1. If ape synonymous mutations behaved like Shen 2022's yeast mutants
   median synonymous effect s = 0.0110;  detection floor 0.005
   Ne = 10000, s = -0.0110:  4Nes = 440;  fixation probability relative to neutral = 3.58e-189
   Ne = 10000, s = -0.00500:  4Nes = 200;  fixation probability relative to neutral = 2.77e-85
   Shen's fraction (0.759) would predict Ks/Kneutral <= 0.241 (+ a little from the 0.013 beneficial)
   observed Ks/Ki, autosomes:  0.635 masked, 0.791 unmasked (Lu & Wu 2005)
   largest strongly-constrained fraction the observation allows:  0.365 (masked) to 0.209 (unmasked)
   chimp consortium like-for-like deficits:  non-CpG 0.3, CpG 0.5
   ratio of Shen's predicted fraction to the largest allowed:  2.08 to 3.63x

2. Lu & Wu 2005's nearly-neutral estimate, as selection coefficients
   2Nes 0.5-0.8 with Ne = 10000:  s = 2.50e-5 to 4.00e-5
   Shen's median effect is 275 to 440 times larger
   fixation probability relative to neutral at s = -2.50e-5:  0.582
   fixation probability relative to neutral at s = -4.00e-5:  0.405

3. Human-chimp codon identity (Theistic Evolution ch. 15, p. 499)
   synonymous divergence Ks = 0.0112;  intergenic Ki = 0.0141 (Lu & Wu 2005, autosomes, unmasked)
   synonymous-site identity 98.89%;  intergenic identity 98.59%
   synonymous sites change at 79.1% of the intergenic rate;  a function that fixed each codon would make this near 0%

4. New synonymous mutations per person and their cost
   new coding mutations per diploid genome per generation:  0.775
   synonymous share 0.25:  0.194 synonymous per person;
      Shen-like:  0.147 deleterious at s ~ 0.0110 -> cost 0.00162 per generation (purged, not accumulated)
      Lu & Wu-like:  0.174 at s = 2.50e-5 -> 4.36e-6 per generation;  over 200 generations (~5800 yr) 0.0871%;  over 10,000 generations 4.26%
      Lu & Wu-like:  0.174 at s = 4.00e-5 -> 6.97e-6 per generation;  over 200 generations (~5800 yr) 0.139%;  over 10,000 generations 6.73%
   synonymous share 0.33:  0.258 synonymous per person;
      Shen-like:  0.196 deleterious at s ~ 0.0110 -> cost 0.00216 per generation (purged, not accumulated)
      Lu & Wu-like:  0.232 at s = 2.50e-5 -> 5.81e-6 per generation;  over 200 generations (~5800 yr) 0.116%;  over 10,000 generations 5.64%
      Lu & Wu-like:  0.232 at s = 4.00e-5 -> 9.30e-6 per generation;  over 200 generations (~5800 yr) 0.186%;  over 10,000 generations 8.88%
   (upper bound for Lu & Wu-like:  assumes every such mutation is fixed and none is offset by a back-mutation)

5. Methods that treat synonymous sites as neutral
   Ks/Kneutral = 0.635:  dN/dS overstates dN/dneutral by 1.57x;  a mutation-rate date from Ks is 36.5% too young
   Ks/Kneutral = 0.791:  dN/dS overstates dN/dneutral by 1.26x;  a mutation-rate date from Ks is 20.9% too young
   A fossil-calibrated clock on the same sites absorbs a constant constraint;  the bias then cancels.

6. ClinVar pathogenic counts per possible mutation
   synonymous share 0.25, missense share 0.70:  missense pathogenic calls per possible mutation are 31.1x synonymous
   synonymous share 0.33, missense share 0.63:  missense pathogenic calls per possible mutation are 46.0x synonymous
   raw count ratio missense/synonymous:  87.0