The literature

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

It is deliberately not a reading list of things that agree with this site.  Several mainstream sources are tagged supports, because they genuinely weaken the standard case against a single couple.  Several creationist and creationist-adjacent sources are tagged undercuts, because their own numbers do not reach a young date.

How to read an entry

Measurement or model.  A count of alleles in a database is an observation.  A population size inferred from a coalescent simulation is a model, and it inherits every assumption in it.  Both are useful and they are not the same kind of thing.

Direction.  Assigned by what the source shows, not by what its author believes.

Verified.  Whether this project read the text directly (full), only its abstract or record (abstract), or only quotations of it inside another source that quotes it verbatim (via).  Nothing is quoted in the main article unless it is marked full or abstract, except Adam and the Genome, which is marked via and is footnoted as such with no page numbers.

Field definitions

Coverage and its limits, stated plainly.  This catalogue covers the published exchange over a single-couple human origin from Ayala 1994 to Swamidass 2022, plus the primary papers each side cites.  Five items could not be obtained and are listed at the bottom;  two of them block checks that matter.

The case against a single couple

Adam and the Genome, chapter 3

Venema, Dennis R. & Scot McKnight.  Adam and the Genome: Reading Scripture after Genetic Science.  Brazos Press, 2017.

camp: mainstream topic: yec-argument method: review direction: undercuts verified: via

Key claims: human effective population size never below about 10,000;  four lines of evidence (allele counts, linkage disequilibrium, PSMC, incomplete lineage sorting);  the Tasmanian devil as the picture of a two-individual founding.

The book that put this question in front of a general audience, and the target of the whole exchange.  No scan could be obtained — archive.org has no item and Open Library's record carries no scan — so every quotation in the article comes through Richard Buggs's verbatim block quotations, which are on disk.  Venema replied to both Buggs posts at length and never disputed the accuracy of any quotation.  The allele-counting argument, which is the strongest of the four, carries no citation in the book;  Buggs asked twice and was pointed to an online HLA primer with no bottleneck model in it.

Adam, Eve and Human Population Genetics (BioLogos 2014, deleted 2021)

Venema, Dennis R.  BioLogos, 2014;  republished with deletions and insertions marked at Peaceful Science Prints.  https://doi.org/10.54739/d67o

camp: mainstream topic: yec-argument method: review direction: complicates verified: full

Key figures: 30% discordance among human, chimp and gorilla gene trees;  ancestral population of about 62,000;  "18 million years" cut to "half a million years."

The long-form 2014 version of the same case, and the only place Venema sets out the incomplete-lineage-sorting arithmetic.  Deleted by BioLogos in June 2021 and replaced with a revised article at a different URL, back-dated to the original date.  The deletions are the substance:  "no point in our evolutionary history over the last 18 million years or more where our lineage was reduced below 10,000 individuals" became "rules out a single couple for at least half a million years," and the entire SNP-counting section went.  The mirror is published by a party to the dispute, so the framing is interested;  the markup is a mechanical diff and is checkable against the live article.

Recent human effective population size estimated from linkage disequilibrium

Tenesa, Albert, et al.  Genome Research 17(4):520–526, 2007.  https://doi.org/10.1101/gr.6023607

camp: mainstream topic: linkage-disequilibrium method: measurement direction: complicates verified: full

Key figures: Ne ≈ 3,100 non-African and ≈ 7,500 African after ascertainment correction;  YRI chromosome-wise mean 6,286 ± 1,357;  alternative recombination estimator moves Ne by −27% on average, range +33% to −45%, mean 1,901;  time window 10,000–200,000 years.

Venema's argument 2, and it does not say what it was used to say.  Its discussion opens "Overall, the estimates of Ne appear to be much lower than the usually quoted value of 10,000."  The 10,000 figure is reached only by adding the African estimate to the non-African one, and the non-African populations are a subset drawn from the African one.  The method also has no reach into the last 10,000 years and assumes a population near equilibrium.

Inference of human population history from individual whole-genome sequences

Li, Heng & Richard Durbin.  Nature 475:493–496, 2011.  https://doi.org/10.1038/nature10231

camp: mainstream topic: coalescent method: measurement direction: complicates verified: full

Key figures: resolution window 20 kya–3 Mya;  an instantaneous tenfold crash at 100 kya smeared over the preceding 50,000 years;  non-African Ne minimum of 1,200 at 40–20 kya.

Venema's argument 3.  A real and powerful measurement whose own validation section says it cannot render a sudden crash as a spike, and which has no power at all more recently than 20,000 years ago.  The copy on disk is the Europe PMC author manuscript;  Buggs quotes the key sentence with different wording, so the two versions differ and the article says which one it read.

The Myth of Eve: Molecular Biology and Human Origins

Ayala, Francisco J.  Science 270(5244):1930–1936, 1995.  https://doi.org/10.1126/science.270.5244.1930

camp: mainstream topic: mhc method: model direction: undercuts verified: full

Key figures: 32 human DRB1 lineages present at 6 Ma;  minimum of 16 individuals;  simulated 510–550 individuals to carry 60 of 70 lineages;  4,490–4,590 with regrowth to 100,000.

The classical form of the argument, and the only one of the four whose logic bites on the human lineage directly.  Its weakness is that the 32-lineage count is derived from exon 2, which is under balancing selection and reshaped by gene conversion.  Bergström et al. 1998 and von Salomé et al. 2007 dated the introns instead and cut the claim.  Ayala's own minimum from the lineage count is 16 people, not thousands.

Molecular genetics of speciation and human origins

Ayala, Francisco J., Ananias A. Escalante, Colm O'hUigin & Jan Klein.  PNAS 91(15):6787–6794, 1994.

camp: mainstream topic: mhc method: model direction: undercuts verified: full

Key figures: 58 DRB1 alleles;  "no fewer than 29 individuals";  "never smaller than two or three thousand individuals";  mean effective size 100,000 over 30 Myr.

The companion paper, and the source of the number everyone quotes.  Worth reading for the gap between its two claims:  the bare lineage count requires 29 people, and the 2,000–3,000 figure is simulation output that depends on the exon-2 mutation model.

The case that the standard arguments do not test the hypothesis

Adam and Eve: a tested hypothesis? / Response to Venema, Part 1 / Adam and Eve: lessons learned

Buggs, Richard J. A.  Nature Ecology & Evolution Community and richardbuggs.com, 29 October 2017, 12 November 2017 and 14 April 2018.

camp: mainstream topic: rebuttal method: commentary direction: complicates verified: full

Key claims: none of the four studies set out to test a single-couple bottleneck;  Venema's addition of African and non-African Ne has no warrant;  incomplete lineage sorting does not exclude a bottleneck in the human lineage alone;  and then, in 2018, that a bottleneck must be at least 500,000 years old to fit.

Buggs is Professor of Evolutionary Genomics at Queen Mary University of London, a Christian, and not a young-earth creationist.  His 2017 posts are the best-credentialed critique of the four arguments and his 2018 post is the honest conclusion of the exchange.  A young-earth article that quotes his 2017 criticisms without his 2018 conclusion is quote-mining him, and he says so himself.  One caution:  his "0.75 of the heterozygosity" figure comes from Nei, Maruyama & Chakraborty 1975, whose Table 3 also shows a founding pair keeping only 23–36% of alleles.  His post prints the first number and not the second.

The Misunderstood Science of Genetic Bottlenecks

Swamidass, S. Joshua.  Peaceful Science Prints, 2022 (revised 2023, 2025).  https://doi.org/10.54739/1w7j

camp: mainstream topic: coalescent method: simulation direction: complicates verified: full

Key figures: median TMR4A 495,000 years;  LD, ILS and Ne called "irrelevant lines of evidence";  trans-species variation permitting a minimum of 5 to 10 individuals over 6 Myr.

The pivot of the whole subject, and it cuts hard both ways.  A mainstream computational biologist concedes three of Venema's four arguments in print, and then supplies the fourth analysis that actually tests the hypothesis and puts the boundary at half a million years.  Not peer reviewed, published in the author's own venue, by a party to the dispute.  The single most load-bearing number against a recent single couple has never been refereed, and this article says so.  Also used (p. 6):  the median TMR10A for Noah's ten lineages, about 180,000 years;  the reply that young-earth criticism of TMR4A "indicates misunderstanding";  note 38, "No simulations of population history are used";  and the claim that the 2021 preprint is "empirically validating TMR4A", which omits the bias that preprint measured.  The p. 6 quotations carry superscript note numbers inside the PDF text layer, so a byte check fails them;  they were read against the bytes by eye.

Brief Population Bottlenecks Are Beyond The Genetic Streetlight

Elliott-Higgins, Sean & S. Joshua Swamidass.  Research Square preprint rs.3.rs-817056/v1, 2021.  https://doi.org/10.21203/rs.3.rs-817056/v1

camp: mainstream topic: bottleneck-detection method: simulation direction: supports verified: full

Key figures: bottlenecks of 2 and 8 for one generation invisible to MSMC2 at 75 kya;  invisible to all methods beyond 250 kya;  Nmin of 6 to 16 individuals at the Youngest Toba eruption.

The strongest thing in print for the creationist side on the population-curve argument.  Not peer reviewed and the second author is an interested party;  the simulations model 2.2 Mb rather than 3 Gb, which could inflate the invisibility result;  and their own Figure 7 says recent bottlenecks are more detectable, so the paper cannot be cited for a Flood bottleneck.  The same figure and p. 7 also show the opposite bias in the dates:  ARG methods "systematically overestimated the coalescence times in simulations with recent bottlenecks", ARGweaver least.  Read from the log axes, a two-person bottleneck at 10 kya comes out near 20 kya with ARGweaver.  And p. 7 gives a three-dataset average TMR4A of 358 kya, lower than the 495 kya usually quoted.  Both are used in the TMR4A section.

Multiple instances of ancient balancing selection shared between humans and chimpanzees

Leffler, Ellen M., et al.  Science 339(6127):1578–1582, 2013.  https://doi.org/10.1126/science.1234070

camp: mainstream topic: mhc method: measurement direction: complicates verified: full

Key figures: 125 shared-haplotype regions, all but two non-coding;  six confirmed ancestral polymorphisms;  the MHC standing out with 11 shared non-synonymous and 7 shared synonymous SNPs.

Trans-species polymorphism is the one form of the lineage-sorting argument that does constrain the human line.  The measured answer is that it is rare and that the number of surviving lineages is small outside the MHC, which is closer to a founding pair's budget than to thousands of ancestors.

The allele-count data

Human triallelic sites: evidence for a new mutational mechanism?

Hodgkinson, Alan & Adam Eyre-Walker.  Genetics 184(1):233–241, 2010.  https://doi.org/10.1534/genetics.109.110510

camp: mainstream topic: allele-counts method: measurement direction: supports verified: full

Key figures: 36,702 transitions, 20,375 transversions and 113 triallelic sites in 896 genes across 90–95 individuals;  0.198% of polymorphic sites triallelic;  no four-allele sites reported.

The hardest single blow to the allele-count argument as it is usually stated.  About 99.8% of human polymorphic sites carry exactly two alleles, so a founding pair's four-allele ceiling is not even binding at the level of a single base.  It does not rescue a recent couple, because the argument that survives is at the haplotype and MHC-lineage level.

A global reference for human genetic variation

1000 Genomes Project Consortium.  Nature 526(7571):68–74, 2015.  https://doi.org/10.1038/nature15393

camp: mainstream topic: allele-counts method: measurement direction: supports verified: full

Key figures: 2,504 individuals, 26 populations, 88 million variants, over 99% of SNPs above 1% frequency.

Cited in the article only for the shape of the data:  the pipeline treats biallelic variants as the main class and adds multiallelic ones onto the scaffold afterwards.  Multiallelic sites are a small residual, not the norm.

IPD-IMGT/HLA Allelelist.txt, release 3.65.0

IPD-IMGT/HLA Database (ANHIG), 14 July 2026.  https://raw.githubusercontent.com/ANHIG/IMGTHLA/Latest/Allelelist.txt

camp: mainstream topic: mhc method: database direction: undercuts verified: full

Key figures computed here: DRB1 4,023 alleles in 13 first-field lineages;  DQB1 3,121 in 5;  A 9,279 in 21;  B 11,258 in 36;  C 9,416 in 14.

The one place the allele-count argument still has force, once it is restated at the lineage level.  Use DRB1's 13:  at DRB1 and DQB1 the first field tracks allelic lineage, while at class I loci it originally tracked serological antigen groups and is not guaranteed to be monophyletic.  DPB1 was excluded because its first field is a running number.

IPD-IMGT/HLA DRB1_gen.txt, release 3.66.0

IPD-IMGT/HLA Database (ANHIG), 14 October 2026.  https://raw.githubusercontent.com/ANHIG/IMGTHLA/Latest/alignments/DRB1_gen.txt  Library record imgt-hla-2026b (YAML only, commit d972ae2).

camp: mainstream topic: mhc method: database direction: undercuts verified: full

Key figures computed here: 680 complete DRB1 gene sequences;  9,042 shared intron sites;  in 500-site windows, the fewest founder sequences that explain every intron family more than 1% apart, allowing patchworks with two switches and two mismatches, is 4 in 2 windows, 5 in 5, 6 in 7, 7 in 2, 8 in 1 and 9 in 1.  Intron divergence between lineage consensuses:  04, 07, 09 against the rest about 10-11%;  01 vs 10 7.06%;  11 vs *13 0.22%.

The measurement the article's Issues list asked for, run by research/drb1-intron-clusters.ts.  It needs no clock and no chimpanzee:  a four-chromosome founding a few thousand years ago allows at most four deep families per window, and 16 of 18 windows need more.  Limits:  named alleles rather than population frequencies;  long gene conversion from the paralogous DRB genes not modelled;  thresholds are choices, though tightening them only raises the counts.  At most nine are needed anywhere, inside the ten-copy ceiling of Noah's family, so the designed-gametes model passes this test.

Recent origin of HLA-DRB1 alleles and implications for human evolution

Bergström, Tomas F., Anne Josefsson, Henry A. Erlich & Ulf Gyllensten.  Nature Genetics 18(3):237–242, 1998.  https://doi.org/10.1038/ng0398-237

camp: mainstream topic: mhc method: measurement direction: complicates verified: abstract

Key figures: alleles within a lineage average 250,000 years old;  over 90% of the 135-plus DRB1 alleles postdate the human-chimpanzee split;  intron coalescence gives Ne ≈ 10⁴ for early hominids.

The single most important source in this dossier for the creationist side, and it also independently confirms the mainstream population size.  It kills the allele-count version of the argument and leaves the lineage version standing.  Abstract only:  closed at the publisher and at every open route tried, so Gauger's "seven ancestral versions" could not be checked against it.  Its companion, von Salomé, Gyllensten & Bergström 2007, is closed in the same way, but its abstract carries the sentence that downgrades the lineage count and is now cited in the article;  see its own entry below.

Full-length sequence analysis of the HLA-DRB1 locus suggests a recent origin of alleles

von Salomé, Jenny, Ulf Gyllensten & Tomas F. Bergström.  Immunogenetics 59(4):261–271, 2007.  https://doi.org/10.1007/s00251-007-0196-8

camp: mainstream topic: mhc method: measurement direction: complicates verified: abstract

Key figures: 15 genomic full-length DRB1 alleles of 10–15 kb, plus 49 complete coding sequences and 322 exon-2 sequences;  the 03 lineage is shared with chimpanzee, bonobo, gorilla and orangutan;  a proto-03 lineage diverged within the last 5 million years into the human-specific lineages 08, 11, 13 and 14;  outside exon 2, most alleles are under 1 million years old.

The single most important source for downgrading the lineage count, and it comes from the same Uppsala group whose 1998 paper destroyed the allele count.  If four of the thirteen first-field DRB1 lineages are a post-split radiation from one ancestor, the pre-split count is about nine, not thirteen.  Abstract only, from NCBI E-utilities, fetched twice from separate sessions.

Reshuffling of ancient peptide binding motifs between HLA-DRB multigene family members

Doxiadis, Gaby G. M., Nanine de Groot, Natasja G. de Groot, Ilias I. Doxiadis & Ronald E. Bontrop.  Molecular Immunology 45(10):2743–2751, 2008.  https://doi.org/10.1016/j.molimm.2008.02.017

camp: mainstream topic: mhc method: measurement direction: complicates verified: abstract

Key claims: most primate DRB alleles are relatively young and species-unique;  no evidence for convergent evolution was found;  the shared exon-2 peptide-binding motifs are therefore reshuffled among duplicated DRB family members rather than inherited across speciation.

This is the paper Gauger's endnotes 7 and 8 actually cite, and it supports her sentence about intron trees grouping by species.  It is why the article's earlier claim that Doxiadis 2012 contradicted her was withdrawn.  Abstract only, from NCBI E-utilities.

Evolution of HLA-DRB genes

Doxiadis, Gaby G. M., Nanine de Groot, Nel Otting, Jeroen H. Blokhuis & Ronald E. Bontrop.  Molecular Biology and Evolution 29(12):3843–3853, 2012.  https://doi.org/10.1093/molbev/mss186

camp: mainstream topic: mhc method: measurement direction: complicates verified: full

Key figures: at least four ancestral DRB gene families before the Old World monkey and hominoid radiation;  most contemporary DRB1 alleles generated in the past 250,000 years.

Open access, mainstream, and recent, so it is the right citation to use in place of the closed 1998 abstract.  It says both halves:  the alleles are young and the lineages are old.  It also contradicts Gauger's claim that DRB intron sequences group by species.

The creationist and design-adjacent models

A Single-Couple Human Origin is Possible

Hössjer, Ola & Ann Gauger.  BIO-Complexity 2019(1):1–20.  https://doi.org/10.5048/BIO-C.2019.1

camp: id-adjacent topic: created-diversity method: simulation direction: complicates verified: full

Key figures: Scenario 1, a couple at about 2 Mya with zero heterozygosity;  Scenario 2, a couple at about 500 kya with primordial heterozygosity 0.012 (roughly twelve times measured human π);  good fits to African allele-frequency spectra and to r² and D′ decay.

The most technically serious pro-single-couple work in print and the only quantitative rebuttal of the linkage-disequilibrium argument.  It is also an old-earth result.  The 100,000-year claim is a conjecture about three model extensions the authors explicitly declined to fit.  No fit at 6,000 years is presented, attempted, or claimed anywhere in this literature.

Genetic Modeling of Human History, Part 1

Hössjer, Ola, Ann Gauger & Colin Reeves.  BIO-Complexity 2016(3):1–15.  https://doi.org/10.5048/BIO-C.2016.3

camp: id-adjacent topic: created-diversity method: review direction: complicates verified: full

Key claims: created founder diversity as a sixth mechanism of genetic change;  four founder chromosome copies, kept explicitly as a ceiling;  haplotype blocks as the falsification test;  African diversity named as the main challenge to a recent Middle Eastern origin.

The clearest statement of the created-diversity model's ceiling anywhere, which is what makes it falsifiable in a way the front-loading models are not.  Its load-bearing sentence, "Each block comes in a few variants, four for many parts of the genome," cites a paper about the human reference genome that does not contain the number.  Its own scenarios are dated at a million years and at 100,000–200,000 years.

Adam and Eve, designed diversity, and allele frequencies

Sanford, John C., Robert W. Carter, Wesley Brewer, John Baumgardner, Bruce Potter & Jon Potter.  Proceedings of the Eighth ICC, pp. 200–216, 2018.  https://doi.org/10.15385/jpicc.2018.8.1.20

camp: creationist topic: created-diversity method: simulation direction: supports verified: full

Key figures: initial allele frequencies of 25%, 50% or 75% in the Designed Alleles Model;  simulated populations of about 1,000 with 100 new mutations per individual per generation;  observed-like frequency distribution reached in 200 generations.

Two very different models in one paper, and they must be separated.  The Designed Alleles Model keeps the four-allele ceiling and is the strongest falsifiable thing the creationist literature has published here.  The Designed Gametes Model abandons the ceiling entirely, and the authors write on p. 213 that it "would appear to be very hard to falsify."  Technical problems:  drift at a census of 1,000 is far stronger than in the real post-Flood population, and the fit is a comparison of normalised curves scored only by a mean squared error that the authors say (p. 206, Table 2) "is not a significance test".  Table 2, ×10⁻³:  Evolutionary 5.75, Evolutionary Adam and Eve 0.70, Designed Alleles 1.16, Designed Gametes 0.84, so the falsifiable model fits worst of the three Adam and Eve models.  The runs last 200 generations (p. 203), so this is a young-earth-date fit, and the article no longer says none exists.  The designed-gametes model is not unfalsifiable once the Flood is counted:  Noah's family caps every locus at ten families.  The paper also miscites the 1000 Genomes paper as "Nature 52:68–74."

On the Origin of Eukaryotic Species' Genotypic and Phenotypic Diversity

Jeanson, Nathaniel T. & Jason Lisle.  Answers Research Journal 9:81–122, 2016.

camp: creationist topic: created-diversity method: model direction: supports verified: full

Key claims: over 80% of an individual's nuclear SNVs created rather than mutated;  rare variants generated in 4,365 years since the Flood;  a predicted number of haplotype blocks that overlaps the observed number.

The only creationist source found here that makes a numerical prediction a mainstream dataset could refute, and it deserves credit for that.  It also made the SNP-level point three years before Buggs did.  Its gap is the same as everyone else's:  the number of haplotype blocks is not the quantity in dispute, because recombination and gene conversion make new blocks and not new divergence between them.

Effective population sizes and loss of diversity during the Flood bottleneck

Carter, Robert W.  Journal of Creation 32(2):124–127, 2018.

camp: creationist topic: created-diversity method: measurement direction: complicates verified: full

Key figures: one individual carries 31.1% of world SNP allelic diversity;  21 individuals for 95%, 57 for 99%;  an Ark founder-equivalent of 1.97–4.7 capturing 55–77%.

An honest measurement on real data that concedes a real shortfall.  It kills the Tasmanian-devil intuition and leaves 23–45% of diversity to be supplied by created diversity or by post-Flood mutation.  Read the statistic carefully:  "allelic diversity" here is a sampling fraction over SNP sites, not a per-locus allele count.  The same paper contains the front-loading escape hatch:  "there is really no limit to the amount of diversity that could have been front-loaded into Adam and Eve."

The Science of Adam and Eve (chapter 5 of Science and Human Origins)

Gauger, Ann.  Discovery Institute Press, 2012, pp. 105–122.

camp: id-adjacent topic: mhc method: commentary direction: complicates verified: full

Key claims: Ayala's exon-2 dating is guaranteed to overestimate;  Bergström's intron study reduced the ancestral count to seven;  intron sequences group by species.

The exon-2 criticism is correct and is confirmed by the immunogeneticists themselves.  The conclusion does not follow from the chapter's own numbers:  seven is still more than four, and the chapter never states when the first pair could have lived.  The "seven versions" figure could not be verified because Bergström 1998's full text is closed.  The intron-tree claim checks out:  her endnotes cite Doxiadis et al. 2008, whose abstract says most primate DRB alleles are young and species-unique.  An earlier draft of the article said Doxiadis 2012 contradicted her;  that reading quoted the tail of a sentence about DRB5 and has been withdrawn.  See quarantine.md, item 4.

Breakthrough for Intelligent Design? (part 3), with Hössjer's reply

Erkell, Lars Johan, and Ola Hössjer.  Biolog(g), University of Gothenburg, 2020–2021;  English translation at Panda's Thumb, 13 November 2022.  https://pandasthumb.org/archives/2022/11/erkell3.html

camp: mainstream topic: rebuttal method: commentary direction: complicates verified: full

Key exchange: Erkell's four-allele objection;  Hössjer's acceptance of the ceiling and of the several-hundred-thousand-year date.

The only substantive response to Hössjer & Gauger 2019 by a working biologist that could be found in any venue.  OpenAlex lists seven citing works and none is critical.  It is a blog exchange, not a journal exchange, and the article says so.  Erkell's headline objection about thousands of alleles is the one Bergström answered in 1998, so it lands more weakly than it should;  his statement about the date is correct and Hössjer does not dispute it.

Genome-Wide Inference of Ancestral Recombination Graphs

Rasmussen, Matthew D., Melissa J. Hubisz, Ilan Gronau & Adam Siepel.  PLoS Genetics 10(5):e1004342, 2014.  https://doi.org/10.1371/journal.pgen.1004342  Library rasmussen-2014.

camp: mainstream topic: coalescent method: inference direction: complicates verified: full

Key figures: human ARGs dated with μ = 1.26 × 10⁻⁸ per site per generation and a constant-size prior of N = 11,534, on 19 time points up to 1,000,000 generations (p. 9).

ARGweaver, the program behind the 495,000-year TMR4A.  Cited for the constant-size prior that Carter objects to, which Elliott-Higgins & Swamidass 2021 show biases dates old after a recent bottleneck.  The PDF's font maps = to ~, μ to m, × to | and − to {, so a byte check fails the p. 9 quotation;  the printed text reads as quoted.

Robert Carter gets everything wrong?

Carter, Robert W.  Creation Ministries International, published 7 June 2021, updated 10 July 2021.  https://creation.com/everything-wrong  Library carter-2023c (key year is wrong;  the YAML date was corrected to 2021-06-07).

camp: creationist topic: coalescent method: commentary direction: complicates verified: full

Key figures: quotes TMR4A as 495 ± 100 kya;  no number of his own.

The only written young-earth response to TMR4A found.  Its argument:  the method assumes constant rates and a population of at least 10,000, and Swamidass's ground rules excluded a specially created or mosaic couple, so it was "not testing the biblical model in any sense."  Half right:  ARGweaver's constant-size prior does bias recent-bottleneck dates old, by about twofold for ARGweaver at 10 kya.  Half wrong:  TMR4A counts four lineages precisely to allow created heterozygosity.  It lists reasons for faster past mutation (radiation, Patriarchal Drive) without pricing any.  Swamidass answered on the Peaceful Science forum thread "Rob Carter Responds to TMR4A" (topic 13962, 2021), read through the Discourse JSON;  that reply is not cited in the article because Swamidass 2022 p. 6 makes the same points in print.

Haplotype blocks

The structure of haplotype blocks in the human genome

Gabriel, Stacey B., et al.  Science 296(5576):2225–2229, 2002.  https://doi.org/10.1126/science.1069424

camp: mainstream topic: haplotype-blocks method: measurement direction: complicates verified: full

Key figures: average common haplotypes per block, 5.0 in Yoruban and African-American samples, 4.2 in European, 3.5 in Asian;  3.1 to 4.9 haplotypes above 5% frequency within any single population;  5.3 in the union across populations;  three to five typically cover ≥90% of chromosomes.

Obtained on 5 September 2026, after being recorded as closed.  Science paywalls it and Unpaywall, OpenAlex and Semantic Scholar all report it closed, but OpenAlex also lists an author copy in the LSU Scholarly Repository, which bin/cdp-fetch.py retrieved through the repository's Cloudflare wall.  It settles the created-diversity model's load-bearing number in both directions:  "four for many parts of the genome" is a fair reading of the measured range, and the African and cross-population values are above four.  What is still missing is the distribution and its maximum, which averages cannot supply.

Assessing the Performance of the Haplotype Block Model of Linkage Disequilibrium

Wall, Jeffrey D. & Jonathan K. Pritchard.  American Journal of Human Genetics 73(3):502–515, 2003.  https://doi.org/10.1086/378099

camp: mainstream topic: haplotype-blocks method: simulation direction: undercuts verified: full

Key finding: real data fit the block model only partially;  a recombination-hotspot model reproduces the observed patterns without any founder interpretation.

This is why the haplotype-block argument is weak evidence in either direction, and why staking the created-diversity model's falsifiability on blocks does not yet buy anything.

The answer that works, and what it costs

The Genealogical Adam and Eve

Swamidass, S. Joshua.  IVP Academic, 2019.

camp: old-earth-christian topic: genealogical-ancestry method: commentary direction: complicates verified: full

Key claims: genealogical ancestry is not genetic ancestry;  Adam and Eve as universal genealogical ancestors 6,000 years ago;  "genetic ghosts" leaving no identifiable DNA;  the condition that their offspring interbred with people outside the Garden.

The most important source here for keeping the article honest.  This model does place Adam and Eve 6,000 years ago and contradicts no genetics, and it buys that by giving up sole genetic ancestry.  A model in which every human descends genetically from one couple gets nothing from it.  Read through archive.org's search-inside on a print-disabled item, with the leaf-to-folio map anchored at both ends of the book.

The Bottleneck Effect and Genetic Variability in Populations

Nei, Masatoshi, Takeo Maruyama & Ranajit Chakraborty.  Evolution 29(1):1–10, 1975.

camp: mainstream topic: bottleneck-detection method: model direction: complicates verified: full

Key figures: a founding pair keeps about 75% of heterozygosity under fast growth and 23–36% of alleles;  worked case is literally two founders.

Cited here only in passing, because it is worked through in full in the sibling article Population Bottlenecks.  Both halves of its result must always be printed together;  each side of this debate has printed one and not the other.

Considered and not used

• Karczewski et al. 2020, gnomAD (141,456 individuals). Fetched and checked for a multiallelic-site fraction. It reports constraint statistics rather than that number, so Hodgkinson & Eyre-Walker 2010 supplies the figure instead.
• Hössjer, Gauger & Reeves 2016, Part 2 (the unique-origin algorithm). Downloaded and stored. It is the algorithm paper; the fitting that the article's argument turns on is in Hössjer & Gauger 2019, so Part 2 adds nothing the article uses.
• Carter & Powell 2016, "The genetic effects of the population bottleneck associated with the Genesis Flood." Worked up in the sibling article Population Bottlenecks, which also records the misprinted fixation-probability formula on its p. 102. Not repeated here.
• Venema 2010, "Genesis and the Genome." In the source library. Superseded by the 2014 article and the 2017 book, both of which state the same case at greater length.

Could not be obtained

1. Adam and the Genome, chapter 3, as a paginated copy. No archive.org scan; Open Library's record has no scan; BioLogos returned HTTP 429 to every direct fetch. The wording is no longer in doubt: Google Books search-inside on volume xRPeDAAAQBAJ returns every sentence quoted in the article, verbatim, from the book itself. It is a reflowable ebook and returns no page numbers, so a print copy or an interlibrary loan is still needed for those.
2. Bergström et al. 1998 and von Salomé et al. 2007, full texts. Both closed at the publisher, is_oa: false on Unpaywall, closed on OpenAlex and Semantic Scholar, subscription required at Europe PMC. Both abstracts were obtained from NCBI E-utilities and both are cited. The gap blocks the check on Gauger's "seven ancestral versions" and on how von Salomé dated the proto-*03 split.
3. Erlich, Bergström, Stoneking & Gyllensten 1996, Science 274:1552–1554, the technical comment on Ayala. Closed, and PubMed carries no abstract. This is the primary for the exon-2 criticism as first published.
4. Doxiadis et al. 2008, Molecular Immunology 45:2743–2751, full text. Closed at Elsevier. Its abstract was obtained from NCBI E-utilities and is what the article cites. The full text would allow the intron trees Gauger redraws to be checked directly.
5. The typeset Nature wording of the Li & Durbin PSMC caveat. The author manuscript on disk and Buggs's quotation of it differ. See quarantine.md.