remain obscure and are not likely to be known in the near future."

The polymerase ribozymes, in order

area: ribozymes camp: mainstream-chemistry type: experiment direction: complicates

The single most useful thing this catalogue can do for this topic is put the numbers in a row, because each paper is quoted in isolation by somebody.

Johnston et al., Science 292 (2001): 1319-1325.  R18, about 189 nucleotides, extends a primer by 14 nucleotides at roughly 96.7% fidelity — the first polymerase ribozyme.  Never opened;  the figures are relayed through Bentin 2011 and Attwater 2013, both of which were opened, and the secondary reports disagree on its length (189, 187, ~200).

Wochner et al., Science 332 (2011): 209-212.  tC19Z.  Never opened (science.org bot-blocks).  The frequently quoted error rate of 8.8 × 10^-3 per nucleotide — 99.12% fidelity — is relayed through Bentin 2011 and does not reconcile with Attwater 2013's 97.3% and 98.3% for related constructs.  Almost certainly different templates and conditions.  Do not quote 99.12% until somebody reads the paper.

Bentin, Artif. DNA PNA XNA 2 (2011): 40-42.  Opened in full.  The commentary that supplies most of the tC19/tC19Z numbers, and its verdict:  "maximally synthesizing a 95 nt RNA transcript would still fall short of copying their own gene of 187 nt."

Attwater, Wochner and Holliger, Nature Chemistry 5 (2013): 1011-1018.  Opened in full.  tC9Y synthesises RNA up to 206 nucleotides — longer than itself — at 97.7% fidelity on full-length products, in 60 hours at 17 °C.  The product came off favourable repeating-sequence templates, not a mixed-sequence gene.  "Longer than itself" is a length claim.  tC9Y did not copy tC9Y.

Horning and Joyce, PNAS 113 (2016): 9786-9791.  PDF obtained via PMC, abstract verified verbatim.  The 24-3 polymerase amplifies short RNA templates more than ten-thousandfold in a ribozyme-catalysed analogue of PCR, and makes a hammerhead ribozyme at 2.4% in 24 hours.  Amplification worked only on templates far shorter than the polymerase, with experimenter-supplied thermal cycling.

Attwater et al., eLife 7 (2018): e35255.  Opened in full.  The t5+1 ribozyme synthesises its own catalytic domain in five pieces at 6-10% yield and 97.4% fidelity — with about a quarter of the triplet junctions pre-ligated using pre-made hexa- and nonanucleotides, and one of two subunits never made at all.  Their own conclusion:  "efficiency of both segment synthesis and assembly will need to be increased significantly to realise a full self-replication cycle."

Tjhung et al., PNAS 117 (2020): 2906-2913.  Opened in full.  The 38-6 polymerase, about 200 nucleotides, synthesises the class I ligase — its own evolutionary ancestor — in three fragments at 1.0-5.7% yield over three days.  The assembled ligase was about 8,000-fold less active than the real thing, and fidelity on the hardest fragment was 83% with insertion hotspots above 40%.  Their own sentence:  the fidelity of RNA polymerisation "should be considered a major impediment," and self-replication reactions "always depended on carefully curated sets of preformed oligonucleotide substrates."

Papastavrou, Horning and Joyce, PNAS 121 (2024): e2321592121.  Opened in full.  The 71-89 polymerase replicates a 34-nucleotide hammerhead through eight rounds, and fitter variants emerge — Darwinian evolution catalysed by RNA.  Yields 7-23% in 24 hours, fidelity 81-91%.  They state what would be needed to close the loop:  97-98% fidelity, which they call "not unattainable."

Attwater et al., Nature Chemistry 17 (2025): 1129-1137.  CC BY, opened in full.  Trinucleotide substrates under pH and freeze-thaw cycling give open-ended exponential replication of a 29-nucleotide fragment of the ribozyme's own core, up to 1.26-fold amplification per cycle.  This is the answer to the strand-separation problem, and it covers 29 of about 180 nucleotides.

A small polymerase ribozyme that can synthesize itself and its complementary strand

Gianni, Edoardo, et al. (Holliger lab).  Science 391 (2026): 1022-1028.  Published online 12 February 2026.

area: ribozymes camp: mainstream-chemistry type: experiment access: OA direction: supports

CC BY author manuscript, opened in full.  Post-cutoff, and the most important single result in this catalogue.

QT45 is a 45-nucleotide ribozyme, pulled from three random pools totalling about 1.2 × 10¹³ sequences through eleven rounds of selection and seven of mutagenesis.  What it does:

• Makes its own complementary strand, full-length, from a complete 64-triplet random pool: 0.24% yield in 72 days, 10.9% of full-length products perfect, 94.1% per-nucleotide fidelity.
• Makes a copy of itself on a QT45 minus-strand template: 0.17% yield in 72 days, using the required triplets plus one pre-made defined hexamer; 43.4% of full-length products perfect.
• Makes a 33-nucleotide hammerhead at 0.39-1.88% over 65 days, at 92.6% fidelity, with the products showing about 19% of wild-type activity.
• Half-life about 117 days, against roughly 16 days for the 200-nucleotide R18.
• Estimated abundance of the QT fold in random sequence space: 9.5 × 10^-17.

What it does not do, and the article must state this precisely.  The two reactions were run in separate tubes, not sequentially and not in one pot.  No closed replication cycle was demonstrated.  Lead author Edoardo Gianni, on the record:  "The two reactions don't happen sequentially or in one pot so we are not yet at the stage of self-replication."

Two obstacles the paper identifies itself.  Strand inhibition:  "the (+) strand outcompetes triplet substrates for binding to the (−) strand template," forming a dead-end duplex — "a key obstacle to closing the self-replication cycle."  And their own quasispecies modelling says that at 92.6-94.1% fidelity "the population of active variants (quasispecies) would be minimal (<0.5%)";  at 97.4% it would be about 12%.

The optimistic reading, in the lead author's own words, and the article should print it:  the result "substantially lowers the bar that needs to be achieved non-enzymatically before ribozyme-catalysed replication kicks in… we can estimate [the probability of life spontaneously emerging] to be higher than before."  David Lilley of Dundee, also on the record, gives both halves in one breath:  "That is unbelievably slow" and "a good step down the road."

Also post-cutoff:  Kent et al., PNAS 123 (2026): e2606892123 — bioRxiv v3 opened in full;  the published version bot-blocks — report that tC19Z under saturating pyrophosphate repairs RNA terminated in a mismatch or a non-extendable cyclic phosphate, a proofreading-like mechanism achieved by RNA alone.  That attacks the fidelity problem directly.  Its PNAS commentary (Murugan 2026) has no abstract obtainable anywhere and was not opened.

Self-sustained replication of an RNA enzyme

Lincoln, Tracey A., and Gerald F. Joyce.  Science 323 (2009): 1229-1232.  ·  and  ·  Robertson, Michael P., and Gerald F. Joyce.  Chemistry & Biology 21 (2014): 238-245.

area: ribozymes camp: mainstream-chemistry type: experiment access: abstract-only direction: complicates

Both abstract only.  Two ligase enzymes cross-catalysing each other's synthesis from four oligonucleotide substrates, amplifying exponentially with a one-hour doubling time in 2009 and a five-minute doubling time in 2014 — "more than 80 logs of growth per day."

The limitation is why this is not the headline result it looks like, and Joyce and Szostak state it themselves:  the system joins two long, pre-made, experimenter-supplied halves.  It copies no sequence information beyond choosing which half, so it cannot explore sequence space or evolve open-endedly.  A real self-replicator and a real dead end.

Protocells and RNA Self-Replication · The eightfold path to non-enzymatic RNA replication

Joyce, Gerald F., and Jack W. Szostak.  Cold Spring Harbor Perspectives in Biology 10 (2018): a034801.  ·  and  ·  Szostak, Jack W.  Journal of Systems Chemistry 3 (2012): 2.

area: rna-world camp: mainstream-chemistry type: review access: OA direction: complicates

Both opened in full.  The single most load-bearing pair of citations in this project, because between them they state every remaining obstacle in the words of the two people best placed to know.

Szostak's eight problems include the one this article leans on hardest:  "the high concentrations of divalent cations required for RNA template-copying reactions catalyze RNA degradation and are incompatible with known vesicle replication systems."  His framing is not despair — "there are potential solutions to all of these problems" — and the article should print that too.  From the same author in the same paper:  "The creation of a true replicase remains a great experimental challenge."

Joyce and Szostak give the numbers:  the error threshold formula, a worked example putting a 25-nucleotide ribozyme genome at q ≈ 0.98, the cross-inhibition and strand-separation and stalling problems, "fatty acid-based membranes become unstable in ≥4 mM Mg²⁺, indicating an apparent incompatibility between the two systems," and the closing statement:  "It also is not clear how to achieve multiple generations of nucleic acid replication within the context of a protocell.  If and only if all of these challenges can be met will it be possible to synthesize an evolving protocell."

The Szostak retraction

Jia, Tony Z., Albert C. Fahrenbach, Neha P. Kamat, Katarzyna P. Adamala, and Jack W. Szostak.  Nature Chemistry 8 (2016): 915-921.  Retracted, Nature Chemistry 9 (2017): 1286.

area: ribozymes camp: mainstream-chemistry type: retraction access: OA direction: complicates

The retraction notice was opened in full and transcribed verbatim.  The original paper's PDF was never obtained — the PMC render endpoint 404s — so only its abstract was read.

The original claim was that oligoarginine peptides slow the annealing of complementary RNA strands by up to several thousandfold, letting primer extension proceed inside a coacervate — a proposed solution to the strand-separation problem.  The notice states that the authors "have been unable to reproduce" it, that the published data "are the result of false positives that arose from an incorrectly designed experiment," that the annealing slowdown "was actually an artefact due to slow coalescence or strand exchange between droplets," and that the circular dichroism result "may also be an artefact."  Two sub-results were reproduced and stand.  "Since the main conclusions of our paper are incorrect, all of the authors are now retracting the Article."

Use this as an example of the field policing itself, and not as evidence that origin-of-life research is unreliable.  A reader who looks it up will find that Szostak comes out of it well.  On the record:  "definitely embarrassing";  "we were totally blinded by our belief [in our findings]… we were not as careful or rigorous as we should have been (and as Tivoli was)";  "The only saving grace is that we are the ones who discovered and corrected our own errors."  The junior scientist who found the errors, Tivoli Olsen, is credited in the notice, and said:  "I fulfilled my obligation to insure that no one after me would waste their time on this."

Two 2026 reviews, and the state of the field

Vázquez-Salazar, A.  Trends in Biochemical Sciences (2026);  and Life 16 (2026): 240.

area: rna-world camp: mainstream-chemistry type: review access: abstract-only direction: complicates

Both abstract only.  Both post-cutoff.  The first reframes the RNA world as "an RNA-centered evolutionary phase rather than as an RNA-only biosphere" and says the classical objections are now "experimentally measurable constraints."  Both halves of that are useful — the retreat in framing, and the confidence in the conclusion.

The second is the best available current statement of the open problem, from a non-creationist:  "the emergence of functional sequence information and of control architectures that couple sequence to reproducible function remains a central open problem, and it sets clear limits on what chemistry alone can explain."

A standing gap in this section.  The whole Holliger line depends on triplet triphosphate substrates, and nobody has shown a prebiotic route to a triplet pool.  Gianni et al. gesture at 5′-adenylated alternatives.  That is potentially the strongest remaining objection to the QT45 result, and it is an open question in the literature, not only for this project.

Protocells and membranes

The compartment literature is the strongest experimental part of the field.  Model protocells grow, divide, compete and encapsulate RNA, and every one of those results comes from a mainstream laboratory.  The unsolved problem is a specific chemical incompatibility, and it was identified and quantified by the laboratory doing the work, not by critics:  RNA copying needs magnesium at 50-200 mM, and a fatty-acid membrane falls apart above about 4 mM.

The first cell membranes · Self-assembled vesicles of monocarboxylic acids and alcohols

Deamer, David W., et al.  Astrobiology 2 (2002): 371-381.  ·  and  ·  Apel, Charles L., David W. Deamer, and Michael N. Mautner.  Biochimica et Biophysica Acta 1559 (2002): 1-9.  ·  and  ·  Monnard, Pierre-Alain, et al.  Astrobiology 2 (2002): 139-152.

area: membranes camp: mainstream-chemistry type: experiment access: abstract-only direction: undercuts

Deamer et al. opened.  Apel et al. and Monnard et al. abstract only — both paywalled or bot-blocked, and their concentration thresholds should be checked before the article prints a table of them.

The chain-length constraint:  above six carbons single-chain amphiphiles form micelles, at eight or more they form bilayers, and stability rises with chain length while the critical bilayer concentration falls.  Apel et al. give the thresholds — octanoic acid above 150 mM, nonanoic acid at 85 mM, dropping to about 20 mM with nonanol added.  Monnard et al. is the founding statement that monocarboxylic acid vesicles are disrupted by low divalent cation concentrations and by high sodium chloride, concluding marine environments would inhibit self-assembly.

Experimental models of primitive cellular compartments

Hanczyc, Martin M., Shelly M. Fujikawa, and Jack W. Szostak.  Science 302 (2003): 618-622.

area: protocells camp: mainstream-chemistry type: experiment access: OA direction: supports

Author manuscript opened in full via PMC.  Montmorillonite accelerates the conversion of fatty acid micelles into vesicles;  clay particles and adsorbed RNA are encapsulated in the process;  the vesicles grow by micelle feeding and divide by extrusion through small pores without diluting their contents.

Szostak's own later assessment supplies the caveat:  Zhu and Szostak 2009 notes that "this artificial division process results in significant loss of protocell contents during each division cycle."  The 2009 paper exists because 2003's division was not good enough.

The emergence of competition between model protocells

Chen, Irene A., Richard W. Roberts, and Jack W. Szostak.  Science 305 (2004): 1474-1476.

area: protocells camp: mainstream-chemistry type: experiment access: OA direction: supports

Author manuscript opened in full.  Concede this one without hedging.  RNA encapsulated in a vesicle exerts osmotic pressure, and the swollen vesicle takes membrane material from relaxed neighbours — up to about 35% of their membrane.  That is Darwinian competition between compartments with no genome doing any catalytic work at all, driven by physics.

RNA catalysis in model protocell vesicles

Chen, Irene A., Kourosh Salehi-Ashtiani, and Jack W. Szostak.  JACS 127 (2005): 13213-13219.

area: protocells camp: mainstream-chemistry type: measurement access: OA direction: undercuts

Author manuscript opened in full.  The sharpest objection in the whole protocell literature, and it was raised by the laboratory doing the work.

Their Table 1.  Decanoic acid vesicles are disrupted at about 0.2 M sodium chloride.  Divalent cations cause aggregation below 1.5 mM magnesium chloride.  The magnesium tolerance ladder:  pure fatty acid leaks at 0.5 mM and goes turbid at 1 mM;  a 4:1 acid/monoglyceride mixture reaches 2 and 3 mM;  a 2:1 mixture reaches 4 and 6 mM.  At 4 mM magnesium the dye-leakage rate rises 27-fold — though a 10-nucleotide RNA showed no leakage after 19 hours.

Even the best mixture tops out at 4 mM.  RNA copying wants 50 to 200.  That is roughly two orders of magnitude, measured in one paper.

Coupled growth and division of model protocell membranes

Zhu, Ting F., and Jack W. Szostak.  JACS 131 (2009): 5705-5713.

area: protocells camp: mainstream-chemistry type: experiment access: OA direction: supports

Author manuscript opened in full.  Multilamellar vesicles fed oleate micelles grow into threads and divide under a shear stress of 0.35 dynes/cm² — the sort a stream or a wind-driven surface supplies.  Surface area rises about fourfold per five equivalents of oleate;  vesicle count rises exponentially over three cycles with r² above 0.99;  RNA cargo is distributed to the daughters.

The RNA is inert cargo.  It is not copied.  That is the honest limit of this result and of every result like it.

Template-directed synthesis of a genetic polymer in a model protocell

Mansy, Sheref S., et al.  Nature 454 (2008): 122-125.

area: protocells camp: mainstream-chemistry type: experiment access: abstract-only direction: supports

Abstract only, and this one matters.  The Europe PMC PDF render endpoint returned HTTP 500 twice and PMC's own PDF route is bot-blocked.  It is the primary citation for "activated nucleotides cross a fatty acid membrane unaided," which the argument leans on, and nobody here has read the body.

The claim is that activated nucleotides cross fatty-acid membranes without transporters and take part in efficient template copying inside the protocell, so a primitive protocell could have been an obligate heterotroph with no transport proteins.

Nonenzymatic template-directed RNA synthesis inside model protocells

Adamala, Katarzyna, and Jack W. Szostak.  Science 342 (2013): 1098-1100.

area: protocells camp: mainstream-chemistry type: experiment access: OA direction: complicates

Author manuscript opened in full.  The best attempt at the magnesium problem, and it comes with a bill the authors state themselves.

Citrate, isocitrate, oxalate, NTA and EDTA all protect oleate vesicles at 10 mM magnesium and above.  At 50 mM magnesium with a fourfold excess of sodium citrate, the primer-extension rate is only halved, from 1.4 to 0.67 per hour, and citrate cuts magnesium-catalysed RNA cleavage from 0.074 to 0.004 per hour at 75 °C.  Inside oleate vesicles most primers extended by six G residues in 24 hours, some by seven.

Their own caveat:  "In the absence of a prebiotic citrate synthesis pathway… it is of interest to consider prebiotically plausible alternatives."  Thirteen years later, a 2026 JACS computational study by the same laboratory is still hoping to inform the "search for metal ion chelators that further enhance nonenzymatic primer extension."  The citrate rescue has not been replaced.

Prebiotic amino acids bind to and stabilize prebiotic fatty acid membranes

Cornell, Caitlin E., et al.  PNAS 116 (2019): 17239-17244.

area: membranes camp: mainstream-chemistry type: experiment access: OA direction: complicates

Author manuscript opened in full.  Serine and glycine, though not leucine, protect decanoic acid vesicles against 10 mM magnesium by increasing lamellarity — from over 80% paucilamellar without amino acids to over 30% with four or more lamellae with serine — and reduce salt flocculation at 300 mM sodium chloride under heat cycling.

Two limits.  10 mM is still far below 50-200 mM.  And as Cohen et al. note, "the assembly of RNA has not been investigated inside these vesicles."

Promotion of protocell self-assembly from mixed amphiphiles at the origin of life

Jordan, Sean F., et al. (Lane lab).  Nature Ecology & Evolution 3 (2019): 1705-1714.

area: membranes camp: mainstream-chemistry type: experiment access: preprint direction: complicates

Accepted manuscript opened in full from UCL Discovery.  The result that stops either camp from declaring the other's compartment impossible.

Mixtures of C10-C15 fatty acids and 1-alkanols form vesicles at modern seawater ionic strength — 50 mM magnesium, 10 mM calcium — at 70 °C.  Mixture complexity is what does the work:  the critical bilayer concentration drops thirtyfold, from 39 mM for pure decanoic acid to 1.3 mM for a twelve-component mixture.

The price is pH 11-12, because the mechanism is hydroxide sequestration of the divalent cations, and "very few" vesicles form below pH 9.  A conflict follows that is this project's inference and not the authors' claim:  RNA hydrolyses fast at that pH, and Cohen et al. had to drop from pH 10 to 9 for ribozyme activity.  No paper was found that tests nucleic acid stability or copying inside Jordan-type vesicles at their working pH, and if the article uses this point it must be labelled as inference.

Amphiphilic Compounds Assemble into Membranous Vesicles in Hydrothermal Hot Spring Water but Not in Seawater · Can Life Begin on Enceladus?

Milshteyn, Daniel, et al.  Life 8, no. 2 (2018): 11.  ·  and  ·  Deamer, David, and Bruce Damer.  Astrobiology 17 (2017): 834-839.

area: membranes camp: mainstream-chemistry type: experiment access: OA direction: undercuts

Both CC BY and opened in full.  The membrane objection to hydrothermal vents, with a field measurement behind it.

Milshteyn et al. took real Yellowstone hot spring water — total cations and anions from micromolar to 27 mM, one sample at 0.026 mM magnesium, two thousand times below seawater — and formed stable dodecanoic acid/monoglyceride membranes in it.  In seawater (0.6 M NaCl, 54 mM MgCl₂, 10 mM CaCl₂) no amphiphile tested formed stable vesicles;  lauric acid made crystals and insoluble soaps instead.

Deamer and Damer state the consequence flatly:  "Self-assembly of membranes composed of amphiphiles has not been demonstrated in hydrothermal vents, either in situ or in laboratory simulations."  They add a second and independent objection:  dehydration-rehydration cycling is the simplest known route to encapsulating polymers, and it cannot happen at a submarine vent.

Natural soda lakes provide compatible conditions for RNA and membrane function

Cohen, Zoe R., et al.  PNAS Nexus 3 (2024): pgae084.

area: protocells camp: mainstream-chemistry type: experiment access: OA direction: complicates

CC BY, opened in full.  The most honest environment paper in this section, because it reports what did not work.

Carbonate precipitation caps soda lakes at about 1 mM magnesium.  Real samples from Last Chance and Goodenough Lakes in British Columbia — roughly 1 M sodium, 1 mM magnesium, pH 10 — supported non-enzymatic RNA-templated polymerisation at rates comparable to standard laboratory conditions with 50 mM magnesium chloride.

And then the caveats, all theirs.  The ligase ribozyme needed the pH dropped to 9.  High monovalent cations lowered ligation yields, and fewer primers extended by three or more nucleotides in 72 hours than under standard conditions.  At 1 M sodium the amphiphiles "immediately floated to the surface" as flocculated aggregates, and the authors "were unable to determine" whether vesicles survive assembly directly in dry-season lake water.  Their working scenario is therefore seasonal:  assemble in the wet season, persist through the dry.

Cyclophospholipids Enable a Protocellular Life Cycle

Toparlak, Ömer Duhan, et al.  ACS Nano 17 (2023): 23772-23783.

area: protocells camp: mainstream-chemistry type: experiment access: OA direction: complicates

CC BY-NC, opened in full.  The best published protocell system, and its opening sentence is the field's own diagnosis:  "There is currently no plausible path for the emergence of a self-replicating protocell."

Cyclophospholipid/fatty-acid mixtures grow and divide while staying stable in 25 mM magnesium-citrate.  What it removes is the cost citrate used to impose — stopped growth and division — not the dependency on citrate itself.  It still rests on an environment nobody can point to.

Hybrid Protocells Based on Coacervate-Templated Fatty Acid Vesicles

Lee, Jaeseung, Fatma Pir Cakmak, Ross Booth, and Christine D. Keating.  Small 20 (2024): e2406671.  ·  and  ·  Joshi, Manesh Prakash, et al.  Langmuir 42 (2026): 13582-13595.

area: protocells camp: mainstream-chemistry type: experiment access: OA direction: supports

Both opened in full via PMC;  Joshi et al. is post-cutoff and CC BY.  Coacervate-supported membranes stay intact and impermeable at 25 mM and 50 mM magnesium, where plain fatty-acid and blended membranes are disrupted.  That is inside the RNA-copying range for the first time.

The caveat is the polycation:  poly(allylamine hydrochloride) is synthetic.  And the 2026 follow-up converges independently with Jordan et al. 2019 on the same finding — compositional heterogeneity beats pure compounds — now reproduced by two laboratories with opposite environmental commitments.

Did the exposure of coacervate droplets to rain make them the first stable protocells?

Agrawal, Aman, et al. (incl. Szostak).  Science Advances 10 (2024): eadn9657.

area: protocells camp: mainstream-chemistry type: experiment access: OA direction: complicates

CC BY, opened in full.  The standing objection to coacervates as protocells is that their contents exchange too fast to evolve — Jia, Hentrich and Szostak, OLEB 44 (2014), abstract only, measured it and concluded such systems "would not allow effective segregation and consequent evolution of RNA."  Baseline exchange times are about a minute for a 6-mer and five minutes for a 49-mer.

Rainwater transfer crosslinks the droplet surface and fixes the long end:  RNA of 35 nucleotides or more stays compartmentalised for days.  Three caveats from the authors.  The polycation is synthetic and "should apply" to natural ones is a promissory note.  The droplets are unstable above 1.3 mM sodium chloride equivalent.  And the boundary is a skin rather than a membrane — short RNAs still exchange in minutes.

What the compartment literature still lacks

area: protocells camp: mainstream-chemistry type: review direction: undercuts

Two more sources bear on the supply side, and one is this project's own arithmetic.

Lai et al., Icarus 319 (2019): 685-700, opened in full.  Meteoritic fatty acids are straight-chain and stop at twelve carbons;  abundance falls with chain length;  chains of eight or more are up to two orders of magnitude scarcer than short ones, but still around 10⁴-10⁵ ppb in Murchison and Tagish Lake.  The authors compare fatty acid abundance to other meteoritic organics, which is the favourable comparison, not to the critical bilayer concentration, which is the relevant one.

The relevant comparison is this project's own and is flagged as such.  Dissolving a kilogram of Murchison-grade meteorite in a litre of water gives about 0.58 mM decanoic acid, against critical bilayer concentrations of 39-150 mM for a single acid or 1.3 mM for a twelve-component mixture.  No published paper performs this comparison, the shortfall is roughly thirty-five to two hundred and fifty fold for a single acid, and mixture complexity plus evaporative concentration could close it.  Check it before printing it, and note that Cohen et al. list three alternative sources — meteorites, metal-surface synthesis, electrochemical sparking — so meteoritic delivery may not even be the intended route.

Shvetsova and Fiore, Life 16 (2026): 497, CC BY, opened in full.  Post-cutoff.  The current survey of abiotic lipid routes:  Fischer-Tropsch chemistry at 175 °C gives C2-C35 mixtures, oxalic acid at 150-250 °C gives C12-C33 lipids, and the field's own verdict is that "reconstructing the nature of the earliest compartments remains highly challenging."

Nunes Palmeira et al., PLoS Biology 24 (2026): e3003544, CC BY, opened in full.  Post-cutoff, and a model rather than an experiment.  Growth-supporting functions must be easier to attain than informational ones, so "growth precedes information at the origin of life."  Cite it as evidence about how the field is thinking, not about what protocells do.

One thing nobody found.  No 2026 experimental protocell couples division and replication.  Europe PMC was searched for 2026 protocell, vesicle and coacervate papers and the Szostak lab's full 2025-2026 output was checked.  Absence of evidence from one database and one laboratory is not proof, and a preprint-server sweep would be a useful check.  Karabasoglu et al., Astrobiology 26 (2026): 227-246, was not opened — SAGE bot-blocks — and it is the Toparlak group's own 2026 assessment of what a replicating protocell still lacks, so it is the most likely place to find a statement that supersedes Joyce and Szostak 2018.  Somebody with journal access should get it.

Metabolism first, and hydrothermal vents

The most useful fact about this literature is that its camps refute each other.  Orgel demolished self-organising metabolic cycles.  Jackson attacked the vent bioenergetics.  Damer and Deamer attacked vents on membranes and water activity.  Russell attacked hot-spring ponds on geology.  Mulkidjanian attacked both on ionic composition.  Every one of those authors is a committed naturalist, and none of them is arguing against natural origins.  Quoting any one of them as a witness against abiogenesis inverts what he was doing, and a reader who checks will see it immediately.

Before enzymes and templates: theory of surface metabolism

Wächtershäuser, Günter.  Microbiological Reviews 52 (1988): 452-484.

area: metabolism-first camp: mainstream-chemistry type: philosophy-paper access: paywalled direction: complicates

Opened in full from the library PDF;  a 1988 scan with imperfect OCR, and every quoted phrase was checked character by character.

The iron-sulfur world.  The energetic core is "pyrite pulling":  FeS + H₂S → FeS₂ + H₂ at ΔG° = −41.9 kJ/mol drives CO₂ + H₂ → HCOOH at +30.2 kJ/mol, for a combined −11.7 kJ/mol, with the carboxylate products bound to the positively charged pyrite surface.

His own laboratory delivered on parts of it.  Huber and Wächtershäuser, Science 276 (1997) and 281 (1998) — both abstract only — made an activated thioester and thence acetic acid on an NiS/FeS slurry, and made peptides from amino acids on (Ni,Fe)S with CO at 100 °C.  Note the inputs:  carbon monoxide and methanethiol, not CO₂ and H₂.

Self-organizing biochemical cycles · The Implausibility of Metabolic Cycles on the Prebiotic Earth

Orgel, Leslie E.  PNAS 97 (2000): 12503-12507.  ·  and  ·  PLoS Biology 6, no. 1 (2008): e18.

area: metabolism-first camp: mainstream-chemistry type: review access: OA direction: undercuts

Both opened in full;  the 2008 paper read line by line.  The strongest technical objection in this section, and it is posthumous — Orgel died in October 2007 and Gerald Joyce submitted the paper.

The 2008 argument, in numbers.  The reverse citric acid cycle has at least eleven reactions;  completing it at 50% overall efficiency requires better than 90% average efficiency per step;  it needs at least six distinct catalytic activities;  and the uncatalysed malate-fumarate interconversion has a half-time of more than ten thousand years at room temperature.  The conclusion is the sentence to quote:  "Lack of specificity rather than inadequate efficiency may be the predominant barrier to the existence of complex autocatalytic cycles of almost any kind."

The 2000 paper is blunter — to postulate a suite of fortuitously catalysed reactions "is to appeal to magic," and the self-organisation of the reductive citric acid cycle "would be a near miracle."  That is Orgel's genuine, verified use of the phrase, about the citric acid cycle rather than about RNA or life, and it is not the sentence apologetics quotes.

Three limits Orgel states himself and that are routinely dropped.  The 2008 paper targets self-organising cycles, not hydrothermal geochemistry generally;  Orgel writes that a theory in which "metal sulfide-catalyzed reactions provided some or all of the organic molecules needed for the formation of a primitive genetic system would have many attractive features."  The 2000 paper says "one can be sure that the complete reverse citric acid cycle is out of range, but it is not obvious that some much simpler cycle relevant to the origin of life is impossible."  And the 2000 paper closes by pointing at the next experiment, not at the end of inquiry.

A trap in the 2008 paper.  It contains Orgel's remark that Ghadiri's experiments illustrate "intelligent design" of the input peptides.  His scare quotes mean the chemist's design of the starting materials.  Quoting that sentence in a discussion of the design movement would be the single worst thing an article on this subject could do, and it is listed here so nobody rediscovers it.

The independent mathematical version of the same objection is Vasas, Szathmáry and Santos, PNAS 107 (2010): 1470-1475 — abstract only, free at PMC, body not read — arguing that "replication of compositional information is so inaccurate that fitter compositional genomes cannot be maintained by selection… the system lacks evolvability."  Orgel plus Vasas is the strongest two-citation combination in this section, and Szathmáry is a mainstream evolutionary theorist, not a critic of the field.

Orgel's reconstruction of Kauffman's autocatalytic-set model — including the charge that Kauffman assumed 1 M concentrations per amino acid — rests on Orgel's summary.  Kauffman 1986 was never opened and no abstract was obtained.  Orgel is a hostile witness, and a fair article would have somebody check him against the original.

On the origins of cells

Martin, William, and Michael J. Russell.  Philosophical Transactions of the Royal Society B 358 (2003): 59-85.

area: vents camp: mainstream-chemistry type: philosophy-paper access: paywalled direction: complicates

Opened in full.  The framing paper for the alkaline vent programme:  life began in structured iron-sulfide precipitates across a redox, pH and temperature gradient.  Their own rhetorical challenge has since been quoted back at them by critics — "once autoclaved, a bowl of chicken soup left at any temperature will never bring forth life."

An Origin-of-Life Reactor to Simulate Alkaline Hydrothermal Vents

Herschy, Barry, et al. (Lane lab).  Journal of Molecular Evolution 79 (2014): 213-227.

area: vents camp: mainstream-chemistry type: experiment access: OA direction: complicates

CC BY, opened in full.  Alkaline fluid at pH 11 and 70 °C injected into an acidic ocean at pH 5 and 20 °C precipitates iron-sulfide, silicate and phosphate tubes.  Thermophoresis concentrated fluorescein by 2,500 to 5,000 times.

The yields are the point, and one of them is internally inconsistent.  The body text says formate in "low micromolar quantities";  the Figure 5a caption says "Estimated concentration is 50 µM."  Both are in the same paper — do not quote either number without the caveat.  Formaldehyde came in around 100 nM and was "variable and inconsistent between runs."  Ribose formed at 0.06% conversion of formaldehyde — but the run used 0.5 M formaldehyde, five million times the reactor's own output.

The authors call it a "preliminary proof of concept."

Natural pH Gradients in Hydrothermal Alkali Vents Were Unlikely to Have Played a Role in the Origin of Life · Ancient Living Organisms Escaping from, or Imprisoned in, the Vents?

Jackson, J. Baz.  Journal of Molecular Evolution 83 (2016): 1-11.  ·  and  ·  Life 7, no. 3 (2017): 36.

area: vents camp: mainstream-chemistry type: calculation access: OA direction: undercuts

Both opened in full, CC BY.  The best technical citations on this side of the ledger, and the author is a bioenergeticist with no commitment to any origin-of-life camp — he thanks Lane and Martin for discussion.

The 2016 paper.  No membranes have been observed at Lost City;  the inorganic barriers there are about 1 µm, more than two hundred times thicker than a 5 nm lipid bilayer;  and a 1 nm machine sitting in a 1 µm channel intercepts a pH difference of only about 0.004 units — 24 J/mol against the 24 kJ/mol available across the bulk gradient, a thousandfold shortfall.  He also identifies a sign error in published diagrams:  an electric potential is claimed across a membrane where no potential difference exists.

The 2017 paper adds the argument this research could find no published answer to.  Even granting everything, the ocean and vent-fluid volumes are too large for a membrane pump to establish a useful gradient, so "ancient organisms utilising a natural pH gradient would have been imprisoned in the vents, unable to escape."  The underlying point is about selection:  an ion pump has no function until the cell seals, and the cell cannot seal until it has a pump.  He also notes four or five incompatible membrane models circulating simultaneously in the same research group's publications.

A third Jackson paper — J. Mol. Evol. 85 (2017): 1-7, arguing that Herschy's formate was an equilibrium artefact — was never opened and no abstract was obtainable anywhere.  Springer redirects to authentication, it is absent from PubMed and Europe PMC, and Semantic Scholar reports the abstract elided by the publisher.  It is known only through Jackson's own one-sentence summary of it in his 2017 Life paper.  That is the specific charge Hudson et al. 2020 answers, so the account of this exchange rests on a paper nobody here has read.

Proton gradients at the origin of life

Lane, Nick.  BioEssays 39, no. 6 (2017): 1600217.

area: vents camp: mainstream-chemistry type: philosophy-paper access: preprint direction: complicates

Accepted manuscript opened in full from UCL Discovery.  Print Lane's complaint, because it is directed at exactly the use this catalogue is meant to prevent:  "Unfortunately, it is already being cited as evidence that life could not have started in alkaline hydrothermal vents.  That is not sustained in his paper."

His substantive replies.  Nobody proposed machines embedded in rock;  they sit in 5 nm lipid bilayers lining the pore walls, so "thickness… is practically irrelevant."  The prebiotic phase needs no machine at all, only electron transfer across the barrier rather than proton transfer.  And he corrects a log-scale error:  hydroxide at pH 10 is a thousand times more concentrated than protons at pH 7, so the ocean-side gradient is about three pH units rather than the vent side's 0.2.  He concedes his own modelling "assumes steep pH gradients across vent pores, which is plainly a best-case scenario," and concludes proton gradients "could in principle have helped power the origins of life."

Score it honestly.  The thickness and geometry objection is answered.  The Lost City observational objection and the "imprisoned in the vents" selection objection are not, and no reply by Lane, Martin, Russell or Sojo to the 2017 selection argument was found.  If somebody repeats that search and also finds none, the article may say it is unanswered.

CO₂ reduction driven by a pH gradient

Hudson, Reuben, et al.  PNAS 117 (2020): 22873-22879.

area: vents camp: mainstream-chemistry type: experiment access: OA direction: complicates

CC BY-NC, opened in full.  The experimental answer to the equilibrium-artefact charge, and the cleanest dose-response in this literature.

A microfluidic reactor, ocean side pH 3.9, vent side pH 12.3, a 30-60 µm iron-nickel-sulfide precipitate, 3.3 seconds residence:  formate at an average of 1.5 µM, confirmed by ¹³C labelling.  Formate rose monotonically with vent-side pH — undetectable, then 0.37, 0.9, 1.5 and 1.8 µM at pH 7.0, 9.1, 12.1, 12.3 and 12.6.  Removing the pH gradient, the precipitate or the hydrogen each killed the product entirely.

And the concession, theirs:  acetate and pyruvate "represent an as-yet unachieved benchmark."  No carbon-carbon bond was formed.  The rate, 2.5 × 10^-13 mol/s, is four orders of magnitude below an electrochemical cell.

A hydrogen-dependent geochemical analogue of primordial carbon and energy metabolism

Preiner, Martina, et al.  Nature Ecology & Evolution 4 (2020): 534-542.

area: metabolism-first camp: mainstream-chemistry type: experiment access: abstract-only direction: supports

Abstract only;  paywalled and the body not opened.  The highest yields in the field:  greigite, magnetite and awaruite catalyse CO₂ + H₂ at 100 °C in alkaline water to give formate up to 200 mM, acetate up to 100 µM, pyruvate up to 10 µM, methanol up to 100 µM, plus methane.

And it quietly complicates the theory it is usually cited to support, because it uses no pH gradient, no membrane and no compartment at all.  A hundred thousand times more formate than the gradient-driven reactor, from minerals in a flask.  That reading is an inference from the abstract — the methods name only minerals, H₂, CO₂ and 100 °C — and somebody should confirm it from the body before the article presses it.

Synthesis and breakdown of universal metabolic precursors promoted by iron

Muchowska, Kamila B., Sreejith J. Varma, and Joseph Moran.  Nature 569 (2019): 104-107.

area: metabolism-first camp: mainstream-chemistry type: experiment access: OA direction: supports

Author manuscript opened in full via Europe PMC.  The best experimental result in this section.

Pyruvate plus glyoxylate plus ferrous iron in water, at 70 °C for three hours, gives nine of the eleven Krebs cycle intermediates and all five universal metabolic precursors.  Add hydroxylamine and metallic iron at one hour and glycine, alanine, aspartate and glutamate appear by two hours.  Ferrous iron alone drives four unrelated reaction types — aldol and retro-aldol, hydration and dehydration, oxidative decarboxylation, and redox — in a single pot.

Four caveats, and the fourth is a nice irony.  It starts from pyruvate and glyoxylate, not CO₂.  It is an open network at 0.1-0.2 mmol scale needing continuous input, not a demonstrated autocatalytic cycle.  Selectivity is reported as a carbon balance rather than clean yields, and the per-run concentrations are in Supplementary Table S3, which was not opened — so "nine of eleven" is verified and "at what yield" is not.  And citrate, the molecule Orgel singled out as hardest, is one of the two missing intermediates.

Autocatalytic chemical networks at the origin of metabolism

Xavier, Joana C., Wim Hordijk, Stuart Kauffman, Mike Steel, and William F. Martin.  Proceedings of the Royal Society B 287 (2020): 20192377.

area: metabolism-first camp: mainstream-chemistry type: calculation access: OA direction: complicates

CC BY, opened in full.  Reflexively autocatalytic food-generated networks embedded in modern microbial metabolism, generating acetyl-CoA, amino acids and bases from H₂ and CO₂ with small-molecule catalysts.

The asymmetry in the result is the finding.  With cofactors supplied in the food set, a large network appears.  With amino acids and bases but no organic catalysts, only a "minute RAF with 33 reactions."

And it is an analysis of the KEGG database of modern metabolism, not an experiment and not a claim about the Hadean earth.  Cofactors such as NAD were placed in the food set by hand.  The paper also notes a real database limit:  "no reactions in KEGG have RNA catalysts assigned in Uniprot."

Clays and the Origin of Life: The Experiments

Kloprogge, J. Theo, and Hyman Hartman.  Life 12, no. 2 (2022): 259.

area: metabolism-first camp: mainstream-chemistry type: review access: OA direction: undercuts

CC BY, opened in full.  A first-hand account of a research programme collapsing, written by one of its participants.

Cairns-Smith's clay-genes hypothesis rested experimentally on Armin Weiss's 1981 claim of clay replication through ten generations.  It was never repeated;  the references it cited turned out to be untraceable or irrelevant, including an unpublished dissertation that Arrhenius and colleagues could not locate.  Hartman, writing as co-author, records that he "was furious that he had invested time and NASA money in what had turned out to be a fraud."  The joint correction letter — Arrhenius, Cairns-Smith, Hartman, Miller and Orgel, Angew. Chem. 25 (1986): 605-606 — was never opened and is known only through this account.

Do not conflate two different claims.  Clay-as-genetic-material collapsed.  Clay-as-catalyst did not:  Ferris et al.'s montmorillonite oligomerisation stands, though Orgel notes it is only about 80% regiospecific — worse than the uncatalysed 2′-5′ preference.

Cairns-Smith's own papers were never opened, and neither was his 1982 book.  Everything about the model here comes through Kloprogge and Hartman.  The one attempt to test crystal inheritance directly — Bullard et al., Faraday Discussions 136 (2007): 231-245 — was reached as an abstract only, and the abstract does not state a result.  The article must not claim that paper falsified Cairns-Smith.  The recorded discussion at pages 309-328 may contain the verdict and nobody here has read it.

The camps refuting each other

area: vents camp: mainstream-chemistry type: review direction: complicates

This is the most useful group of citations in the section and the least used by either side.

Damer and Deamer, Astrobiology 20 (2020): 429-452.  CC BY-NC, opened in full.  Section 3.3 is a critique of vents by the hot-spring camp:  a proton gradient one micrometre thick has almost no energy available, membranes do not assemble at seawater ionic strength, and vents offer no source of activated monomers because they sit in equilibrium with the ocean bulk.  Their reply to Jordan et al. is honest and non-dismissive — they do not dispute the chemistry, only note that "a plausible prebiotic source of such mixtures remains uncertain."  They do not engage Lane's 2017 rebuttal despite using the same geometry argument as Jackson.

Russell, Life 11, no. 5 (2021): 429.  CC BY, opened in full.  The vent camp firing back on geology:  the early ocean "was never less than 4 km deep," there were no Icelands or Hawaiis, and the magma beneath was too mushy to support them.  He calls the warm-little-pond idea "a reductionist fantasy."  He also quotes Sutherland's own verdict against his own camp:  a requirement for ultraviolet light "would rule out deep sea environments… the idea that life originated at vents should, like the vents themselves, remain 'in the deep bosom of the ocean buried.'"  (That Sutherland quotation is relayed through Russell and was not independently opened.)

Mulkidjanian et al., PNAS 109 (2012): E821-E830.  Opened in full.  A third camp against both:  cells contain far more potassium, phosphate and transition metals than any ocean, and "no marine environment could ever provide" that composition.  Their answer is inland geothermal vapour-condensate fields.

Three mainstream camps, each with a technical argument that the other two cannot answer.  That is the honest state of this subfield, and it is more informative than any single camp's case.

Recent work, and what was never chased

González-Martínez et al., Nature Communications 17 (2026): 6115 — abstract only, post-cutoff.  A prebiotic NAD⁺ analogue mediates oxidative decarboxylation and reductive amination without enzymes, producing succinate, acetate, formate, glutamate, alanine and glycine from α-ketoacids, and identifies the structural features that rationalise nicotinamide's selection.  A partial answer to Orgel's specificity problem — one cofactor coupling an oxidation to a reduction — not a closed cycle.

McAnally et al., PNAS 122 (2025): e2501839122 — abstract only.  Laboratory simulation of dense molecular cloud conditions produces "the complete suite of biorelevant molecules central to the Krebs cycle."  A fourth environment, and it changes the shape of the question if the intermediates are delivered rather than built in place.

Never opened, and named so nobody re-chases them:  Kauffman 1986;  Cairns-Smith 1966 and 1982;  Arrhenius et al. 1986;  Jackson's 2017 J. Mol. Evol. paper;  Bullard et al. 2007's actual result;  Wächtershäuser 2007 (Chem. Biodivers. 4: 584-602), on which rests Orgel's strong historical claim that Wächtershäuser abandoned surface metabolism and the reverse citric acid cycle — that claim currently stands on one hostile summary and should be confirmed;  Wächtershäuser 2016, "In Praise of Error" (J. Mol. Evol. 82: 75-80), cited by Damer and Deamer as a critique of the vent programme, which if it is what it appears to be is the founder of one metabolism-first school attacking the other;  and a 2018 paper by Ross on vent thermodynamics that could not even be located with a full citation.

The genetic code and the ribosome

Where the codon table came from is a separate problem from where the molecules came from, and it is less often argued carefully.  The "one in a million" result is misreported in both directions, and the misreport turns on a denominator most treatments never state.

The Genetic Code Is One in a Million

Freeland, Stephen J., and Laurence D. Hurst.  Journal of Molecular Evolution 47 (1998): 238-248.

area: genetic-code camp: mainstream-chemistry type: calculation access: abstract-only direction: complicates

Abstract only, verbatim from PubMed;  the body is paywalled and was never opened.

They built about a million alternative codes, scored each on a translation-error cost function weighted for transition/transversion bias and position-dependent mistranslation, and found that only one in a million beat the real code.

The load-bearing detail, and it is almost never stated.  The random codes are not arbitrary mappings of 64 codons onto 21 meanings.  They are built by shuffling the twenty amino acids among the standard code's own synonymous codon blocks, keeping block structure and degeneracy fixed.  That space has 20! ≈ 2.4 × 10¹⁸ members, not the ~10⁸⁴ of the unrestricted space.

Three independent sources opened in full confirm the constraint — Haig and Hurst 1991 (the paper being extended, abstract only), Koonin and Novozhilov 2009, and Miyachi and Ichihashi 2026.

So both standard misreports are wrong, and the arithmetic shows why.  A hit rate of 10^-6 inside a space of 2.4 × 10¹⁸ implies roughly 2.4 × 10¹² codes that beat the standard code — over two trillion.  "One in a million" is a percentile, not a rank.  Against that, the code really is in the top millionth of its structural class, which is not nothing.

Early Fixation of an Optimal Genetic Code

Freeland, Stephen J., Robin D. Knight, Laura F. Landweber, and Laurence D. Hurst.  Molecular Biology and Evolution 17 (2000): 511-518.

area: genetic-code camp: mainstream-chemistry type: calculation access: abstract-only direction: complicates

Abstract only (OUP bot-blocks).  Restricting code space to "plausible biological constraints" and using empirical substitution-frequency data, the canonical code is "at or very close to a global optimum for error minimization," and "significantly better codes… are extremely rare."

Do not print this next to Koonin's 70-78% minimisation percentage as though they were in conflict.  They use different constraint sets and different measures of amino acid similarity.  A frequently circulated body quote from this paper — a "96-100% optimization" figure — could not be verified and must not be printed without opening the paper.

Evolution of the genetic code: partial optimization of a random code

Novozhilov, Artem S., Yuri I. Wolf, and Eugene V. Koonin.  Biology Direct 2 (2007): 24.

area: genetic-code camp: mainstream-chemistry type: calculation access: OA direction: supports

CC BY, opened in full.  The result that undercuts "the code is unreachable by search," and it is from the same laboratory that flags the one-in-a-million misreport.

Working inside the same block-structured space, the standard code is reachable from a random code in ten to twelve swaps on average, and it is not even a local peak — one further swap improves it.  Their sentence:  the standard code "appears to be a point on an evolutionary trajectory… about half the way to the summit of the local peak."

Origin and evolution of the genetic code: the universal enigma · Frozen Accident Pushing 50

Koonin, Eugene V., and Artem S. Novozhilov.  IUBMB Life 61 (2009): 99-111.  ·  and  ·  Koonin, Eugene V.  Life 7 (2017): 22.

area: genetic-code camp: atheist-explicit type: review access: OA direction: undercuts

Both opened in full — the 2009 paper as a PMC author manuscript, the 2017 as CC BY.  The single strongest citation on this topic, and it comes from the field's leading reviewer rather than from a critic.

The minimisation percentage — how far the standard code has travelled from an average random code toward the best code found — is 70% on the polar requirement scale and 78% relative to a locally optimised code.  The 2017 paper gives the hit rate as 10^-7 to 10^-8 within the block-structured space, and states the denominator problem in the same sentence:  "among the ~10⁸⁴ possible random codes, there is a huge number with a higher degree of error minimization than the SGC."  Mixing those two denominators is the commonest error in popular treatments, on both sides.

The methodological warning applies to every optimality figure ever quoted on this subject:  rearranging the existing table "allows formal… estimation of robustness but not the evolutionary processes that lead to it."  Koonin adds the harder version:  if the cost function is a poor proxy for the real optimisation criterion, "the general assessment of the entire field… would have to be particularly somber."

And the verdict:  fifty years of code studies have elucidated "notable features of the standard code, such as high robustness to errors, but failed to develop a compelling explanation for codon assignments."  From 2009:  the question "why is the genetic code the way it is and how did it come to be… might remain pertinent even in another 50 years."

Experimental verification of the error minimization theory using non-standard genetic codes constructed in vitro

Miyachi and Ichihashi.  2026.

area: genetic-code camp: mainstream-chemistry type: experiment access: OA direction: complicates

Opened in full from the on-disk PDF.  Post-cutoff, and the most important entry in this section, because every optimality number above it is computational and this is the first direct experimental test.

Ten non-standard genetic codes were built in a reconstituted PURE translation system with 21 tRNAs, reassigning alanine, serine and leucine.  Three reporter genes were mutagenised and the loss of protein activity measured under each code.

The prediction failed.  In their words:  "In contrast to this prediction, translation experiments using the non-SGCs and mutagenized libraries revealed that the effects of mutations on protein functions were similar across all genetic codes and reporter genes examined here."

And the same paper contains a result that cuts the other way, so honest reporting needs both.  When three cost measures were combined into a single index, the near-standard code came out lowest of all 19,683 codes constructible in their system.  So the code is the multi-property optimum, and the robustness that optimum is supposed to buy could not be detected in the experiment.

The authors are careful about reach and the article should be too:  the finding is limited to individual reporter proteins in a reconstituted system, only three amino acids could be reassigned, and whether it holds at the level of cellular fitness or over evolutionary time "remains an open question."

The stereochemical theory, and its author's own retreat

Yarus, Michael, J. Jeffrey Widmann, and Rob Knight.  Journal of Molecular Evolution 69 (2009): 406-429.  ·  and  ·  Yarus, Michael.  Life 7 (2017): 13.

area: genetic-code camp: mainstream-chemistry type: calculation access: OA direction: complicates

Both opened in full.  The 2009 paper reports 337 independent RNA binding sites for eight amino acids with cognate triplets enriched at P ≈ 5.3 × 10^-45 for codons and 2.1 × 10^-46 for anticodons, and concludes that about 75% of modern amino acids "entered the code" stereochemically.  The limit is stated in the same abstract and is never quoted:  only 21% of possible triplets appear in binding sites at all.

By 2017 Yarus had walked the figure back himself:  "75% overstates the results… the majority of triplets may have entered the code another way, rather than via RNA-amino acid specificity."  Only 12 of 48 possible associations were detected in the fuller survey.

Koonin's objection is decisive on its own terms:  the stereochemical signal is strongest for arginine, tryptophan, tyrosine, histidine and isoleucine — precisely the amino acids prebiotic chemistry does not make in quantity.  The signal is strongest for the amino acids that arrived last.  Koonin and Novozhilov put it more bluntly:  "many years of experimentation including the latest extensive studies on aptamer selection show that the code is not based on a straightforward stereochemical correspondence between amino acids and their cognate codons."

The coevolution theory, and what happens when its inputs are made empirical

Wong, J. Tze-Fei.  PNAS 72 (1975): 1909-1912.

area: genetic-code camp: mainstream-chemistry type: calculation access: closed direction: complicates

Never opened — PNAS bot-blocks and the PMC PDF route returned an interstitial;  the one-sentence abstract came from NCBI.  The theory is that code structure reflects the order in which amino acid biosynthesis pathways emerged.

The numbers that decide it are in Koonin and Novozhilov 2009, which was opened, reporting Amirnovin's tests.  With Wong's original eight precursor-product pairs, the standard code beats 99.9% of random codes.  Remove two contested pairs and 3.6% of random codes do better.  Use E. coli's actual characterised metabolism instead of inferred primordial pairs, and 34% of random codes score better than the real one.  A wobble-corrected recount removed the statistical support entirely.

The signal nearly disappears when the precursor list stops being assumed.

The frozen accident, and what has replaced it

Crick, Francis H. C.  Journal of Molecular Biology 38 (1968): 367-379.

area: genetic-code camp: mainstream-chemistry type: philosophy-paper access: closed direction: complicates

Never opened;  Elsevier paywalled and PubMed carries no abstract.  Everything here is quoted second-hand through Koonin 2017.  The claim:  any code change would now be lethal because it corrupts every evolved protein at once, so universality implies a single ancestor and the assignments were "entirely a matter of chance."

The "instantly lethal" framing is falsified and the article must not use it.  Living recoded organisms exist, and at least 27 distinct translation-table variants are catalogued.  The defensible version is a fitness barrier, not an impossibility.

Two mainstream alternatives have grown up around it, and neither was opened.  Vetsigian, Woese and Goldenfeld, PNAS 103 (2006), argue the code was fixed by horizontal gene transfer, because a code change is prohibitive for genetic exchange — not opened, PNAS bot-blocks.  And Worning and Ibarra-Chávez, Trends in Genetics (2026) — abstract only, post-cutoff — argue the same, more strongly:  gene sharing rather than frozen accident explains near-universality, and they supply the hard number of 27 NCBI translation-table variants.

The variant codes, measured

Shulgina, Yekaterina, and Sean R. Eddy.  eLife 10 (2021): e71402.  ·  McGowan, Jamie, et al.  PLOS Genetics 19 (2023): e1010913.  ·  DeBenedictis, Erika A., Dieter Söll, and Kevin M. Esvelt.  eLife 11 (2022): e76941.

area: genetic-code camp: mainstream-chemistry type: measurement access: OA direction: supports

All three opened in full.  Together they kill "the code never changes" and support "changes are marginal," which are different claims.

Shulgina and Eddy screened 251,571 genomes covering 48,693 unique species and found five new arginine-codon reassignments — the first sense-codon changes documented in bacteria.  The mechanism is the interesting part:  reassignment follows codon capture, meaning codons are reassigned only after genome-wide pressure such as low GC content or genome reduction has driven them near extinction.

McGowan et al. broke a rule everyone thought was solid:  an uncultured ciliate reassigns UAA to lysine and UAG to glutamate, decoupling the two stop codons.

DeBenedictis, Söll and Esvelt answer the "could it have been otherwise" question experimentally.  All twenty canonical tRNA isoacceptor classes can be converted to functional quadruplet-codon tRNAs, though only nine of twenty aminoacyl-tRNA synthetases tolerate quadruplet anticodons.  Their conclusion:  the "triplet codon code was selected because it is simpler and sufficient, not because a quadruplet codon code is unachievable."

And the cost side, post-cutoff.  Nyerges et al., Nature Communications 17 (2026): 5311, opened in full, pushed six E. coli strains up to 45.8% of a synthetic genome on a deliberately deleterious 57-codon code and found widespread defects:  "synonymous codon replacement — recoding — is frequently lethal."  Changing the code is possible and expensive.  That is the fitness-barrier picture, not the impossibility picture and not the free-parameter picture.

The ribosome's derivable history

area: ribosome camp: mainstream-chemistry type: calculation direction: supports

The claim that the ribosome has no reconstructible history is false, and the article should say so plainly.

Bokov, Konstantin, and Sergey V. Steinberg, Nature 457 (2009): 977-980.  Never opened by anyone on this project.  No open-access copy exists anywhere — Unpaywall, OpenAlex and Semantic Scholar all confirm it, ResearchGate bot-blocks — and everything here comes from the PubMed abstract plus three papers that describe the method and were opened in full.  Using A-minor motif dependency asymmetry as a timing signal, they derive an order of addition for 23S rRNA and strip the molecule back to a small core around the peptidyl transferase centre.

Five limits.  The peptidyl transferase centre's primacy is assumed rather than derived — Fox 2010, opened in full, says so.  It covers the large subunit only.  Fox notes that "acquisition of the GTPase center and L1 protuberance and even the addition of the 30S subunit are relatively late."  And it is contested from inside the mainstream:  Caetano-Anollés, J. Mol. Evol. 80 (2015) and a 2017 commentary, both opened in full, argue that a branch-to-trunk insertion assignment error contradicts the peptidyl-transferase-first model and that translocation is the ancestral function instead.

Petrov et al., PNAS 111 (2014) and 112 (2015), both opened in full, reach a converging decomposition by a different method with different assumptions, and count 59 ancestral expansion segments in the large subunit and 27 in the small.  Two independent methods converging is the reason this counts as a result rather than a story.

Bose et al. (Yonath lab), Nucleic Acids Research 50 (2022): 1815-1828.  Opened in full via PMC.  Cut-down, dimerised reconstructions of the peptidyl transferase centre's symmetric pocket — "protoribosomes" — form peptide bonds, verified by MALDI-TOF and LC-MS.  The authors state the limit themselves:  the construct "did not possess the ability to support substrate motion from one site… to the other, as required in the contemporary mode of protein biosynthesis."  It makes a bond;  it cannot make a chain;  and the chemistry is uncoded.  (A 2022 correction to this paper is an attribution and reference fix, not a retraction — confirmed by direct read.)  Agmon 2024, opened in full, reports that three independent groups have replicated the dimeric protoribosome result.

A symbiotic origin of the ribosome?

Lynch, Michael, and Andrew Ellington.  PNAS Nexus 5 (2026): pgag019.

area: ribosome camp: mainstream-chemistry type: philosophy-paper access: OA direction: undercuts

CC BY-NC, opened in full.  Post-cutoff, and the best-sourced "this is unsolved" citation in this section, because of who wrote it — Michael Lynch is among the most cited evolutionary geneticists alive — and how recent it is.

"The origin of the translation system… remains the most enigmatic" of the unsolved origin-of-life problems.  Their hypothesis is that the protoribosome began as a molecular parasite of an RNA-world host and that translation emerged from host-parasite coevolution, explicitly as a way around the chicken-and-egg problem.

The number they supply is the useful part.  Non-enzymatic templated RNA replication runs at in-vitro error rates of 0.01 to 0.1 per nucleotide site;  modern RNA polymerases run at 10^-6 to 10^-4.  That is a gap of two to four orders of magnitude between demonstrated prebiotic copying fidelity and what a translation system needs.

Their own limits are stated:  "exact answers may be beyond reach," and "short of a time machine, we may never know with certainty how the pre-LUCA endowments of life became established."  The hypothesis is offered as testable, not proven.

Breaking the synthetase circularity

area: genetic-code camp: mainstream-chemistry type: experiment direction: supports

The circle is that aminoacyl-tRNA synthetases are proteins made by translation, and translation needs synthetases.  The strong form of that objection — "only a protein can charge a tRNA" — has been dead since 1995, and the article should not run it.

Illangasekare et al., Science 267 (1995): 643-647 — abstract only.  A selected ribozyme aminoacylates its own terminus, accelerating the reaction at least a hundred-thousandfold.  Xiao et al., Nature 454 (2008): 358-361 — abstract only.  The flexizyme is a 45-nucleotide ribozyme with more than 10⁵-fold rate enhancement and more than 10⁴-fold discrimination against some non-cognate amino acids.  Lee et al., Nature Communications 10 (2019): 5097 — abstract only.  Thirty-two of thirty-seven non-canonical substrates acylated and incorporated by in vitro translation.

What none of them supplies is a codon assignment.  Flexizymes are promiscuous by design;  they are charging catalysts, not coding catalysts.  They dissolve the "must be protein" step and leave the "which amino acid on which anticodon" rule untouched.  Chen, Chen and Zhu 2021, never opened and with no abstract obtainable, is characterised through Miyachi and Ichihashi's summary:  "protein synthesis using ribozyme-based aminoacylation is still a challenge."

Two further lines belong here.  Radakovic et al., Science Advances 11 (2025), opened in full, built a self-assembling amino-acid-bridged ribozyme sustained over indefinite serial dilution — but there the amino acid staples two RNA fragments together, which is aminoacylation in the service of RNA rather than of coding.  And the urzyme programme of Charles Carter — Li, Francklyn and Carter, JBC 288 (2013), and Martinez-Rodriguez et al., JBC 290 (2015), both opened in full — builds 120-130 residue synthetase cores that acylate tRNA about a millionfold faster than uncatalysed peptide synthesis while retaining ~60% of modern catalytic proficiency, and 46-residue peptides coded by opposite strands of the same designed gene, both of which activate amino acids.  Their own concession is the honest one:  "Urzymes appear highly evolved, implying that they had even simpler ancestors," and those ancestors are undiscovered.  Carter and Wills, MBE 35 (2018), opened in full, argue formally that the synthetases must be interdependent.  This is a serious minority dissent from the RNA-world premise, made from inside the mainstream, and the article should present it as such.

Amino acid recruitment order

Wehbi, Sawsan, et al.  PNAS 121 (2024): e2410311121.  ·  and  ·  Hashmi, S. A., et al.  Genome Biology and Evolution 18 (2026): evag012.  ·  and  ·  Goldman, Aaron D., Gregory P. Fournier, and Betül Kaçar.  Cell Genomics 6 (2026): 101140.

area: genetic-code camp: mainstream-chemistry type: calculation access: OA direction: supports

All three opened in full;  the last two are post-cutoff.  (Hashmi et al. has a June 2026 correction that was not read — flag it before citing.)

Wehbi et al. show that the textbook recruitment order was partly an artefact of Miller's apparatus lacking sulfur:  metal-binding and sulfur-containing amino acids were recruited much earlier than thought, smaller ones earlier still, and ancestral pre-LUCA sequences carry amino acid frequency signatures hinting at "earlier, alternative genetic codes."  That last claim is an inference from reconstructed frequencies, not a recovered ancient code.  Hashmi et al. reach a converging chronology from sixteen independent molecular-complexity metrics — an independent line of evidence, derived from chemistry rather than from a spark-discharge apparatus.

Goldman, Fournier and Kaçar supply direct sequence evidence of the code being built.  The pre-LUCA ancestor of the tryptophan and tyrosine synthetases is "devoid of tryptophan amino acids" even though its descendants contain it — tryptophan entered the code late.  And the pre-LUCA leucine/isoleucine/valine synthetase ancestor "did not discriminate between these three amino acids."  The code was still under construction after protein synthesis was already working, which is a fact neither side's simple story accommodates.

Resurrected LUCA tRNAs have now been tested:  Yamanouchi, Moriya and Yokobori, J. Mol. Evol. (2026) — abstract only, post-cutoff — synthesised reconstructed LUCA tRNA-Ala and tRNA-Ser variants and found they decode ACN codons at efficiencies comparable to native E. coli tRNA-Thr, with decoding patterns that "differed from Crick's and extended wobble rules."  Narrow — two tRNA species in a wholly modern system — but it shows ancient translation is not beyond experimental reach.

A biophysical basis for the emergence of the genetic code in protocells

Harrison, Stuart A., Raquel Nunes Palmeira, Aaron Halpern, and Nick Lane.  BBA Bioenergetics 1863 (2022): 148597.

area: genetic-code camp: mainstream-chemistry type: philosophy-paper access: OA direction: complicates

Accepted manuscript opened in full via UCL Discovery.  The most ambitious retrodiction of the codon table from chemistry, and it is a model.

Three rules are claimed to predict most codon assignments:  the first base tracks metabolic distance from CO₂ fixation, the second anticodon base tracks hydrophobicity, and third-position redundancy tracks side-chain length.

The authors' own status report is the sentence to quote:  "Further work to confirm each rule… is underway."  And the hypothesis names its own missing pieces — proto-mRNA, proto-tRNA and the proto-ribosome are all listed as absent from it.

A related proposal, Agmon 2022 (opened in full via PMC), sets out proto-tRNA self-assembly and self-aminoacylation from acceptor-stem coding triplets.  Its verbs are the tell:  "evoke the possibility," "proposed to underlie."  The experiments were not performed.

The honest summary for the article:  four families of theory compete — stereochemical affinity, coevolution with biosynthesis, error minimisation, and the frozen accident.  Each explains part of the pattern.  None explains the actual assignments.  They are not four facets of one settled account, and Koonin says so in print.

The minimal cell and LUCA

These are the two ways to measure the far end of the gap — the simplest thing that has ever lived, and the reconstructed last common ancestor.  Both numbers are softer than either side usually admits, and in the LUCA case the disagreement between studies is itself the finding.

Design and synthesis of a minimal bacterial genome

Hutchison, Clyde A., III, et al. (J. Craig Venter Institute).  Science 351 (2016): aad6253.

area: minimal-cell camp: mainstream-chemistry type: experiment access: paywalled direction: complicates

Opened in full from the on-disk text.  The most-cited number in this part of the argument, and the one that most needs its caveats printed first.

JCVI-syn3.0:  531,560 base pairs, 473 genes — 438 protein-coding and 35 RNA — doubling every 180 minutes.  Smaller than any autonomously replicating cell found in nature.  For scale:  syn1.0 was 1,078,809 bp and 901 genes at 60 minutes;  M. genitalium is 580 kbp and 525 genes at 16 hours.

The famous sentence, printed twice by the authors:  "Unexpectedly, it also contains 149 genes with unknown biological functions, suggesting the presence of undiscovered functions that are essential for life."  Their confidence breakdown of all 473:  equivalog 232, probable 58, putative 34, generic 84, unknown 65 — the last two summing to the 149, about 31% of the genome.  A stricter and sharper version of the same claim is also in the paper:  79 genes (17%) could not be assigned even to one of four broad functional categories;  19 of those are essential;  and 13 of those are of completely unknown function.

Three caveats belong in front of the number, not behind it.

It is not a claim about the first cell.  The authors state it plainly:  "the genetic requirements for survival, and therefore the minimal genome size, depend on the environment in which the cell is grown.  The work described here has been conducted in medium that supplies virtually all the small molecules required for life."  Syn3.0 grows in SP4 medium with serum, yeast extract and peptones, imports amino acids and peptides, and salvages nucleotides and lipids rather than making them.

Nothing was built from chemistry.  Every syn3.0 genome was booted by transplantation into a living M. capricolum cell whose own genome was then removed.  Ribosomes, membrane, cytoplasm and machinery were all inherited from something already alive.

473 is not a floor.  "More genes could probably be removed… but growth rate would be compromised."

And the result the article should pair with the 149, because it is the more interesting one.  Top-down design from first principles failed.  The hypothetical minimal genome built from the literature and transposon data was non-viable — seven of eight segments would not support life.  Their sentence:  "we did not have sufficient knowledge to design a functional minimal genome from first principles."  They also record being puzzled by six annotated efflux systems:  "It is somewhat disconcerting to imagine that all of these exclude or remove toxic substances."

Genetic requirements for cell division in a genomically minimal cell

Pelletier, James F., et al.  bioRxiv 2020.10.07.326892;  published as Cell 184 (2021): 2430-2440.

area: minimal-cell camp: mainstream-chemistry type: experiment access: preprint direction: complicates

The CC0 preprint was opened in full.  The peer-reviewed Cell version was never opened by anyone on this project, and the two differ in one number:  the preprint abstract says "five genes of unknown function" while its body says "the three genes in the dcw cluster, as well as the four genes of unknown function outside this locus."  Those reconcile — one dcw gene is a hydrolase of unknown substrate — but somebody should confirm against the published paper before the article prints "five."

Adding nineteen genes to syn3.0 restores near-normal division, and seven of the nineteen — including ftsZ, sepF, and five genes of unknown function — are jointly necessary and sufficient.  No single one is dispensable, and ftsZ alone is insufficient while being dispensable once the full set is present.

That is an irreducibility of phenotype under a specific genomic background, not irreducible complexity in the argumentative sense.  Syn3.0 still divides;  it divides badly.

Essential metabolism for a minimal cell · Improving the annotations of JCVI-Syn3a proteins

Breuer, Marian, et al.  eLife 8 (2019): e36842.  ·  and  ·  Kilinc, Mesih, Kejue Jia, and Robert L. Jernigan.  Methods in Molecular Biology 2867 (2025): 153-168.

area: minimal-cell camp: mainstream-chemistry type: calculation access: OA direction: complicates

Breuer et al. CC BY, opened in full.  Kilinc et al. abstract only, chapter not read.

The unknowns figure has moved, and "149 unknowns, unchanged since 2016" can no longer be written honestly.

Breuer et al. built a metabolic model of syn3A — 543 kbp, 493 genes — covering 338 reactions and 155 genes.  91 genes (20%) are of unclear function, of which 30 are essential.  For comparison they give M. pneumoniae at 45% unclear and E. coli at 38%:  the minimal cell has the smallest proportion of unknowns of any genome they compared.  Attach each figure to its own paper;  473/149 and 493/91 are different strains and different annotation rounds and must not be blended.

Kilinc et al. assigned annotated homologs to 93% of the minimal genome using protein language models.  That is a computational prediction, not an experiment — a remote-homology hit is a hypothesis about function, and this research does not know what accuracy threshold "high level of accuracy" refers to, nor how many of Breuer's thirty essential unknowns were actually resolved.  No paper was found that systematically determines the biochemical function of the 149 experimentally.

Evolution of a minimal cell

Moger-Reischer, R. Z., et al.  Nature 620 (2023): 122-127.

area: minimal-cell camp: mainstream-chemistry type: experiment access: OA direction: supports

CC BY;  PDF downloaded, abstract read verbatim, body skimmed.  Streamlining cost the minimal cell more than 50% of its fitness — and it regained the entire deficit within 2,000 generations, evolving 39% faster in relative fitness than the non-minimal cell.

This kills any argument that the minimal cell sits on a knife-edge of viability.  A 2026 preprint (abstract only, post-cutoff, not peer reviewed) adds that syn3B evolved to grow at 25 °C, which it previously could not, in forty passages:  "even a minimal cell retains substantial evolvability."

The smallest natural genomes, and why they prove nothing here

Nakabachi, Atsushi, et al.  Science 314 (2006): 267.  ·  Bennett, Gordon M., and Nancy A. Moran.  Genome Biology and Evolution 5 (2013): 1675-1688.  ·  Michalik, Anna, et al.  Nature Communications 17 (2026): 2473.

area: minimal-cell camp: mainstream-chemistry type: measurement access: OA direction: complicates

All three opened in full;  Nakabachi including both errata;  Michalik is post-cutoff and CC BY.

Carsonella ruddii:  159,662 bp, 182 ORFs, 90% of them overlapping, 97.3% coding density, with total loss of cell-envelope and nucleotide and lipid biosynthesis.  The authors' own verdict:  "Its gene inventory seems insufficient for most biological processes that appear to be essential for bacterial life, and possibly the host bacteriocyte compensates."

Nasuia deltocephalinicola:  112 kb, retaining synthesis of two amino acids plus the translation machinery, and — the striking part — "both symbionts have lost genes underlying ATP synthesis through oxidative phosphorylation."  It cannot make its own ATP.

Vidania:  barely 50-52 kb, convergently evolved twice, able to make only phenylalanine, and described by its discoverers as "very close to organelles of symbiotic origin… further blurring the bacteria-organelle boundary."

This is a line to retire, not to deploy, in either direction.  These are not free-living organisms and they are not models of a first cell.  They are the terminal stage of reductive evolution inside a host that supplies everything else.

The physiology and habitat of the last universal common ancestor

Weiss, Madeline C., et al.  Nature Microbiology 1 (2016): 16116.

area: luca camp: mainstream-chemistry type: calculation access: paywalled direction: complicates

Opened in full.  The sparest LUCA:  355 protein families out of 286,514 clusters — about 0.1% — traced to LUCA by a domain-monophyly criterion.  An anaerobic, CO₂-fixing, hydrogen-dependent, nitrogen-fixing thermophile using the Wood-Ljungdahl pathway, "only half-alive and… dependent upon geochemistry," with just nine nucleotide- and five amino-acid-biosynthesis proteins.

The instability is disclosed by the authors themselves and is the reason the number cannot be quoted as a measurement.  Relax the lateral-gene-transfer criterion by one notch and the count rises to 479;  by another, to 576.  The "355-gene LUCA" is a filter setting.

It has been attacked from two directions.  Berkemer and McGlynn, MBE 37 (2020) — abstract and introduction read — added more sequence homologs and found the analysis "rejects a simple LUCA hypothesis based on phylogenetic separation of the bacterial and archaeal domains for ~82% of the previously identified LUCA proteins," with the surviving signal concentrated in translation and transcription rather than metabolism — which is exactly where Weiss's physiological portrait rests.  And Gogarten and Deamer, Nature Microbiology 1 (2016): 16229, attacked "half-alive" directly:  sequence data "support the idea that LUCA was a prokaryotic cell using nucleic acids… 20 genetically encoded amino acids, ribosomes… and membranes.  To consider a cellular organism with these properties as only 'half-alive' reflects an uncommon definition of life."

Two honesty notes on that exchange.  Only the first three paragraphs of Gogarten and Deamer were readable in the free preview.  And Weiss et al.'s published reply, Nature Microbiology 1 (2016): 16230, is visible only in its first paragraph — its substance was never read by anyone here.  This catalogue currently carries Gogarten's attack without Weiss's answer, and that is not fair reporting.  Somebody should get it.

A consensus view of the proteome of the last universal common ancestor

Crapitto, Andrew J., Amanda Campbell, Andrew J. Harris, and Aaron D. Goldman.  Ecology and Evolution 12 (2022): e8930.

area: luca camp: mainstream-chemistry type: survey access: OA direction: undercuts

CC BY, full text opened via Europe PMC including Table 1.  The single most useful entry in this section, and it is four authors auditing their own field.

Across eight published LUCA reconstructions, raw prediction counts run from 80 to 571.  The consensus set — families predicted by at least four of the eight — is 366.  The mean across studies is 768;  the median 702.

Their verdict, stated twice:  "none of them showed a high or even moderate level of similarity with any other individual study… Even the studies with fewest predictions, which we consider to be the most conservative, were not in strong agreement with each other."  And:  "Surprisingly, we found that studies of the genome or proteome of the LUCA do not uniformly agree with one another."

An article that quotes only the largest LUCA is cherry-picking.  Report the spread.

The nature of the last universal common ancestor and its impact on the early Earth system

Moody, Edmund R. R., et al.  Nature Ecology & Evolution 8 (2024): 1654-1666.

area: luca camp: mainstream-chemistry type: calculation access: OA direction: complicates

CC BY, opened in full.  The largest reconstruction on record:  LUCA at ~4.2 Ga (4.09-4.33 Ga), a genome of at least 2.5 Mb — modelled at 2.75 Mb, range 2.49-2.99 — encoding about 2,657 proteins (2,451-2,855), "comparable to modern prokaryotes," with ATP synthase subunits at posterior probability 0.82-0.99 and nineteen class-1 CRISPR-Cas effector families, meaning LUCA already had antiviral immunity and therefore predecessors and parasites.

Three things must travel with that number.

The nameable list is far shorter than the modelled total.  The gene families identified at probability 0.75 or above number 399, against a modelled 2,657 integrated over all 9,365 families' probabilities.  Goldman, Fournier and Kaçar flag that disparity explicitly.

The authors say the reconstruction is incomplete.  "It is more difficult to identify the specific gene families that might have already been present in LUCA… our reconstruction is probably incomplete."

The early date is bought by discarding the Late Heavy Bombardment.  Moody et al. explicitly reject the LHB as "a credible maximum constraint on the age of LUCA," citing Boehnke & Harrison.  So an article cannot cite Moody's short window and cite the LHB's short window at the same time.  The two cancel.

Their own framing of the window is the quotable one:  "the inferred age of LUCA… suggests that the process required a surprisingly short interval of geologic time."  And they conclude LUCA was "prokaryote grade rather than progenotic."

A New View of the Last Universal Common Ancestor

Goldman, Aaron D., and Arturo Becerra.  Journal of Molecular Evolution 92 (2024): 659-661.

area: luca camp: mainstream-chemistry type: review access: OA direction: complicates

CC BY, opened in full.  The concession that must be printed with every LUCA number.

On the spread:  "the specific predictions of individual studies often diverge, rendering them somewhat unreliable."  On the window:  the 4.2 Ga date, "if correct, suggests that early evolution, from the origin of life to the LUCA, may have occurred within only 100-200 million years."  And on what LUCA is:  this "depiction of a physiologically and ecologically complex LUCA provides strong evidence that this stage of evolutionary history was far removed from the origin of life."

LUCA is not the first cell.  A great deal of evolution stands between it and any prebiotic chemistry, and any argument that treats LUCA's complexity as the first cell's complexity is wrong.

(A minor numeric note:  Goldman and Becerra give Moody's interval as 4.09-4.32 Ga;  Moody's own abstract says 4.09-4.33.  Use Moody's.)

The progenote, LUCA, and the root of the cellular tree of life

Gogarten, J. Peter.  Chapter 8.1 in Handbook of Astrobiology (2019).

area: luca camp: mainstream-chemistry type: review access: OA direction: complicates

Author's version opened in full.  Read this before writing anything about progenotes, because the claim is the reverse of what is usually attributed to him.

Gogarten does not argue LUCA was a progenote.  He argues the opposite, and calls the conflation "a frequent misconception."  He also names a specific error in Weiss 2016 — the pheRS "LUCA branch" reflects a known archaeal-to-Spirochete horizontal transfer rather than LUCA's true position — and states the methodological floor:  "At present, reliable inference of genes to be present in LUCA is limited to genes with nearly universal distribution."

Universal paralogs provide a window into evolution before the last universal common ancestor

Goldman, Aaron D., Gregory P. Fournier, and Betül Kaçar.  Cell Genomics 6 (2026): 101140.

area: luca camp: mainstream-chemistry type: review access: OA direction: complicates

CC BY-NC-ND, opened in full.  Post-cutoff.  The field's own consensus statement, and its own statement of where its methods stop:  "The LUCA… represents a cellular organism with a complete tripartite genetic system and complex metabolism.  Evolutionary stages preceding the LUCA must have seen the development of nearly all of these core molecular systems."  And:  "the dependence of molecular phylogenetics on conserved gene sequences means that periods of evolution that predated the emergence of the genetic system cannot be studied."

The programme they propose is the interesting part — gene families duplicated before LUCA give a window into pre-LUCA evolution:  the synthetase families, IF2/EF-Tu/EF-G, ATP synthase subunits, the signal recognition particle, ABC transporters, HisA/HisF, aminotransferases.  Their results are catalogued under the genetic code above.

How secure is the date?

Bell, Elizabeth A., and Gregory P. Fournier.  Astrobiology 26 (2026): 633-643.

area: luca camp: mainstream-chemistry type: review access: abstract-only direction: undercuts

Abstract only.  Post-cutoff.  This is the entry that should stop the article from leaning too hard on the 100-200 Myr window.

Molecular clocks "provide clues to the nature and timing of… LUCA, although the latter methodology remains challenging and highly uncertain."  Fournier's own aminoacyl-tRNA-synthetase-based clock gives a wider and younger range, 3.9-4.2 Ga — but that figure comes from a December 2024 seminar abstract, not a paper, and this research could not find a published version.  Use the seminar only to establish that the date is contested.  Never cite it for a number, and do not attribute the critique to anyone by name without a citation.

Where the gap actually stands

area: minimal-cell camp: mixed type: review direction: complicates

Setting the two ends beside each other is the point of this section, and the entries above supply both.

At one end, the most complex thing demonstrated prebiotic chemistry has produced without a chemist steering the sequence:  a 45-nucleotide ribozyme copying its own strand at 0.2% yield over 72 days, in separate tubes (Gianni et al. 2026, catalogued above).

At the other, the simplest thing that has ever lived and the earliest cell that can be reconstructed:  473 genes with 149 unknown functions in a rich medium, booted from a living cell;  or a LUCA with somewhere between 355 and 2,657 proteins depending on the study, which already had a genetic code, ribosomes, a membrane and ATP synthase, and which the field's own reviewers describe as "far removed from the origin of life."

Every one of those numbers has been argued down in this catalogue.  What survives is that no chemical route between the two ends has been proposed by anyone.  That claim is deliberately smaller than the one usually made, and it is the one the sources will bear.

(One item of library housekeeping noticed in this work and not acted on:  index.tsv carries a duplicate moody-2024 row differing only in author-name diacritics and identifier formatting, and a duplicate white-2025 row.  Also, Lecompte et al., Nucleic Acids Research 30 (2002), the source of the "34 universal ribosomal-protein families" figure, is abstract only and is a 2002 count from 66 genomes;  it should be replaced with a modern source.)

Sequence space and functional information

This is the section where the argument is usually won or lost, and where both sides are usually careless.  The question has a precise form:  out of all possible sequences of a given length, what fraction do something useful?  The published estimates span fifty-three orders of magnitude, and every one of them is inside the peer-reviewed literature, between people who are not arguing about God.  Arthur Hunt's one-line summary — "10^-10 → 10^-63 (or thereabout): this is the range of estimates of the density of functional sequences in sequence space that can be found in the scientific literature" — is the single most important fact on this page.

Two methods exist and they disagree by construction.  The forward method builds a library of genuinely random sequences and counts the hits.  It measures the fraction directly, but it cannot see below about one in 10¹³, because that is as large as a library gets.  The reverse method mutates a working protein, measures how much abuse it tolerates per position, and raises that tolerance to the power of the chain length.  It can produce any exponent you like, and every one of those exponents assumes positions fail independently.  Every entry below states which method it used.

Functionally acceptable substitutions in two alpha-helical regions of lambda repressor

Reidhaar-Olson, J. F., and R. T. Sauer.  Proteins 7 (1990): 306-316.

area: sequence-space camp: mainstream-chemistry type: experiment access: abstract-only direction: complicates

The oldest reverse-method benchmark, and the source of the ~10^-63 figure for a 92-residue protein.  Abstract only — Wiley bot-blocks and no open copy exists.

Two cautions.  The authors' own summary sentence runs the opposite way from the use made of them:  "The results reveal the high level of degeneracy in the information that specifies a particular protein fold."  And the 10^-63 figure is attributed to them by Axe;  this research could not confirm from their own paper that they state it.

Searching sequence space for protein catalysts

Taylor, S. V., K. U. Walter, P. Kast, and D. Hilvert.  PNAS 98 (2001): 10596-10601.

area: sequence-space camp: mainstream-chemistry type: experiment access: OA direction: complicates

Opened in full.  The other reverse-method benchmark:  chorismate mutase, 93 residues, ~10^-24 — which implies a per-residue tolerance of 0.55 against Reidhaar-Olson & Sauer's 0.21, on proteins of essentially the same length.

Their 10²⁴ is not a measured frequency.  It is a required library size, built on a stated worst-case assumption:  "the assumption that only a single amino acid is tolerated at each highly conserved position in the active site."  Their own conclusion is a research programme, not a barrier.

Functional proteins from a random-sequence library

Keefe, A. D., and J. W. Szostak.  Nature 410 (2001): 715-718.

area: sequence-space camp: mainstream-chemistry type: experiment access: OA direction: supports

Opened in full and read line by line from the free author manuscript at PMC4476321.  The forward count, and the strongest single result against the rarity argument.  Six trillion random proteins, each carrying eighty contiguous random amino acids, passed over a column that grabs anything binding ATP.  Four unrelated protein families came out.  Corrected for folding efficiency the frequency is roughly 1 in 10¹¹.

Szostak states the design point himself:  "This library randomly samples the whole of sequence space, rather than the vicinity of a known protein."  And he states the other half in the same conclusion:  "this frequency is still low enough to emphasize the magnitude of the problem faced by those attempting de novo protein design."

Five caveats are in the paper and all five are real.  The function is binding, not catalysis.  The winners were poorly folded, 5-15% active.  The best affinity came from eighteen rounds of directed evolution rather than the initial draw.  The winning family needs environmental zinc it did not pay for.  And the method has a detection floor around one in 10¹³ — if the true frequency for a harder function were one in 10³⁰, this experiment would return nothing at all.

The follow-up hurts the isolated-islands picture more than the original does:  the selected protein's structure was solved and turned out to be a novel fold unknown in biology, stabilised by a zinc ion — Lo Surdo, P., M. A. Walsh, and M. Sollazzo, "A novel ADP- and zinc-binding fold from function-directed in vitro evolution," Nature Structural & Molecular Biology 11 (2004): 382-383, abstract only from PubMed:  "the first protein structure derived from in vitro evolution and the first nucleotide-binding protein stabilized by a zinc ion."

Estimating the prevalence of protein sequences adopting functional enzyme folds

Axe, Douglas D.  Journal of Molecular Biology 341, no. 5 (2004): 1295-1315.  PMID 15321723.

area: sequence-space camp: id-movement type: experiment access: closed direction: complicates

The most-cited number in origin-of-life apologetics, and the entry that needs the longest access note on this page.

Access, stated exactly.  The paper is closed.  Unpaywall reports "oa_status": "closed", "has_repository_copy": false, "oa_locations": [];  OpenAlex agrees;  Semantic Scholar reports the publisher has elided even the abstract from its record;  ScienceDirect, ResearchGate and CORE all bot-block.  For most of this project's history the body had never been opened by anyone here, and the account of the method rested second-hand on Arthur Hunt's 2007 exposition.  That is no longer the state of play:  the full typeset PDF was eventually recovered from a Wayback Machine capture of a now-dead personal site — 21 pages, pdfinfo Creator Elsevier, CreationDate 2004-08-06 — and read directly, and every page-numbered quotation below comes from that file.  The recovered file is a publisher-typeset copy from an unauthorised third-party host:  excerpt it, do not redistribute it.  The main article's own sidebar still says the body could not be opened;  that sidebar is out of date and should be corrected.

What it actually measured.  The target was the large domain of TEM-1 β-lactamase, 153 residues, scored in vivo — E. coli carrying the variant either grew on ampicillin or did not.  Axe first engineered a temperature-sensitive reference variant differing from wild type at 22% of the 153 positions and working only at 25 °C.  Four spatially separated clusters of ten side chains were then randomised within the fold's hydropathic signature.  Two clusters recovered functional variants at ~3 × 10^-4;  one at ~10^-5;  one cluster gave no functional variants at all.  Averaging gives a per-residue tolerance of 0.38, and 0.38¹⁵³ ≈ 1 in 10⁶⁴.

What the headline figure is.  Not that.  The 10⁷⁷ is 10^-64 multiplied by a foldability factor of 10^-10 taken from Lau and Dill's theoretical lattice model at its least stringent criterion, and by a factor of 10^-3 that Axe introduces with the words "A reasonable estimate of the average proportion suited to a particular task is 0.1%" (p. 1310).  One measured number, one computed number, one reasonable-sounding guess.

The fact that decides how the figure may be used.  Axe published a range, and 10⁷⁷ is its pessimistic end.  Verbatim, p. 1310:  "the proportion of all sequences of large-domain length that perform the specified function by means of any tertiary fold … is estimated to be in the range of one in 10⁷⁷ to one in 10⁵³."  Twenty-four orders of magnitude, in the author's own sentence.  Axe reports the range again in his 2010 BIO-Complexity paper.  His popularisers drop it, and the article should say plainly that the problem is downstream of Axe rather than in him.

What the paper concedes about evolution.  Almost nobody quotes this.  Discussion, pp. 1310-1311:  about one in four random single-residue changes are functionally neutral, and the proportion under stricter conditions would be lower "but not so low as to preclude progressive sequence divergence by gradual accumulation of point mutations."  Axe's argument is about the origin of new folds, not about protein evolution generally, and any use of his number against protein evolution as such is contradicted by the paper it cites.

Verdict.  It is real science, peer reviewed at a serious journal, with its methods and hedges stated in the text;  attacks on it as pseudoscience are wrong.  It may be cited for the claim that the density of functional sequences around a given enzyme fold is low, on the order of 10^-40 to 10^-64 for a domain-sized protein.  It may not be cited for "the probability of a functional protein is 1 in 10⁷⁷," and the 10⁷⁷ may never appear without the 10⁵³.

Meyer's Hopeless Monster

Gishlick, Alan, Nick Matzke, and Wesley R. Elsberry.  The Panda's Thumb, 24 August 2004.

area: sequence-space camp: atheist-explicit type: web-page access: OA direction: supports

Opened.  The first published criticism, three weeks after Axe's paper, aimed not at Axe but at Stephen Meyer's use of him:  "Meyer only reports the lowest extreme.  One in 10⁵³ is still a small number, but Meyer apparently didn't feel comfortable mentioning those 24 orders of magnitude to his reader."  Checked against Axe's own text and correct.  Highest verdict-to-effort ratio of any objection in this literature — it requires no biochemistry, only reading the Discussion.

A standing attribution error:  this piece is frequently credited to Ken Miller.  It is not his.

Protein stability promotes evolvability

Bloom, Jesse D., Sarah T. Labthavikul, Christopher R. Otey, and Frances H. Arnold.  PNAS 103 (2006): 5869-5874.

area: sequence-space camp: mainstream-chemistry type: experiment access: OA direction: supports

Opened in full.  The measurement that gives the objection to Axe's construct quantitative teeth, and the better citation precisely because it was not written against him and does not mention him.

Same enzyme, same mutagenesis, same assay, two variants of cytochrome P450 BM3 differing only in stability:  "61% of 5H6 mutants contained at least half the folded protein of the parent versus 33% for 21B3."  Per-mutant tolerance nearly doubles with starting stability.  Since moving a per-residue tolerance from 0.38 to 0.60 moves a 153-residue estimate by thirty orders of magnitude, a destabilised parent is not merely a lower bar — it is a steeper hill.

Axe (2004) and the evolution of enzyme function

Hunt, Arthur.  The Panda's Thumb, 14 January 2007.

area: sequence-space camp: atheist-explicit type: web-page access: OA direction: supports

Opened in full.  Four objections:  the construct was deliberately destabilised and therefore "exquisitely sensitive to mutation";  neighbouring activities that share the structural theme are not counted;  wholly unrelated solutions to the same activity (metallo-β-lactamases) are not counted;  and the forward method typically returns 10^-10 to 10^-15 for the same parameter.

Credit in both directions.  Hunt thanks Axe for helping with early drafts, and Axe confirms in his own reply that "His summary of the experiments I reported in the JMB paper was largely correct."  The description of the experiment is not in dispute;  only the inference is.

Functional information and the emergence of biocomplexity

Hazen, Robert M., Patrick L. Griffin, James M. Carothers, and Jack W. Szostak.  PNAS 104, suppl. 1 (2007): 8574-8581.

area: information camp: mainstream-chemistry type: calculation access: OA direction: complicates

Opened in full and read line by line.  The most valuable single entry in this section, and the article should adopt its definition explicitly.

I(E_x) = −log₂[F(E_x)]:  count what fraction of all possible configurations do the job at least this well, then take the negative log base two.  Three properties come with it, and each one disqualifies a common argument.  Functional information is defined only relative to a stated function — which kills the whole genre of claims about "the information in DNA" that never say information for what.  It is defined relative to a stated threshold, because function is not binary.  And it "requires a deep understanding of the system's agents and their various interactions" — you must know the distribution over all configurations, not just the sequences that happen to exist today.

Measured values in the paper:  a random Avida instruction sequence performs at least one logic or arithmetic operation with probability ~10^-3;  a tenfold increase in GTP-binding strength costs about 10 additional bits, i.e. a thousandfold decrease in abundance among random sequences.  And eleven distinct classes of GTP-binding RNA were identified, distinguished by both sequence motif and secondary structure — the same multiple-islands result as Szostak's four protein families, and the thing the reverse method structurally cannot see.  The authors caution that even the islands picture is "a visual simplification."

This is the concept intelligent-design writing has been reaching for, defined properly, by mainstream researchers, in PNAS, with a Nobel laureate as senior author.  It concedes that information is a legitimate measurable property of biological systems.  It also fixes rules that most apologetics uses of the word fail.

Measuring the functional sequence complexity of proteins

Durston, Kirk K., David K. Y. Chiu, David L. Abel, and Jack T. Trevors.  Theoretical Biology and Medical Modelling 4 (2007): 47.

area: information camp: id-movement type: calculation access: OA direction: undercuts

Opened in full.  The "Fits" measure:  take a Pfam alignment, compute the Shannon uncertainty at each aligned position, subtract from log₂20 = 4.32 bits, sum over positions.  Thirty-five families, values from 0 to about 2400.  The apologetics payload is one sentence:  a Fit value of 379 for ribosomal S12 is converted into "an evolutionary search target of approximately 10^-106 percent of 121-residue sequence space."

The measure is computed entirely from sequences that already exist, which measures how much one lineage has varied — not how many sequences anywhere in the space could do the job.  Matlock and Swamidass demonstrated the failure directly (below).

Two things in its favour, stated fairly:  it is genuinely peer reviewed and open access, not a house journal;  and applied to ubiquitin, the seven highest-Fit sites cluster on the binding domain, six of seven near the Lys-48 site.  As a conservation-detection heuristic it works.  That is a real, modest, checkable result, and the article should say so.

Reception, measured.  An OpenAlex query on the works citing this paper returns twenty-eight citations in nineteen years.  The largest block is by co-author David Abel (six);  then Dembski, Marks and Ewert (three);  then Thorvaldsen and Hössjer (three), Royal Truman in Journal of Creation, and Durston's own follow-up.  The clearly independent technical engagements are two.  The paper cleared peer review and was then taken up almost entirely by the movement that wanted its conclusion — against Hazen et al., published six months earlier, defining essentially the same quantity, which became a research programme.

The cosmological model of eternal inflation and the transition from chance to biological evolution

Koonin, Eugene V.  Biology Direct 2 (2007): 15.

area: information camp: atheist-explicit type: calculation access: OA direction: complicates

Opened in full, including the Appendix and all four signed referee reviews with Koonin's responses.  The source of 10^-1018, the largest number in origin-of-life apologetics.

What the number is.  A self-described toy calculation:  10²² stars in the observable region, one habitable planet per ten stars, a 10 km habitable layer, 1% of that volume doing RNA synthesis, one n-mer per cm³ per second, over 10¹⁰ years — about 1.5 × 10⁶¹ distinct oligomers ever tried.  For a coupled replication-translation system requiring n = 1800 nucleotides, "E < 10^-1018."  The calculation assumes exactly one required sequence out of 4ⁿ, which is Hoyle's assumption and is measurably false.  Koonin's own caveat, same Appendix:  "The model considered here is not supposed to be realistic by any account."

The paragraph directly above the famous one.  For the RNA-world version, n ≈ 100:  "A ribozyme replicase consisting of ~100 nucleotides is conceivable, so, in principle, spontaneous origin of such an entity in a finite universe consisting of a single O-region cannot be ruled out in this toy model."  Anyone quoting Koonin's big number against abiogenesis is quoting the paragraph immediately below one that says the opposite.

The mandatory disclosure.  Koonin addresses intelligent design twice, unprompted.  In the main text:  the model "leaves no room whatsoever for any form of intelligent design."  And in his published reply to referee Eric Bapteste, who had warned the paper "might open too broad an avenue to the supporters of intelligent design":  "Properly interpreted, the anthropic principle is a death knell to ID."  He then anticipates being quote-mined:  "Will they, actually?  No doubt they will!"  The paper is CC BY, so it may be quoted at length, and if the article quotes Koonin at all it must print the death-knell sentence.

How much of protein sequence space has been explored by life on Earth?

Dryden, David T. F., Andrew R. Thomson, and John H. White.  Journal of the Royal Society Interface 5 (2008): 953-956.

area: sequence-space camp: mainstream-chemistry type: calculation access: OA direction: complicates

Opened in full.  The number every probability claim on this subject has to beat, and it is missing from almost every treatment on both sides.  A probability is meaningless without the number of trials.

Their extreme upper bound:  "Let us assume that every single gene in this total of 10³⁴ is unique and that evolution has been working on these genes for 4 Gyr completely changing each gene to some other unique, new gene every single year.  This gives an extreme upper limit of 4×10⁴³ different amino acid sequences explored since the origin of life."  Their lower bound in the same paragraph is 4 × 10²¹.

That number is where the Axe-versus-Szostak dispute stops being academic.  At F ≈ 10^-11, a search resource of 10⁴³ is overwhelming.  At F ≈ 10^-64, it is twenty to thirty-four orders of magnitude short, and deep time does not help because deep time is already inside the 10⁴³.

Two honest caveats.  The headline conclusion — that "it is instead quite plausible for all of functional protein sequence space to have been explored" — depends on shrinking the effective space by allowing reduced amino-acid alphabets and short chains:  "all sequences containing 20, 8 and 3 types of amino acids have been explored if the chains are 33, 50 and 100 amino acids in length."  Whether a 100-residue protein built from a three-letter alphabet models a real enzyme is exactly what is in dispute.  And it is a four-page Report with deliberately generous assumptions.  Quote it as an upper bound.

The paper also attacks the premise by name:  "Protein sequence space is often viewed as a limitless desert of maladjusted sequences with only a few oases of working sequences linked by narrow pathways (Axe 2000, 2004)."

Protein dynamism and evolvability

Tokuriki, Nobuhiko, and Dan S. Tawfik.  Science 324 (2009): 203-207.

area: sequence-space camp: mainstream-chemistry type: review access: abstract-only direction: supports

Abstract only, paywalled.  Enzymes are promiscuous — most catalyse side reactions weakly — so the functional distance between activities is much smaller than a sequence-space picture suggests:  these properties "facilitate the divergence of new functions within existing folds and the evolution of entirely new folds."

Sequence space and the ongoing expansion of the protein universe

Povolotskaya, Inna S., and Fyodor A. Kondrashov.  Nature 465 (2010): 922-926.

area: sequence-space camp: mainstream-chemistry type: measurement access: abstract-only direction: complicates

Abstract only, verbatim from PubMed.  Concedes the sparseness and denies the conclusion in one sentence:  "approximately 98 per cent of sites cannot accept an amino-acid substitution at any given moment but a vast majority of all sites may eventually be permitted to evolve when other, compensatory, changes occur."  The best single citation for "the landscape is rugged and navigable."

The case against a Darwinian origin of protein folds · Correcting Four Misconceptions about my 2004 Article in JMB

Axe, Douglas D.  BIO-Complexity 2010(1): 1-12;  and Biologic Institute, 4 May 2011, reposted by Discovery Institute 3 March 2016.

area: sequence-space camp: id-movement type: philosophy-paper access: OA direction: complicates

Both opened.  Axe's own replies, and they deserve to be presented rather than waved at.

His answer on the destabilised construct is a real argument about internal consistency:  if you accept a barely-working mutant as functional, you must start from a barely-working parent, or you are measuring the wrong thing.  In his words, the method "was deliberately generous toward Darwinism."  Why it does not fully answer the objection:  Axe's reply is about the threshold;  Hunt's objection is about the slope, and Bloom et al. 2006 measured exactly the quantity Axe's reply assumes away.  The objection survives weakened;  so does the reply.

He is also candid about what the reverse method is for:  the ideal forward experiment would need a library "about a hundred billion times more massive than the sun."  Fine — but then the answer is an extrapolation, with an extrapolation's error bars.

What he has never answered:  the four misconceptions do not include the range objection.  Axe has never publicly addressed the charge that 10⁷⁷ is quoted without its companion 10⁵³ — though he himself gives the range in the 2010 paper.  BIO-Complexity reserved a DOI for a critique of that paper;  it resolves to HTTP 404, and no critique was ever published there.

A trap.  Two recent papers defending protein rarity — a 2024 PLoS ONE percolation analysis and a 2025 rarity methodology — are by Brian J. Miller of the Biologic Institute, Axe's own institute, not by Kenneth R. Miller of Brown University.  On the evidence available, Kenneth Miller has no published criticism of Axe 2004:  the Kitzmiller transcripts contain zero occurrences of "Axe" in his testimony, his own resource site never mentions Axe, and "Meyer's Hopeless Monster" is not his.

Evolution and Functional Information

Matlock, Matthew K., and S. Joshua Swamidass.  bioRxiv preprint 10.1101/114132, 5 March 2017.

area: information camp: christian type: calculation access: preprint direction: undercuts

Opened in full.  The decisive demonstration against Durston's method, written by a Christian computational biologist, and a preprint with no journal version found — label it as such.

They simulate sequences a fixed number of mutational events from one ancestor with no functional constraints at all, so the true functional information is by definition zero.  The alignment-based measure does not converge to zero:  it converges to 4.3, 2.6, 1.8, 1.2 and 0.8 bits per amino acid at mutation rates of 0.0 through 1.0.  Durston et al. report real protein families at "about 1 to 4 bits per amino acid."  A simulated family under zero functional constraint lands in the same range.

Their conclusion:  "we must conclude either that (1) a mindless process driven entirely by randomness can reliably generate arbitrarily high FI or that (2) MISA is an extremely bad estimate of FI.  Either conclusion falsifies the FI argument entirely."  Three further biases all push the same way — codon structure adds 0.183 bits per residue by itself, multiple families with the same function are not aggregated, and selection operates at the boundary of high and low function rather than function and non-function.

On the roles of function and selection in evolving systems

Wong, Michael L., et al. (incl. Robert M. Hazen).  PNAS 120, no. 43 (2023): e2310223120.

area: information camp: mainstream-chemistry type: philosophy-paper access: abstract-only direction: complicates

Abstract only, verbatim from PubMed;  body not opened and the responses to it not surveyed.  Proposes a "law of increasing functional information."  Do not cite it as evidence of anything.  Cite it only to show that functional information is now a live mainstream research programme rather than a creationist coinage.  It drew immediate criticism as unfalsifiable and is philosophy of science more than measurement.

Energy Flow in Biology

Morowitz, Harold J.  Academic Press, 1968.

area: sequence-space camp: atheist-explicit type: book access: book direction: undercuts

Read page by page through Internet Archive controlled digital lending (item energyflowinbiol0000moro;  loan taken and returned).  The clearest and most fully documented quote-mine in the whole abiogenesis debate, and it has been almost entirely unworked.

Two numbers, two pages, two conduits.  The famous 1 in 10^339,999,866 comes from Table 4-14 on p. 99, and Morowitz never prints that figure anywhere.  James F. Coppedge did the subtraction, in Evolution: Possible or Impossible? (Zondervan, 1973), p. 235 — the scanned chapter was opened in full.  Morowitz's table gives 10^-3.4×108¹³⁴oppedge divided by Morowitz's own 10¹³⁴ "chances in the history of the universe," and 340,000,000 − 134 = 339,999,866.  The derivation checks exactly.  A second figure, 1 in 10^{100,000,000,000}, comes from pp. 7, 11-12 and 67 by way of Robert Shapiro's Origins (1986), p. 128.

What it actually calculates.  Not the origin of life.  A deliberately artificial equilibrium thought experiment, stated on pp. 5-6:  grow a large batch of E. coli, centrifuge it into a pellet, "raise the temperature to some very high value (on the order of 10,000 °C) so as to destroy any traces of the original chemical state," cool it, and ask what fraction of the resulting equilibrium ensemble is alive.  No researcher has ever proposed that as a model of the origin of life.  Morowitz built it to show that equilibrium is the wrong frame.

He says so on the same page, and every ellipsis in the apologetics version cuts it out.  Page 12, immediately after the calculation:  "It is always possible to argue that any unique event would have occurred.  This is outside the range of probabilistic considerations, and really, outside of science.  We shall subsequently show that it is not necessary to invoke such arguments to explain biological organization."  Page 68:  "if equilibrium processes alone were at work, the largest possible fluctuation in the history of the universe is likely to have been no larger than a small peptide… On the other hand, an energy enrichment of 0.07 eV/atom, the average amount required for biomass, is easily achieved in an energy flux situation."  Note the conditional, and note that his sentence is about "the impossibility of considering life originating as a fluctuation" — not the impossibility of life originating.

His own premise, from the preface, p. vii:  "I have rejected the prebiological notion of 'historical accident' as a kind of know-nothingness which renders further study impossible."

And he addressed the misuse under oath.  Morowitz testified for the plaintiffs in McLean v. Arkansas in December 1981.  The transcript was opened in full via a Wayback capture of antievolution.org, the live host having failed to respond.  Verbatim, pp. 500-501:

"A:  Well, I find the use of probabilistic arguments to be somewhat deceptive. …  To do the probabilistic calculations, we would have to know all the kinetic and mechanistic details by which the processes have come about…  We are simply lacking the information to do the calculations now, so to present them on the basis of the random model is somewhat deceptive.

Q:  Is it also in your view unscientific?

A:  Since deception is unscientific, the answer to that is yes."

The man whose number is being used said in federal court, in 1981, that using probability arguments that way is deceptive — and he was describing the use being made of his own work.

One caution on the arithmetic:  the numeric columns of Table 4-14 are damaged in the scan's OCR, and the table value should be re-checked against a print copy before anything further is built on it.

The Universe: Past and Present Reflections

Hoyle, Fred.  Engineering and Science 45, no. 2 (November 1981): 8-12.

area: sequence-space camp: unaffiliated type: popular access: OA direction: complicates

Opened in full, on disk.  Use this and never the Nature 294:105 citation, which is a news story about Albania (see the failed-verification section above).  Hoyle published the 747 argument himself, in plain prose:  "a typical enzyme has a chain of perhaps 200 links and … there are 20 possibilities for each link … This is for one enzyme, and there are upwards of 2000 of them."

All three fatal flaws are visible in Hoyle's own paragraph.  He counts the modern enzyme set all at once, which is a scenario no researcher proposes — and which is closer to what young-earth creationism believes than to anything in the origin-of-life literature.  He counts one acceptable sequence per enzyme, which is measurably false.  And he dismisses the gradual alternative as "deceitful" while conceding in the same breath that small enzymes are "sufficiently elementary to be discoverable by chance."

Evolution and the second law of thermodynamics

Bunn, Emory F.  American Journal of Physics 77 (2009): 922-925.  arXiv:0903.4603.

area: philosophy camp: mainstream-chemistry type: calculation access: preprint direction: supports

Opened in full.  The quantitative version of the open-system reply, and better than the usual hand-wave.  Bunn computes an upper bound on the entropy reduction represented by all life on Earth, rigging every assumption to overstate it — every atom plucked from a Dead-Earth atmosphere into its exact present quantum state, biomass assigned zero entropy — and gets |ΔS| ~ 10⁴⁴ k.  Sunlight pays that bill in under a year.  Life took four billion years.  The margin is about eleven orders of magnitude.

Bunn's paper corrects Daniel Styer's earlier treatment (Am. J. Phys. 76 (2008): 1031-1033), which rested on "an unjustified and probably incorrect assumption."  Styer was not obtained — the AIP page bot-blocks and no abstract was reached;  it is cited only through Bunn's description.  Two physicists arguing about how to do the sum, both concluding there is no conflict.

Timing, and the earliest biosignatures

The apologetics argument here is that life appears in the rock record almost as soon as the earth cools, leaving no time for chemistry.  The mainstream geology of the last decade has moved against that argument, and the movement is not creationist in origin.  The sterilising impact spike that made the window look impossibly narrow is probably an artefact of argon loss in the Apollo samples, and the oldest biosignature claims have mostly been withdrawn or contested down to about 3.48 Ga.

Evidence for life on Earth before 3,800 million years ago

Mojzsis, S. J., et al.  Nature 384 (1996): 55-59.

area: biosignatures camp: mainstream-geology type: measurement access: abstract-only direction: complicates

Abstract only, from Europe PMC.  The founding Isua/Akilia claim:  isotopically light carbon in apatite inclusions in ~3,800 Myr banded iron formation, and possibly in a >3,850 Myr Akilia rock;  "no known abiotic process can explain the data."  Largely abandoned after van Zuilen 2002.

Isotopically light carbon in the early Archaean of west Greenland

Rosing, Minik T.  Science 283 (1999): 674-676.

area: biosignatures camp: mainstream-geology type: measurement access: abstract-only direction: complicates

Abstract only;  science.org bot-blocks.  The stronger Isua case:  2-5 µm graphite globules in ordinary marine turbidites at δ¹³C ≈ −19‰, "likely biogenic."  Light carbon alone is not proof — siderite disproportionation and Fischer-Tropsch synthesis also make it.

Reassessing the evidence for the earliest traces of life

van Zuilen, M. A., A. Lepland, and G. Arrhenius.  Nature 418 (2002): 627-630.

area: biosignatures camp: mainstream-geology type: measurement access: abstract-only direction: supports

Abstract only.  The rebuttal:  the Isua graphite is metasomatic, injected by hot fluid after deposition, and "clearly lack[s] biological relevance."

Stromatolite reef from the Early Archaean era of Australia

Allwood, Abigail C., et al.  Nature 441 (2006): 714-718.

area: biosignatures camp: mainstream-geology type: measurement access: abstract-only direction: supports

Abstract only.  Seven stromatolite morphotypes across a multi-kilometre peritidal carbonate platform in the Strelley Pool Chert at ~3,430 Ma, which "refute contemporary abiogenic hypotheses."  This, not Isua, is the field's accepted floor.

Evidence for early life in Earth's oldest hydrothermal vent precipitates

Dodd, Matthew S., et al.  Nature 543 (2017): 60-64.

area: biosignatures camp: mainstream-geology type: measurement access: paywalled direction: complicates

Author accepted manuscript on disk, read in full including Supplementary.  Hematite tubes and filaments in Nuvvuagittuq ferruginous seafloor-vent precipitates "at least 3770 and possibly 4290 million years old."

Do not lead with Nuvvuagittuq.  The 520 Myr age spread is between two dating systems on the same belt — ¹⁴⁶Sm-¹⁴²Nd on metabasalts gives 4290 Ma, U-Pb on zircons (Cates et al. 2013) gives 3750-3780 Ma.  Worse, Lan et al. 2022 report ~2.5 Ga U-Pb ages on apatite and zircon in the fossil-bearing jasper itself, which would make the host rock Neoarchean.  Lan et al. was never opened and no abstract was obtainable — it is known only through one clause in Westall 2025.  The discoverers' own 2022 follow-up concedes the filaments are of "uncertain origin," and a 2026 book chapter by Papineau and O'Neil, also not opened, appears to be a critical re-evaluation by the original claimants.

Rapid emergence of life shown by discovery of 3,700-million-year-old microbial structures

Nutman, Allen P., et al.  Nature 537 (2016): 535-538.

area: biosignatures camp: mainstream-geology type: measurement access: abstract-only direction: complicates

Abstract only, cross-checked against the rendered page.  1-4 cm stromatolites in 3,700 Myr Isua metacarbonates with a seawater-like rare-earth signature.  The title is the argument, and it is the single most-quoted source for "life appeared immediately."

Reassessing evidence of life in 3,700-million-year-old rocks of Greenland

Allwood, Abigail C., et al.  Nature 563 (2018): 241-244.

area: biosignatures camp: mainstream-geology type: measurement access: abstract-only direction: supports

Abstract only, verified twice independently — from Europe PMC and off the rendered page, and the two match.  The body was not opened, and it contains the actual three-dimensional evidence.

The rebuttal in the same journal:  three-dimensional analysis shows a "non-biological, post-depositional origin" in tectonic deformation, and the authors frame it as a cautionary tale for Mars biosignature searches.  Note who wrote it.  A co-author is Minik Rosing, whose own 1999 carbon work is the strongest surviving Isua biosignature claim, and Abigail Allwood is the author of the Strelley Pool paper above.  These are not people hostile to ancient life;  they are hostile to badly evidenced ancient life.

Reappraisal of purported ca. 3.7 Ga stromatolites from the Isua Supracrustal Belt

Zawaski, M. J., et al.  Earth and Planetary Science Letters 545 (2020): 116409.

area: biosignatures camp: mainstream-geology type: measurement access: preprint direction: supports

Accepted manuscript opened in full from Oxford ORA.  The structures "lack convex upward internal laminae" and "fail to meet the criteria to be referred to as stromatolites."  Their own hedge is worth quoting:  they do not rule out life playing "some role," they shift the burden of proof.

The article should note what it is missing here.  Nutman et al. 2021 in EPSL is the direct reply to this paper and was never opened, with no abstract obtainable anywhere;  neither was Van Kranendonk et al.'s 2025 Earth-Science Reviews defence.  Section-by-section, the sceptics get the last word in this catalogue by default, and that may not be fair.

Earth's earliest and deepest purported fossils may be iron-mineralized chemical gardens

McMahon, Sean.  Proceedings of the Royal Society B 286 (2019): 20192410.

area: biosignatures camp: mainstream-geology type: experiment access: OA direction: supports

Abstract only;  the paper is open at PMC.  The most damaging entry in this group, because it is not scepticism but a demonstration:  abiotic "chemical gardening" reproduces the hollow, branching, lifelike filament morphology and the iron mineralogy of the Nuvvuagittuq and Dresser structures in a beaker.  Not "I doubt it" but "here is the process, run on a bench."

SIMS analyses of the oldest known assemblage of microfossils document their taxon-correlated carbon isotope compositions

Schopf, J. William, et al.  PNAS 115 (2018): 53-58.  ·  and  ·  Brasier, Martin D., et al.  Nature 416 (2002): 76-81.

area: biosignatures camp: mainstream-geology type: measurement access: abstract-only direction: complicates

Both abstract only.  The Apex chert dispute, thirty years old and unresolved.  Schopf's 1993 "eleven species at ~3,465 Ma" was the textbook oldest fossil;  Brasier argued the structures are secondary artefacts of Fischer-Tropsch hydrothermal carbon with no primary biological morphology;  Schopf's 2018 SIMS work reports taxon-correlated δ¹³C from −31‰ to −39‰ and claims to "confirm the biogenicity."  No clean resolution exists.  Report it as contested.

Nano-porous pyrite and organic matter in 3.5-billion-year-old stromatolites record primordial life

Baumgartner, Raphael J., et al.  Geology 47 (2019): 1039-1043.

area: biosignatures camp: mainstream-geology type: measurement access: abstract-only direction: supports

Abstract only;  GSA bot-blocks.  Probably the strongest single Archean biosignature paper:  diamond-drilled below the weathering zone, nano-porous pyrite with entombed organic filaments, δ¹³C_OM = −29.6 ± 0.3‰ VPDB.  With Djokic et al. 2017 on the ~3.48 Ga Dresser hot-spring deposits, this is what the 3.48 Ga floor actually rests on.

What the earliest evidence for life tells us about the early evolution of the biosphere

Westall, Frances.  Philosophical Transactions of the Royal Society B 380 (2025): 20240106.

area: biosignatures camp: mainstream-geology type: review access: OA direction: complicates

CC BY;  opened in full.  The most recent even-handed review, and the right source for the current floor:  "the abundance of biosignatures as of the Palaeoarchaean, and as early as 3.48 Ga, suggests a diverse community including chemotrophs and phototrophs."  It also flags a contamination hazard nobody else discusses — modern phototrophic endoliths living in the Isua outcrops "could confuse measurements."

Illusory Late Heavy Bombardments

Boehnke, Patrick, and T. Mark Harrison.  PNAS 113 (2016): 10802-10806.

area: timing camp: mainstream-geology type: calculation access: OA direction: supports

Opened in full via Europe PMC.  Partial ⁴⁰Ar/³⁹Ar resetting under a monotonically declining impact flux manufactures a spurious 3.9 Ga spike in the Apollo data, so "impacts may have played a minimal role in terrestrial habitability."  Their own caveat:  this "does not preclude the existence of such events."

Cataclysm No More: New Views on the Timing and Delivery of Lunar Impactors

Zellner, Nicolle E. B.  Origins of Life and Evolution of Biospheres 47 (2017): 261-280.

area: timing camp: mainstream-geology type: review access: OA direction: supports

Open access, read in full.  The single most damaging sentence in this catalogue for the timing argument, and it is not a creationist writing:  the evidence supports "a prolonged lunar (and thus terrestrial) bombardment from ~4.2 to 3.4 Ga and not a cataclysmic spike at ~3.9 Ga," and therefore "the timeline for the assembly of the first biomolecules has been lengthened."

The Late Heavy Bombardment

Bottke, William F., and Marc D. Norman.  Annual Review of Earth and Planetary Sciences 45 (2017): 619-647.

area: timing camp: mainstream-geology type: review access: paywalled direction: complicates

PDF on disk, read;  the DOI bot-blocks.  The counterweight, and the reason the LHB should not be reported as dead.  Convergent Rb-Sr, Sm-Nd, U-Pb and ⁴⁰Ar-³⁹Ar clustering still supports a discrete impact episode:  "excellent agreement… provides confidence that the ages are meaningful."

Lunar chronology model with the Chang'e-6 farside samples

Yue, Z., et al.  Science Advances 12 (2026): eady9265.  ·  and  ·  Zhang, W. F., et al.  Science Advances 12, no. 30 (2026): eaee8718.

area: timing camp: mainstream-geology type: measurement access: OA direction: supports

Post-cutoff, and the most consequential 2026 development in this area.  The first is CC BY and was opened in full:  2807 ± 3 Ma basalts and 4247 ± 5 Ma norites from the lunar farside, with an impact rate that "supports a smooth decay instead of abrupt changes."  The second was read as an abstract only and is not yet in PMC:  28 single-clast ⁴⁰Ar/³⁹Ar ages spanning 4.33-1.13 Ga, of which only two of twenty-eight fall in the 4.0-3.7 Ga window a cataclysm should dominate.

The Apollo samples all came from the lunar nearside, which is why a farside sample return was the decisive test.  Nobody appears to have written the reconciliation between these results and Bottke & Norman's hybrid model — a "discrete early, post-accretion" population is not the same thing as a 3.9 Ga cataclysm.  The honest statement is that the 3.9 Ga spike is dead, not that the LHB is.

Constraining the Time Interval for the Origin of Life on Earth

Pearce, Ben K. D., Andrew S. Tupper, Ralph E. Pudritz, and Paul G. Higgs.  Astrobiology 18 (2018): 343-364.  arXiv:1808.09460.

area: timing camp: mainstream-geology type: review access: preprint direction: supports

arXiv version opened in full.  These are the numbers the article should use.  A habitability boundary at 4.5 Ga (early) or 3.9 Ga (late), a biosignature boundary at ~3.7 Ga, and the conclusion:  "the time taken for life to appear could… be within 200 Myr or as long as 800 Myr."  Not ten million years.

One caveat the paper cannot carry:  its biosignature boundary partly rests on the Nutman 2016 stromatolites, and the arXiv version predates Allwood 2018.

Does the rapid appearance of life on Earth suggest that life is common in the universe? · Bayesian analysis of the astrobiological implications of life's early emergence

Lineweaver, Charles H., and Tamara M. Davis.  Astrobiology 2 (2002): 293-304.  ·  and  ·  Spiegel, David S., and Edwin L. Turner.  PNAS 109 (2012): 395-400.

area: timing camp: mainstream-geology type: calculation access: preprint direction: complicates

Both PDFs on disk, read in full.  The pair that kills the inference in both directions, and the article should use them that way.

Lineweaver and Davis derive from rapid biogenesis that the probability of life on an Earth-like planet older than 1 Gyr exceeds 13% at 95% confidence — and flag their own load-bearing assumption, that "rapid biogenesis is not such a prerequisite" of our existence.  Spiegel and Turner attack exactly that assumption:  under anthropic selection, rapid appearance "is consistent with an arbitrarily low intrinsic probability of abiogenesis for plausible uninformative priors."

One datapoint, observed by the life it produced, supports no inference at all.  That kills the naturalist's "fast means easy" and the apologist's "fast means no time" with the same argument.

The nature of the last universal common ancestor and its impact on the early Earth system

Moody, Edmund R. R., et al.  Nature Ecology & Evolution 8 (2024): 1654-1666.

area: luca camp: mainstream-chemistry type: calculation access: OA direction: complicates

PDF on disk, read.  Catalogued in full under LUCA below.  It belongs here too because of a trap:  Moody et al. date LUCA to ~4.2 Ga and remark that this "suggests that the process required a surprisingly short interval of geologic time" — but they reach that date only by explicitly rejecting the Late Heavy Bombardment as "a credible maximum constraint," citing Boehnke & Harrison.

An article cannot cite Moody for a short window and cite the LHB for a short window at the same time.  The two arguments cancel, and noticing that is worth more than either.

What space delivers, and panspermia

Space is full of organic molecules.  That has not been in dispute for fifty years, and any apologetics text still saying the meteorite organics are contamination is wrong on the data.  What space has never been shown to deliver is anything above a monomer.

Extraterrestrial nucleobases in the Murchison meteorite

Martins, Zita, et al.  Earth and Planetary Science Letters 270 (2008): 130-136.

area: meteoritics camp: mainstream-chemistry type: measurement access: preprint direction: supports

Opened in full as the author preprint (arXiv:0806.2286);  Elsevier paywalls the journal version.  The paper that settles the contamination question by isotope rather than by assertion.

Uracil at δ¹³C = +44.5‰ and xanthine at +37.7‰, against a soil control uracil at −10.6‰.  Terrestrial biology does not make carbon that heavy.  The values "indicate a non-terrestrial origin."

A detail worth carrying:  cytosine degrades with a 17,000-year half-life and guanine converts to xanthine with a 1.3 Myr half-life at 0 °C and pH 7, so meteoritic nucleobase distributions reflect degradation as much as synthesis.

Oba et al., Nature Communications 13 (2022): 2008 — abstract only, open access — later reported all five nucleobases including cytosine, uracil and thymine in carbonaceous meteorites down to parts per trillion.

Abundant ammonia and nitrogen-rich soluble organic matter in samples from asteroid Bennu

Glavin, Daniel P., Jason P. Dworkin, Conel M. O'D. Alexander, et al.  Nature Astronomy 9 (2025): 199-210.

area: meteoritics camp: mainstream-chemistry type: measurement access: paywalled direction: complicates

PDF opened in full, including Extended Data Table 4.  The cleanest sample-return chemistry ever measured, because the material never touched the ground.

OSIRIS-REx returned 121.6 g of Bennu in September 2023.  The inventory:  33 identified amino acids, fourteen of the twenty life uses;  glycine most abundant at 44 nmol/g, about three parts per million by mass, and mostly unhydrolysed;  all five DNA and RNA nucleobases;  about 10,000 nitrogen-bearing chemical species;  and ammonia twelve times higher than Murchison and seventy-five times higher than Ryugu, itself reported as a lower limit.

The extraterrestrial origin is established rather than assumed — racemic amino acids, a violation of Chargaff's rules, and non-protein amino acid diversity "strongly supports an extraterrestrial origin."

The chirality result is catalogued in full under homochirality above and is the finding that matters most:  every chiral non-protein amino acid that could be resolved is racemic, and the authors say this "challenges the hypothesis that the emergence of left-handed protein-based life on Earth was influenced by an early Solar System bias toward L-amino acids."

The 2025-2026 Bennu follow-ups

area: meteoritics camp: mainstream-chemistry type: measurement direction: complicates

All post-cutoff, and they are the reason this section could not be written from memory.

Oba, Y., et al., Communications Chemistry 9 (2026): 132, published 2 April 2026.  CC BY-NC-ND, full text opened via Europe PMC.  The distribution of nucleobases and N-heterocycles in Bennu:  pyrimidines are more abundant than purines, like Orgueil but unlike Murchison and Ryugu, suggesting "preferential synthesis in ammonia-rich ices from the outer Solar System."  High urea concentrations are implicated in the synthesis.  And the detail that matters most for the argument:  non-canonical nucleobase pairs — structural isomers sitting alongside the canonical five.  What space delivers is a mixture, not a reagent bottle.

Furukawa, Y., et al., "Bio-essential sugars in samples from asteroid Bennu," Nature Geoscience, published online 2 December 2025.  Never opened — nature.com redirects to an identity provider and there is no Europe PMC record.  The reported figures — ribose at about 0.097 nmol/g and glucose at about 0.35 nmol/g, the first extraterrestrial glucose, plus lyxose, xylose, arabinose and galactose — come from press coverage only.  Do not quote them without opening the paper.  If it holds, every chemical component of a ribonucleotide has now been found in one body.  No ribonucleotide has.

Mojarro, A., et al., PNAS 122 (2025): e2512461122.  Abstract only, open access.  A tentative fifteenth amino acid, tryptophan, which "has not been detected previously in extraterrestrial materials" — hedged in the paper as "tentatively" and "trace signals," a hedge popular coverage dropped.  Also reports that soluble and insoluble organic distributions differ between stones, so the parent body was processed heterogeneously.

Endogenous production of organic compounds by comets and asteroids

Chyba, Christopher, and Carl Sagan.  Nature 355 (1992): 125-132.

area: meteoritics camp: mainstream-chemistry type: calculation access: abstract-only direction: supports

Abstract only;  body not opened.  The mainstream result the article concedes.  If the early atmosphere was only weakly reducing — which current geochemistry says it was — then interplanetary dust particles "may have been the dominant source of prebiotic organics" on earth.

And it stops exactly where the problem starts.  Every compound in the delivered inventory is a monomer.  No nucleoside — no base joined to a sugar — has been reported from any meteorite or returned sample.  No nucleotide.  No peptide longer than a trace dipeptide.  No oligonucleotide.  No membrane vesicle assembled in place.  The Solar System is good at making bricks and has never been shown to make a wall.

Directed Panspermia

Crick, Francis H. C., and Leslie E. Orgel.  Icarus 19 (1973): 341-346.

area: panspermia camp: atheist-explicit type: philosophy-paper access: paywalled direction: complicates

Opened in full from the PDF.  (A library note:  the on-disk scan is a bundle that also contains two pages of unrelated 2003 Lancet correspondence — an index.tsv error that should be corrected.)

The argument, which is more careful than its reputation.  Natural panspermia is "unlikely" — radiation kills spores in transit, and the probability of an interstellar meteorite arriving is "so small."  So they propose deliberate seeding, while stating that "the scientific evidence is inadequate at the present time to say anything about the probability."

Their pre-emptive answer to "you have just moved the problem" deserves engagement rather than a wave.  "It is quite probable that planets not unlike the Earth existed as much as 6.5 × 10⁹ yr before the formation of our own Solar System… there may be other types of planet on which the origin of life ab initio is greatly more probable than on our own… such a planet may possess a mineral, or compound, of crucial catalytic importance, which is rare on Earth."

Why it nonetheless fails.  Fifty-plus years later nobody has named the mineral.  And the field's actual difficulties — the water paradox, the asphalt problem, cross-inhibition, ribose instability, the error threshold — are thermodynamic, kinetic and informational.  They travel with the chemistry regardless of which planet it happens on.

Their two pieces of evidence have both collapsed, and they hedged both at the time.  Anomalous biological molybdenum abundance turns out to track ancient ocean bioavailability rather than crustal abundance.  And the genetic code is not universal — twenty-plus variants are catalogued — so a frozen accident is not the only explanation for the similarity.  Their own hedges:  "we cannot conclude anything from this single example," and "such prejudices find no scientific support of any weight."

Did Crick recant?  Not demonstrably, and the article must not say he did.  Crick wrote a whole book defending the hypothesis eight years later, Life Itself (1981);  four archive.org scans exist and all are lending-restricted, the djvu text returns HTTP 401, and search-inside 404s — the book was never opened.  The 1993 Orgel and Crick FASEB J. piece "Anticipating an RNA world" was not opened (Wiley bot-blocks, Europe PMC has a record but no abstract);  a talk.origins FAQ claims it never mentions directed panspermia, and that claim is unverified.  A 2013 Scientific American guest blog, read in full, says Orgel abandoned the idea while "Crick continued to advocate for its viability" — and supplies no quotation from either man.

The defensible narrower claim:  Crick and Orgel offered the hypothesis on evidence they themselves called too feeble, and both of their evidential hooks have since collapsed.  That needs no recantation.

And it is not a friend of the young-earth position.  Their scenario requires a 13-billion-year-old galaxy and transit times of 10⁵ to 10⁷ years, and their proposed senders are finite, evolved, physical beings who "resembled us psychologically."  Cite the paper for the difficulty, never for the conclusion.

The Hoyle-Wickramasinghe programme, and where it ended

area: panspermia camp: unaffiliated type: book access: book direction: undercuts

None of the Hoyle and Wickramasinghe books were opened, and the page citations for the famous quotations disagree with each other across secondary sources — 24, 141 and 148 for the same line.  Do not use these quotations without somebody reading the book.  Hoyle's own argument is available in a source that was opened in full, his 1981 Caltech lecture, catalogued under sequence space.

Where the programme ended.  Steele et al. (33 authors), Progress in Biophysics and Molecular Biology 136 (2018): 3-23 — PDF opened in full — proposes that the Cambrian explosion was caused by new genes arriving in "cryopreserved and matrix protected fertilized Octopus eggs" delivered by icy bolides.

A theory whose defenders end up proposing frozen octopus eggs from space is not one to cite approvingly.

The 2003 Lancet correspondence on a cosmic origin for SARS, read in full from the same bundled scan as Crick and Orgel, is the cleanest case.  The decisive rebuttal is that the reported organisms' 16S rRNA sequences match terrestrial Bacillus and Staphylococcus — independently originated life would not share our ribosomal RNA, so the match proves contamination.  Pushpa Bhargava, co-designer of the cryosampler, reported that his own institute "was unable to detect any microorganisms" using the same method and declined co-authorship.

Transfer survival, and what it does and does not show

Kawaguchi, Yuko, et al.  Frontiers in Microbiology 11 (2020): 2050.  ·  and  ·  Mileikowsky, Curt, et al.  Icarus 145 (2000): 391-427.  ·  and  ·  Wesson, Paul S.  Space Science Reviews 156 (2010): 239-252.

area: panspermia camp: mainstream-chemistry type: experiment access: OA direction: complicates

Kawaguchi and Wesson opened in full (the latter as arXiv:1011.0101);  Mileikowsky abstract only.

Kawaguchi et al. exposed Deinococcus pellets to open space for three years on the ISS and extrapolate that 1 mm pellets would endure two to eight years.  A typical Mars-to-Earth meteoroid transit is about 10⁷ years, so the fast transits that fit the survival window are rare.  Mileikowsky et al. conclude that "if microbes existed or exist on Mars, viable transfer to Earth is not only possible but also highly probable" — note the conditional opening;  this is a paper about transport, not about origins.  And Wesson closes the interstellar case:  "the vast majority of organisms reach a new home in the Milky Way in a technically dead state."

Panspermia's most-repeated proof should stop being told.  Rummel, Allton and Morrison, presented at the LPI Sample Return Systems workshop 2011, opened in full — John Rummel was NASA's Planetary Protection Officer — dismantle the Surveyor III Streptococcus mitis story:  the organism is normal mouth flora and was found on the astronauts themselves;  no negative control was run;  only one of thirty-three samples was positive;  the chain of custody was open;  and recovered 16 mm film shows a worker in short sleeves without gloves exhaling directly at the foam sample immediately before the only positive result.

What panspermia actually buys, quantified

Behroozi, Peter S., and Molly Peeples.  MNRAS 454 (2015): 1811-1817.

area: panspermia camp: mainstream-geology type: calculation access: preprint direction: complicates

Opened in full.  About 10²⁰ Earth-like planets in the observable universe, and the Solar System formed after 80% of them already existed.

So panspermia buys roughly twenty orders of magnitude in sites and about one in time — twenty-one in total.  That is a real number, and against probabilities whose exponents nobody can pin down it is negligible.  Both sides should use it and neither does.

Mars, briefly

McKay, David S., et al.  Science 273 (1996): 924-930.

area: biosignatures camp: mainstream-geology type: measurement access: abstract-only direction: complicates

Abstract only;  body not opened, and the rebuttal papers were not individually chased.  The ALH84001 "possible relic biogenic activity" claim.  All four lines of evidence — polycyclic aromatic hydrocarbons, carbonate globules, magnetite, and "nanofossils" — have since been given abiotic explanations, and the claim is not the consensus.  Usable only as a one-sentence cautionary tale.

The apologetics literature, audited

Every source in this group was fetched with its URL and HTTP status recorded on 2 August 2026, and every quotation comes from a page actually opened.  A note on access that contradicts this project's own standing assumption:  creation.com returned HTTP 200 to ordinary desktop-user-agent curl on that date, having returned 403 to other runs the same day.  The block is intermittent or user-agent-dependent.

These entries are tagged by the direction they cut against the apologetics case, which is why so many of them read undercuts.  That is the point of the page.

Evolution, Thermodynamics, and Entropy · Entropy and Open Systems

Morris, Henry M.  ICR Impact, May 1973 and October 1976.

area: apologetics camp: creationist type: web-page access: OA direction: undercuts

Both opened in full, live at HTTP 200.  The founding statement:  the second law is "the most devastating and conclusive argument against evolution," and the open-system reply is "specious."  Morris then proposes his own restated law invoking an "informational program" and an "ingestion-storage-converter mechanism."

That restatement is not a thermodynamics statement at all.  It is prose Morris wrote.  The physics is answered quantitatively by Bunn 2009, catalogued under sequence space:  sunlight pays the entropy bill for all life on earth in under a year, against four billion years available.  Dead.

The Mystery of Life's Origin, chapters 7-9

Thaxton, Charles B., Walter L. Bradley, and Roger L. Olsen.  1984.

area: apologetics camp: id-movement type: book access: OA direction: undercuts

Author-authorised excerpts opened.  The founding book of the movement on this subject, and it concedes the physics on the first pass:  "Without a doubt, the atoms and molecules which comprise living cells individually obey the laws of chemistry and physics, including the laws of thermodynamics.  The enigma is the origin of so unlikely an organization of these atoms and molecules."  Elsewhere:  "it is thermodynamically possible to develop complex living forms."

Their real argument is about configurational entropy work — the work of specifying one sequence out of many — which has no known prebiotic coupling mechanism.  That is a probability argument wearing a thermodynamics costume, and it should be evaluated as one.

And their own arithmetic undercuts the emphasis.  In their tables the coding-work term is the smallest of the three thermodynamic terms:  for a protein, 15.8 cal/gm of coding work against totals dominated by chemical and thermal work;  for DNA, 2.39 cal/gm of coding work against 65 cal/gm of thermal entropy work — twenty-seven times smaller.

The substantive creationist literature stopped making the thermodynamic claim in 1984.  The article should say so, and then say what follows.

The Second Law of Thermodynamics: Answers to Critics · Thermodynamics and evolution · Arguments we think creationists should NOT use

Sarfati, Jonathan.  Creation Ministries International, 2006, 2006 and 2011.

area: apologetics camp: creationist type: web-page access: OA direction: complicates

All opened, HTTP 200.  The finding here corrected this project's own assumption and is worth stating carefully.

Neither major young-earth ministry has retired the scientific second-law argument.  Answers in Genesis' "Arguments to Avoid" index and CMI's "Arguments we think creationists should NOT use" page each list only the theological version — the claim that the second law began at the Fall.  Both maintain separate pages defending the scientific version.  CMI's, by Sarfati, explicitly anticipates and rejects the open-system reply:  "Open systems still have a tendency to disorder… Raw energy cannot generate the specified complex information in living things.  Undirected energy just speeds up destruction."

And in the same body of work Sarfati gives the right advice twice.  "I suggest that thermodynamic arguments are excellent when done properly, and the 'open systems' canard is anticipated.  Otherwise I suggest concentrating on information content."  And, from 2011:  "I tend not to use entropy arguments at all for biological systems… if I don't use it, then few people should."

That is a concession in substance while keeping the claim in name, and it points at where the real argument is.  Sarfati also draws a distinction the article should adopt:  the origin of the first life is different from biological evolution, "since natural selection can't occur without two or more self-reproducing entities."  That distinction is correct, and it is the reason the narrow prebiotic version of the information argument survives while the biological version does not.

15 loopholes in the evolutionary theory of the origin of life · Origin of life: the chirality problem

Sarfati, Jonathan.  Creation Ministries International, 2000 and 1998, both updated 2024.

area: apologetics camp: creationist type: web-page access: OA direction: complicates

Both opened, HTTP 200.  The chirality piece is catalogued under homochirality above.  Of the fifteen loopholes:  the wrong-atmosphere claim is overstated (see Cleaves 2008);  the ribose-and-cytosine claim has been dead since 2009 (see Powner 2009);  and the coding-system loophole is overstated by relying on Gitt's circular definitions.

The 10^-5000 enzyme improbability figure is dead for the same reason Hoyle's is:  it treats two thousand enzymes as simultaneous independent single-target draws, against a directly measured functional-protein density of about one in 10¹¹.

Information, science and biology

Gitt, Werner.  Journal of Creation 10, no. 2 (1996): 181-187.

area: information camp: creationist type: web-page access: OA direction: undercuts

Opened in full, HTTP 200.  Numbered "theorems":  a code is necessary for information (4);  only semantics is information (9);  information always traces to a mental source (10).  Conclusion:  any code implies a mind.

It is circular by construction.  Theorem 9 defines information as carrying semantics;  Theorem 10 asserts that semantics requires a mind;  so anything called information has a mind behind it by stipulation, not by evidence.  It is also not peer-reviewed outside a creationist journal.

The replacement is better and comes from a hostile witness.  Koonin and Novozhilov 2009 — mainstream, NCBI, opened in full — say the code is not literally universal, that "much of the evolution… could be a combination of frozen accident with selection for error minimization," and also that the origin of the coding principle is "a dreary vicious circle" and "perhaps the most formidable problem of all evolutionary biology."  A design advocate can have the strong half of what Gitt wanted, from a source that cannot be dismissed, without the circularity.

Why the Miller-Urey research argues against abiogenesis

Bergman, Jerry.  Journal of Creation 18, no. 2 (2004): 28-36.

area: apologetics camp: creationist type: web-page access: OA direction: undercuts

Catalogued in full under the atmosphere section above.  The single worst item found in the corpus:  a factual claim about Hadean geochemistry — the atmosphere "was probably O₂-rich" — footnoted to a companion book for a 1998 PBS television series, and wrong by six orders of magnitude against the measured constraint.

The Origin of Life: DNA and Protein · The Origin of Life

Parker, Gary.  Answers in Genesis, from Creation: Facts of Life, 2006.  ·  and  ·  Menton, David.  Answers in Genesis, first published 1993.

area: apologetics camp: creationist type: web-page access: OA direction: undercuts

Both opened, HTTP 200.  Catalogued together because between them they demonstrate the provenance problem.  They give two mutually incompatible sources for the same Hoyle quotation — Menton attributing it to "a recent address at Cal Tech" in 1993, against the standard Nature 294:105 citation that turns out to be a news story about Albania.  A quotation with three provenances has none.

To Parker's credit, his page notes honestly that Hoyle was agnostic and Wickramasinghe an atheistic Buddhist.

Can mutations create new information?

Carter, Robert.  Creation Ministries International, 2011, updated 21 February 2025.

area: information camp: creationist type: web-page access: OA direction: undercuts

Opened in full, HTTP 200.  The strongest concession in the entire apologetics corpus, and it is by CMI's own geneticist.

"The phrase, 'Mutations cannot create new information' is almost a mantra among some creationists, yet I do not agree."  And:  "Can mutations produce new information?  Yes… Can they account for the evolution of all life on Earth?  No!"

That retires the "information cannot increase" claim as stated, for biological evolution, from inside the movement.  What survives is the narrow prebiotic version Sarfati identifies — that no mechanism has been demonstrated for generating the first sequence-specific functional polymer in the absence of replication — and that version is defensible.

One warning about the retreat position.  AiG and CMI's reformulation — "we have yet to find a mutation that increases genetic information in ways needed for microbes-to-man evolution to be feasible" — is unfalsifiable as worded, because no observed increase will ever be judged sufficient.  It also owes the reader a definition of information satisfying Hazen's three requirements, and does not supply one.

Two Experiments in Abiogenesis · An Open Letter to My Colleagues · personal statement on evolution and creation

Tour, James M.  Inference 2, no. 2 (2016);  Inference 3, no. 2 (2017);  jmtour.com, updated August 2019.

area: apologetics camp: christian type: popular access: OA direction: complicates

All three opened in full.  Tour is a synthetic chemist at Rice and his chemistry objections stand on their own merits — but he carries no weight as a hostile witness, because he is an open apologist.  The article should disclose that rather than presenting him alongside Orgel, Benner, Shapiro, Joyce, Koonin and Szostak.

His strongest material is a chemist's objection rather than an impossibility claim.  On a purine-synthesis result:  "the team purchased its ribose," and the steps were purified, ordered and temperature-controlled.  That is "a person made it," not "it cannot be made."  From the Open Letter:  "Life should not exist… We are nowhere near solving this problem."

Three weaknesses the article should know about before citing him.  He cites Wikipedia's "Minimal Genome" article for a load-bearing number — 256 genes — when JCVI-syn3.0's 473 was already published.  His figure for the smallest bacterial genome, 1,354 proteins, is far off:  Carsonella ruddii has about 182 and M. genitalium about 482.  And his best section is on lipids and membranes, which is his own field, while his weakest are on genomes and geology.

His own disclaimer is worth printing:  "I do not know how to use science to prove intelligent design."

Signature in the Cell, Appendix A: "Some Predictions of Intelligent Design"

Meyer, Stephen C.  Pages 481-496.

area: apologetics camp: id-movement type: book access: book direction: undercuts

Read in full from an archive.org scan, used only for quote verification.  This is the movement's own most detailed published statement of what it predicts, and auditing it is worth more than any hostile source could be.

The Discovery Institute's own public list of four testable predictions contains nothing about the origin of life — three concern the fossil record and one concerns junk DNA.

The one genuinely unfalsified prediction.  That ribozyme research "will reveal an insufficient number of enzymatic functions to sustain a primitive cell."  Score it honestly:  twenty-six years of in vitro selection have not produced a self-replicating ribozyme, and no RNA-only route to translation exists.  It stands.  But it does not discriminate, because the identical prediction is made by Robert Shapiro, Leslie Orgel and Steven Benner, none of them design advocates, all arguing that the RNA world as usually stated cannot work — and by metabolism-first researchers who infer no designer from it.

The second half of the same paragraph is the problem.  Successful ribozyme engineering will show that "active information provided by ribozyme engineers will account for much of this sequence specificity."  That is an accommodation generator.  Every successful in vitro evolution result is absorbed in advance, because any experiment has an experimenter.  No result could count against it.

The same structure recurs.  The genetic-algorithm prediction is a disjunction — programmer input, or lack of biological realism, "or both" — covering every possible outcome.  The dysteleology prediction is that apparently poor design will reveal "either (a) functional reasons for their design features or (b) evidence of degenerative evolution."  Secretly good or secretly decayed.  There is no third option and therefore no possible falsifier.

And the appendix concedes that design advocates predict opposite things.  On phylogenetics, common-descent design advocates "predict that phylogenetic analyses performed on different molecules and structures in two species should yield harmonious trees of life," while polyphyly advocates "predict that phylogenetic analyses would often yield conflicting trees of life."  Meyer's answer is that hypotheses generate predictions only with auxiliary hypotheses, "in this, design advocates are no different from advocates of other scientific theories."

That reply is technically correct and it costs him the argument.  Auxiliaries are legitimate when they make an additional risky commitment that gets checked — Newtonian mechanics plus a specific unseen mass at a specific position predicted Neptune.  ID's front-loaded and discrete auxiliaries make no such additional checkable commitment and are chosen after the data are in.

The claimed success story fails on four counts.  Junk DNA:  it is not about the origin of life;  the premise is false as stated, since population genetics predicts junk in large-genome species with small effective population sizes and streamlined genomes in bacteria, which is what is observed and is a far more specific prediction;  non-coding function was predicted by mainstream molecular biologists before ID existed;  and no quantity is specified, so "should perform important biological functions" is compatible with 1% and with 100%.

And Meyer removes his own best example from the ledger.  On Wells's centriole work he writes:  "The outcome of his work won't directly confirm or disconfirm intelligent design, or neo-Darwinism for that matter."  That is an honest sentence and it settles the matter.  (Wells's own paper was not opened;  the description above is from Meyer's summary.  The polar ejection force is now attributed in the mainstream literature to chromokinesins rather than a rotating centriole, and that was established from review-level secondary sources only — verify before asserting it.)

The fair summary, and the article is stronger for making it in its own voice:  ID has not produced a risky, discriminating, origin-of-life prediction.  That is not a claim that ID is false.  It is a claim about what the programme has so far delivered, and the primary source for it is the programme's own most detailed statement.

The Sewell withdrawal

Sewell, Granville.  "A Second Look at the Second Law."  Accepted then withdrawn from Applied Mathematics Letters, 2011.

area: apologetics camp: id-movement type: retraction access: closed direction: complicates

The withdrawal notice was never obtained.  What is recorded above is recollection of the episode rather than a fetched source, and the article must not state it until somebody pulls the notice.  The outline as generally reported is that the journal published a withdrawal notice and paid Sewell's legal costs without retracting the substance of its editorial judgment.  A rebuttal by the physicist Dan Styer was downloaded from a university page, but it is a personal web document, not peer-reviewed.

The claims, and where each verdict comes from

area: apologetics camp: mixed type: survey direction: complicates

The audit graded twelve claims.  The evidence for each verdict is catalogued in the relevant area group above.

Four things survive intact, and none of them needs a creationist source.  The water and polymerisation problem;  the interdependence of replication and translation;  the absence of any self-replicating ribozyme;  and the fact that the field's own leaders say in print that the problem is unsolved.  All four are in the mainstream literature, argued by people who expect the problem to be solved eventually.

Hubert Yockey and his own repudiation

Yockey, Hubert P.  Reply to the Foundation for Thought and Ethics amicus brief, Kitzmiller v. Dover, 16 November 2005.

area: apologetics camp: unaffiliated type: philosophy-paper access: OA direction: complicates

Public PDF hosted by the NCSE, downloaded and read in full.  The most instructive single document in this group, because the man who supplied apologetics with its numbers put his objection on the record under his own name.

Yockey's 1977 and 1981 papers are the upstream source of most of the probability figures in The Mystery of Life's Origin and thence of nearly everything downstream.  In 2005 he wrote:

"As the attached correspondence shows, Dr. Thaxton consulted me in 1982 and 1983 and was impervious to anything that refuted his preconceived conclusions."

"I believed, and believe, Dr. Thaxton's theories of creationism, now called intelligent design, must be entirely discarded.  Science has no need for an Intelligent Designer any more than the planets and stars needed angels to move them around after Newton discovered the law of gravity."

On what his own papers were for:  "the purpose of my paper was to give evidence why no origin of life theory based on 'self-organization' was credible… (The missing ingredient needed for the origin of living matter is the genome, not Intelligent Design.)"  And:  "My publications on information theory show that the origin of life is unknowable through scientific methods."

And the complication, which a fair article prints too.  In a 1985 letter that Yockey himself quotes in the same filing, he had written to Thaxton:  "I have no trouble with the conclusions in your book… They are well reasoned and they are supported by the scientific evidence."  His 2005 explanation is that the agreement was narrow — chance and self-organisation scenarios fail — and that his cover blurb was "limited to the point on which we agree."  He also notes that when he refers to materialism he means "the dialectical materialism of Marxism-Stalinism," which is idiosyncratic and worth knowing before quoting him on the word.  He was not an atheist by his own account:  "Theism and atheism both are irrelevant to science."

The article may cite Yockey for the negative claim in his own words — that chance and self-organisation theories of the origin of life are not scientifically valid — provided it reports in the same breath that he rejected the design inference and named the author who misused him.  It must never cite a Yockey probability figure as evidence for design.

Philosophy, method, and the god-of-the-gaps objection

Science and Creationism: A View from the National Academy of Sciences

National Academy of Sciences.  2nd edition, 1999.

area: philosophy camp: mainstream-chemistry type: survey access: OA direction: complicates

Full text read from the NCBI Bookshelf.  Two sentences from the same document, and each side quotes one of them.

"The question is no longer whether life could have originated by chemical processes… The question instead has become which of many pathways might have been followed."

And:  "the consensus among scientists is that none of the current hypotheses has thus far been confirmed."

Print both or neither.

The Ascent of Man · The Perimeter of Ignorance

Drummond, Henry.  Third edition, 1894.  ·  and  ·  Tyson, Neil deGrasse.  Natural History, 2005.

area: philosophy camp: mixed type: popular access: OA direction: undercuts

Both read in full;  Drummond from archive.org, Tyson from the magazine text.

Drummond is the founding statement of the god-of-the-gaps objection, and it was made by a Christian evangelist against Christians.  His theological version is the harder one and is the one the article should answer:  "If God appears periodically, he disappears periodically."  A God invoked only where science has not yet reached is a God who shrinks.

Tyson supplies the modern secular formulation and the Newton-Laplace example, charging design arguments with "embracing ignorance."

A note the article should carry:  Bonhoeffer's 1944 "we shouldn't think of God as the stopgap" line was never checked against a print edition and translations differ.  Do not quote it without one.

Is there anything wrong with 'God of the gaps' reasoning?

Larmer, Robert.  International Journal for Philosophy of Religion 52 (2002): 129-142.

area: philosophy camp: christian type: philosophy-paper access: paywalled direction: complicates

Full text read.  The best available reply, and it is honest about what it costs.

Larmer's distinction is that a gap argument is fallacious only when it rests on ignorance rather than on positive knowledge of causal limits.  And he concedes the corollary that makes the debate empirical:  a gap argument can legitimately strengthen as science progresses, if naturalistic explanation grows more remote rather than less.

That is the version of the argument this literature can actually adjudicate, and it is the reason the audit above matters:  on the atmosphere, on nucleotides, on homochirality, on timing and on ribozyme self-copying, the naturalistic explanation has grown less remote since 1953.  On the origin of coding and on the sequence-specification problem it has not.

Prebiotic plausibility and networks of paradox-resolving independent models · Do-Nothing Prebiotic Chemistry

Benner, Steven A.  Nature Communications 9 (2018): 5173.  ·  and  ·  White, S. B., and P. B. Rimmer.  Accounts of Chemical Research 58 (2025): 1-10.

area: philosophy camp: mainstream-chemistry type: philosophy-paper access: OA direction: complicates

Both opened in full.  Two attempts to say what "prebiotically plausible" should mean.  Benner argues plausibility comes from interconnections across independently grounded models, and traces the word back to Orgel and Lohrman in 1974.  White and Rimmer propose replacing the argument with measurement:  rate constants, kinetic networks, and predictions of what a real geochemical environment does "without direct intervention from a chemist."

That this is now an organised research programme is itself a concession that the intervention objection is real.  Cite it that way, in both directions.

Evolution at the Origins of Life?

Schoenmakers, Ludo, Thomas A. C. Reydon, and Andreas Kirschning.  Life 14 (2024): 175.

area: philosophy camp: mainstream-chemistry type: philosophy-paper access: OA direction: complicates

CC BY, full text read.  The most careful recent statement of where Darwinian explanation starts, and its own summary of the position:  "A comprehensive theory of evolution… starting from prebiotic chemistry… cannot yet be told."

Is abiogenesis part of evolution?

area: philosophy camp: mixed type: review direction: complicates

Two sources settle the historical question and neither settles it the way either side would like.

Peretó, Bada and Lazcano, Origins of Life and Evolution of Biospheres 39 (2009): 395-406 — abstract verbatim, open access.  Darwin held both positions at different times and said so.  Publicly, in an 1863 letter:  it is "mere rubbish thinking… of origin of life."  Privately, in the 1871 warm-little-pond letter:  a natural chemical origin, prefaced with "oh what a big if."

The formulation the article should use:  the origin of life and the subsequent history of life are logically separable questions, and a failure on the first does not touch the evidence for the second.  That concession costs the apologist nothing he can afford to keep and gains him a reader.

The Fermi and rare-earth calculations

Sandberg, Anders, Eric Drexler, and Toby Ord.  "Dissolving the Fermi Paradox."  arXiv, 2018.

area: philosophy camp: philosophy type: calculation access: preprint direction: complicates

Included as background, and with a caution.  Propagating realistic scientific uncertainty through the Drake equation rather than using point estimates dissolves the apparent paradox, because the distribution over the number of civilisations is enormously wide.  The abiogenesis term is the widest factor in it, and nobody can constrain it — which is exactly the point this catalogue keeps arriving at from other directions.  It cannot be used as evidence that abiogenesis is rare;  it is a demonstration that the uncertainty is unbounded.

A closing note on how this catalogue was built

area: philosophy camp: mixed type: review direction: complicates

Every DOI cited on this page resolves.  An HTTP 403 from science.org, pnas.org, royalsocietypublishing.org, Wiley, Elsevier, MDPI, SAGE, Springer, OUP, GSA, Annual Reviews or Liebert means a publisher bot-block on an automated fetch, not a dead link — the same pages load normally in a browser.  Where a page returned 403, the entry names the route by which the text was obtained instead:  Europe PMC, PMC, arXiv, bioRxiv, an institutional repository such as UCL Discovery or Oxford ORA, an author's own site, or the Wayback Machine.

No quotation on this page comes from a page nobody here read, and every entry that rests on an abstract, a supplementary file, a free preview or somebody else's summary says so in those words.  That is the discipline the page exists to demonstrate, and it is the reason a reader who checks one entry should be willing to trust the next.