An explanatory audit of constraints on the genetic code's redundancy ratio: five candidate pathways

The standard genetic code (SGC) distributes 64 codons across 23 signals (20 amino acids and 3 stops), giving an average redundancy ratio r = 64/23 ≈ 2.78. Rather than proposing a further explanation of this value, I ask which existing explanations genuinely constrain it and which merely remain compatible with it. Five candidate pathways are subjected to an explanatory audit: comparative genomics of natural variant codes (P1), information-theoretic channel capacity (P2), translation-error mutation load (P3), tRNA codon–anticodon discrimination capacity (P4), and proteome viability under reduced alphabets (P5). The pathways are found to operate at different epistemic levels. Only P1 yields a narrow empirical window, r ∈ [2.67, 3.05] across twenty-four fully characterised NCBI codes (widening to 3.20 once context-dependent ciliate codes are admitted), and even this is genealogically caveated: the codes are perturbations of a single ancestral attractor, not independent samples. P4 supplies the only binding theoretical constraint, a sharp but loose floor r ≥ 1.83 under standard cytoplasmic wobble. P5 is recast as an informal sanity check, because its per-class minima are calibrated against natural proteomes. P2 and P3 are nonbinding at biological parameter values. The contribution is methodological: an ablation showing that, of five persistent explanations, only comparative genomics carries evidential weight, while the theoretical pathways either fail to bind, fix only a weak floor, or reproduce calibration against the observed value.