Recently watched @dsterncardinale’s latest video where he and @talkpopgen react to Rob Carter getting population genetics wrong. Interesting video in any case, but Zach mentioned a reddit post where someone finally figured out how John Sanford rigged his Mendel’s Accountant program to always produce fitness decline unless completely ridiculous values are used.
The reddit post is here:
There’s also a post by Zach himself further down in the comments explaining how the program artificially punishes beneficial mutations with a hidden maximum fitness scaling factor.
Basically the program reduces the impact of beneficial mutations compared to deleterious mutations such that even for mutations with ostensibly the same magnitude of fitness effect (0.01 vs -0.01), the beneficial mutation is then further reduced in effect so it will never be able to compensate for a deleterious mutation even though it is drawn from the distribution of fitness effects with an equal but opposite magnitude of effect.
As someone else points out in the comments, this basically means reversals are impossible in MA, since switching a deleterious mutations with a -0.01 fitness effect back to a beneficial one with a +0.01 will not fully restore the fitness as the beneficial is artificially reduced below it’s actual value. So if you switched the same mutation back and forth, you’d get a fitness decline anyway since you could never recover the fitness loss with the reversal.
So it’s fraud. The program is literal fraud.
Let’s a get a comment from some resident Sanford fans (if any are still around)? @Giltil @stcordova