Drs. Sanford and Carter respond to PS Scientists

For those of you interested, we have compiled a list of objections and responses.


Here are the PS threads quoted in the article, which are now all public:

What are the Substantive Critiques of Genetic Entropy?

SFT: On Genetic Entropy

The Distribution of the Effects of Mutations

1 Like

Just looking at the parts relevant to stuff I’ve said/written…


Section “1. Mutations & Equilibrium” misses the point entirely. The argument I make is not about the nature of the mutations, it’s simply about the math of Sanford’s claims. The response is all about how well that equilibrium would be past the point of extinction anyway, not all mutations are equally likely, etc.

Not relevant. The objection is that if you take Sanford’s claims of constant fitness effects for a given mutation at face value, you eventually reach an equilibrium point. Period. If they’re conceding that, then great.


Section “2. Natural selection equilibrium” seems to disregard basic evolutionary dynamics, things that are definitionally true based on the meaning of words like “fitness” and “deleterious”.


And in part 6, about the 2012 H1N1 paper…wow. Yeesh. I’ll just comment on the overarching methodological problem: Carter and Sanford purported to track mutation accumulation over decades, but the reference sequence they used (a 2009 pandemic H1N1 strain) was from a distinct lineage relative to the 1918 H1N1 lineage. Just two different lineages of H1N1. So tallying up the differences from 1918 to 2009…they’re not using the correct baseline reference genome. The reference they use is the result of reassortment of a non-human H1N1 lineage with human H3N2.

In other words, the reference sequence and the 1918 lineage do not share a MRCA that was H1N1. The former became H1N1 through reassortment.

Yes, mutations accumulated in the 1918 H1N1 lineage since 1918. But are they accurately characterized here? No. It’s such a fundamental error, if this paper had been reviewed by evolutionary virologists, it would have been caught in peer review.

And that’s putting aside all the other problems, like their claims that fitness declined without actually measuring fitness at any point.


Also, bonus points for the gratuitous potshots from Sanford and in the conclusion.


I read a couple of them that featured me. They were both wildly inaccurate.

@swamidass, what is the point in taking these people’s statements seriously?


What are you alleging was inaccurate? Everything was fully footnoted with word-for-word quotations.

First mention of me:

There is indeed a reference and a quotation from me. The quote is accurate and says nothing like the claim it’s attached to.

I didn’t refuse to explain why nonselectable mutations should have a different spectrum of fitness effects than selectable ones. I explained it at at length. In fact, I explained it in the quotation they inaccurately applied to the previous claim.


There is indeed a reference and a quotation from me. The quote is accurate and says nothing like the claim it’s attached to.

… I was waiting for an explanation that never came. Just an assertion that it was inaccurate. This brings into view the relevance of what we said in the article:

What follows is a summary of six arguments against GE recently made at Peaceful Science , followed by our answers. The references are to forum posts (sub-posts within longer discussions). Each statement will be documented in the footnotes in case these links eventually change or disappear. A disclaimer is also appropriate: we have done our best to give accurate summaries of the intended arguments of these experts, but often we have had to piece together responses from different places, as the intended meanings were often shrouded in opaque or indirect language.

Your motto must be, “It’s hard to get refuted when nobody can figure out what you’re saying.”

Specifically, your statement,

“The precise fitness of the genome will drift up and down very slightly, but changes are as likely to be positive as negative.”

…implies equilibrium (drifting up and down). After review, I don’t believe you’ve been misrepresented, or at least certainly not in a big enough way to warrant any changes, in light of our disclaimer (and in light of the fact that you’re still choosing to remain opaque).

That’s simply false. You did assert it, but at no point have you given any explanation (even when directly asked) as to why very small changes to a machine would not adhere to the same general principle of “easier to break than to improve”. This was, in fact, what you acknowledged as the reason why the distribution is uneven.

You asked what was inaccurate. I told you, assuming that the only way you would fail to recognize the inaccuracy was if you hadn’t read the statement and footnote. Clearly I was mistaken – and didn’t notice that you were one of the people responsible. Now that I’ve gone back and re-read the claim they say I made and the quotation from me, I still have no idea why you think it might be accurate. Could you explain the connection between what I said and what you say I said?

Exactly. What it doesn’t imply is that slightly deleterious mutations accumulate until their overall effect becomes large, and then that they become selectable – which is what the ‘claim’ is. I said that mutations of very slight effect are never selectable and therefore beneficial and deleterious mutations are always in equilibrium, since they were never selected for in the first place. You’re saying that they become selectable. Those are two very different idea. Are you genuinely unable to tell the difference between them?

I’m really trying not to be rude, but before concluding that I’m trying to be opaque, first consider the possibility that you’ve overestimated your own reading comprehension.

As I said above, I provided an explanation in (among other places) the text you quoted above. If you don’t understand the explanation, ask. Telling me that I haven’t provided an explanation while quoting my explanation would be insulting if it weren’t so ridiculous.


That is an egregious falsehood. I seem to recall a lengthy attempt by @glipsnort to get you to wrap your head around an analogy using printers that use toner to print pages.


From the article:

First, Sanford et al. are claiming that these mutations prevent reproduction. That’s not nearly neutral. That is strongly deleterious.

More to the point, there is evidence for an upper limit for accumulated nearly neutral deleterious mutations.

The conclusion says it all.


You are one of many participants here who have contributed to that particular claim, which, when taken as a whole, are represented by our synopsis. You may not have said that exact bit about “becoming selectable” (but you did say the other related things included there), and your exact words were noted for clarity and transparency. That means you were not misrepresented. Obviously space and time would not allow for us to write a full article responding to every statement made by every contributor in the debate, so we consolidated where possible into 6 “classes” of objections. We also made that clear up front.

If you want, I can look into the possibility of editing your footnote there, explaining that you only adhere to two of those three sentences in the synopsis, rather than all three.

I’d forgotten that. Clearly, I wrote all of those posts in an attempt to be opaque.


That was an example of your claim about the genome being mostly junk, which is responded to in a different part of our article. Can you connect the dots for me please? What is the relevance between your claim about the genome being junk and the claim about effectively neutral mutations being even? It would seem to imply you’re saying the e. neutral mutations only happen in junk, but you’ve denied that before as well.

Not a population geneticist, but doesn’t this conclusion require that there be considerable variance in mutational load, at least the portion due to nearly neutral, slightly deleterious alleles among individuals? It seems to me that the GE model must assume, among other things, that there is no significant variance, and that seems like a reasonable assumption to me.

The model I have in my head is the “straw that breaks the camel’s back”. If what Sanford says is true then there will be a mutation that pushes the individual over the line where they can no longer reproduce. That wouldn’t be a nearly neutral mutation, even if that same mutation would have been nearly neutral many generations in the past. It would seem to me that Sanford’s own model has an upper limit on mutational load after which selection will kick in.


Why would you think there would be a significant variance of mutational load between individuals in the same population?

Here you go @glipsnort, here’s an example of the “becomes selectable” claim that was included in our article. Feel free to explain to T_aquaticus why you disagree with this.

A mutation that causes infertility would be selectable, wouldn’t it?

Of course, you would first have to demonstrate how many nearly neutral deleterious mutations would have to accumulate before there are problems. If the number is well past the point of equilibrium for back mutations and compensating mutations then the entire discussion of an upper limit is moot, at least in complex sexually reproducing eukaryotes.

That is only going to happen once the whole population has arrived at the precipice of mutational meltdown.

There is no such thing as an equilibrium point. Back mutations are a non-factor here. This is addressed in the article. Please read it.

And they won’t fall off that precipice because any additional deleterious mutations would be strongly deleterious.

That’s obviously false. As the number of mutations accumulates the chances of a new mutation occurring at the same base increases. If 50% of the human genome is the product of past mutations then there is a 50% chance of each new mutation occurring at a base that was previously substituted. This is why Sanford needs an estimate of how many nearly deleterious mutations it takes before there are issues.

On top of that, you also have to consider the relative effect of nearly neutral deleterious mutations and beneficial mutations. If each nearly neutral deleterious mutation causes a reduction in fitness of 1 unit but a beneficial mutation causes an increase of 50,000 units of fitness then you only need a few beneficial mutations to compensate for many nearly neutral deleterious mutations.

Epistasis is also a factor. Nearly neutral deleterious mutations can become strongly beneficial in the future due to interactions with new mutations.