The larger catalogue for the whole series is on the parent page, the origin of life: the literature.
Sources cited
| Source | What it is | Access | Direction |
|---|---|---|---|
| Wong 1975, PNAS 72:1909 | The coevolution theory, on eight precursor-product pairs. | Cited through Koonin & Novozhilov 2009 | complicates |
| Freeland & Hurst 1998, J Mol Evol 47:238 | "The genetic code is one in a million," within a space of block-preserving shuffles. | Abstract only (Springer paywall) | complicates |
| Novozhilov, Wolf & Koonin 2007, Biol Direct 2:24 | The greedy-optimisation computation behind the swap-count and mediocre-peak results. | Opened in full (CC BY) | undercuts the optimality argument |
| Koonin & Novozhilov 2009, IUBMB Life 61:99 | The review that states the two-trillion consequence, the 70-78% minimisation, the 10-12 swaps, the "not locally stable" finding, the cost-function caveat, and the Amirnovin recount. | Opened in full (PMC author manuscript) | complicates |
| Koonin 2017, Life 7:22 | The 10-7-to-10-8 figure with the 1084 denominator in the same sentence, the methodological warning, and the fifty-year verdict on codon assignments. | Opened in full (CC BY) | supports the unsolved claim |
| Shulgina & Eddy 2021, eLife 10:e71402 | 251,571 genomes screened; five new reassignments; the codon-capture route. | Opened in full (CC BY) | undercuts universality |
| DeBenedictis, Söll & Esvelt 2022, Nucleic Acids Res 50:5906 | All twenty transfer RNA isoacceptor classes converted to quadruplets; only nine synthetases tolerate them. | Opened (CC BY) | undercuts "only triplets work" |
| McGowan et al. 2023 | A ciliate decoding UAA and UAG as two different amino acids. | Opened (CC BY) | undercuts universality |
| Wehbi et al. 2024, PNAS 121:e2410311121 | The recruitment order re-derived from ancestral sequences, and the Miller-Urey sampling bias in the textbook order. | Opened (CC BY) | complicates |
| Nyerges et al. 2026 | Recoded E. coli: "synonymous codon replacement — recoding — is frequently lethal." Post-cutoff. | Opened | complicates |
| Miyachi & Ichihashi 2026, eLife 15:RP111164 | The first experimental test of error minimization. Prediction not detected; standard code the multi-property cost optimum among 19,683. Post-cutoff. | Opened in full (CC BY) | complicates, both ways |
| Hashmi et al. 2026 | A converging amino acid chronology from sixteen molecular-complexity metrics. Post-cutoff. | Opened (CC BY) | complicates |
| Goldman, Fournier & Kaçar 2026, Cell Genomics 6:101140 | Reconstructed pre-LUCA synthetases lacking tryptophan, and an ancestor that could not tell leucine from isoleucine from valine. Post-cutoff. | Opened | complicates |
Considered and not used
- Freeland & Hurst 1998 body. Springer paywall. Only the abstract was read. The description of the sampled space comes from Koonin and Novozhilov's account of it, which was opened.
- Freeland et al. 2000, Mol Biol Evol 17:511. The source of the stronger "at or very close to a global optimum" claim. Not opened (Oxford University Press blocks automated access). This project's notes carry a "96% and 100%" figure from its body that could not be verified, and the article prints neither number. Printing 96-100% beside the 70-78% minimisation percentage would be a factual error, because they are different measures on different constraint sets.
- Crick 1968, the frozen-accident paper. Named as a position, not quoted.
- Amirnovin 1997. The recount of the coevolution theory. Not opened; Koonin's report of the three figures is what the article cites, and the footnote says so.
- Ronneberg, Landweber & Freeland's wobble recount, which Koonin says removes the statistical support for coevolution altogether. Not opened; mentioned in the footnote rather than argued from.
- Worning & Ibarra-Chávez 2026, on the twenty-seven NCBI translation tables and near-universality as a horizontal-transfer effect. Not opened; the table count is used and the interpretation is not.
- Koonin 2014's "ominous specter of irreducible complexity" passage. It belongs to the translation article, not to the optimality question, and is used there.
Open questions
- Freeland et al. 2000 should be obtained. It is the source of the strongest optimality claim in circulation and nobody on this project has read it.
- Whether the Miyachi and Ichihashi result holds with more than three reassigned amino acids, and with proteins other than the three reporters used.
- Every figure in this article depends on a cost function chosen by the analyst. Koonin's own caveat about that is quoted; nobody appears to have tested how the rankings move under a genuinely different measure of amino acid similarity.
- An attribution error in this project's own dossier was found and corrected here. The "10–12 evolutionary steps" and "not locally stable" sentences were credited to Novozhilov, Wolf and Koonin 2007; they are in the 2009 review. Any other article quoting them should be checked.