Rob Carter Responds to TMR4A

Rob Carter (from CMI) just responded to a video by @dsterncardinale and @GutsickGibbon. This article engages for the first time with TMR4A…

Labeling Courses in Resumes?

Before getting into the scientific details, it is worth addressing this comment by Carter:

Dan is an associate of Joshua Swamidass, who is also a troubler of the creation world and recently called for all creationist courses to be labelled on resumes so that future employers could spot them.1

This is just false. I have not called for “all creationist courses to be labelled on resumes so that future employers could spot them.” This is just false. You can read what I actually called for here: Q&A on “A Compromise on Creationism”.

About TMR4A

A year or two ago, there was much discussion on the “time to the most recent four alleles” (TMR4A) on Swamidass’ Peaceful Science blog.7 This was supposedly a refutation of the biblical timeline and, at first glance, appears to be a formidable argument against us.

It is not a refutation of the bible timeline. I just wrote a book, the GAE, showing that with people outside the garden there is no conflict with the biblical timeline. Instead, TMR4A is a measurement of the genetic data that conflicts with YEC models of human origins that do not have people outside the garden.

Using a program called ARGWeaver,8 one can run models of human history that account for recombination and inheritance. Given a constant mutation rate, a constant recombination rate, a constant generation time, and human population of at least 10,000 individuals, how long do you think it would take for you to reach the common ancestor of most people using this method?

Given a constant mutation rate, a constant recombination rate, a constant generation time, and human population of at least 10,000 individuals, how long do you think it would take for you to reach the common ancestor of most people using this method?

This is just false. ARGWeaver does not run simulations of human history.

TMR4A = 495 +/- 100 kya

A half a million years (give or take)! People see numbers like that and are instantly intimidated. However, Swamidass made this admission when he set out the criteria for the calculations:

“In this conversation we are not allowing for any miracles. For example, the bottleneck couple would not be specially created or be genetic mosaics (with different genomes in every sperm/egg).”

This is not an admission, as if something wrong was done. It is just a clear statement of the model considered. Carter does the same thing in the next paragraph where he considers different models of Adam and Eve in his graph.

Turns out that we also consider genetic mosaic progenitors. The YEC models still require a Noahic bottleneck of 10 genomes in this case, which would mean TMR10A (at 180 kya) becomes the problem. To summarize,

  1. IF (1) Adam and Eve have biology like ours and (2) there are no people outside the garden THEN (3) TMR4A is a measurement of the genetic data that indicates Adam and Eve were more ancient than about 500 kya.

  2. IF (1) Adam and Eve are genetic mosaics with totally different genomes in each gamete and (2) there was a Noahic bottleneck THEN (3) TMR10A is a measurement of the genetic data that indicates Noah were more ancient than about 180 kya.

In that same area, he says they were only considering bottleneck that happened 100 thousand to three million years ago and they never allowed the population to dip down to a single couple.

This is false. Remember, that ARGWeaver doesn’t even simulate populations. It is a measurement of the data that does not depend on any simulations.

Thus, Swamidass admits to excluding all aspects of the creationist model from the get go. How, then, is this a refutation of anything?

Well, I did not exclude the creationist model from the get go. I directly tested Carter’s model, as he has laid it out. It seem that Carter still has not understood how TMR4A is measured.

About this same time, Hössjer and Gauger published a paper based on their own computer model that also claimed a 2-person bottleneck was possible.9 They did not consider “young earth” models and specifically assumed the two founders evolved from a large population that evolved over millions of years. At least the claim that it is “impossible” that the human race started off with but two people has been falsified.

This paper presents data that shows there could be a single couple founding 500 kya ago, basically replicating findings from TMR4A with a different approach, one that requires simulation. You can see our discussion of this paper here: Gauger: A Single-Couple Human Origin is Possible. So this is two papers that show a bottleneck at 500 kya might be consistent with the data.

It seems that Carter is confused here about findings of this paper. These results do not undermine the TMR4A calculation at all.

[Note added August 20, the Gauger study is focused on a very week line of evidence in not sure is valid any ways.]

The TMR4A calculation is an interesting intellectual exercise, but it is assuming too much in favour of the evolutionary position. They were not testing the biblical model in any sense. Instead, they were testing to how far back a single-couple origin for modern humanity could be, given millions of years and a consistently large population with stable mutation rates, etc.

This is just not true. It is a direct test of Carter’s model. TMR4A does not assume “millions of years and a consistently large population with stable mutation rates, etc.” There is no simulation that embeds these assumptions either. Rather, this is just a measurement of the data. In Carter’s model, we expect TMR4A to be at about 6 kya, and TMR10A to be at about 4 kya. Instead, we find that they are at 500 kya and 180 kya, respectively.

Some Other Points

Not everything Carter says in this article is incorrect.

  1. He claims I misunderstand the substitution rate (per year) vs the mutation rate (per generation).

I am perfectly well aware that the long-term substitution rate is not the same as the genealogical mutation rate. However, over long periods of time in a population in mutation-drift equilibrium, the two are nearly equal.

It is true that mutation rate and substitution rates are approximately equal if you use the same units. This comment by Carter is confusing though. The distinction between mutation and substitution rates is not that one is in years and the other is in generations. In fact, both rates can be expressed in either years or generations.

As for the rest of the points in this, it seems It seems that many of these objections would be resolved by replicating his experiments in SLiM. Why not?


A post was split to a new topic: Comments on Rob Carter Responds to TMR4A

Dr. Carter actually missed the point with the substitution rate thing; that specific rate - ~1/10^8 - is per year, but he applied it per generation and assigned a generation time of 30 years. That’s the problem I pointed out. He responded by saying that such rates can be reported differently by different authors (true), but that’s not relevant to what I said.


You have to be more precise as to what exactly Carter wrote to make sense of this. With references of course.

It’s in a video, so references are not available, but he’s talking about mutations accumulating, rather than merely occurring, which means we’re talking substitution rate rather than mutation rate. The number he uses is the approximate per year rate (it’s about 4/10^8 per site per year if I recall correctly, he says about 1, close enough), but he then calls it per generation and gives a generation time of 30 years.

4/10^8 is closer to the per generation rate than it is to the per year rate.

As I understand it the per year rate is ~1.25/10^9 which yields a generation rate (with generation time of 30) to be 3.75/10^8, which is basically 4/10^8. So Carter might not have made a mistake here, though his rebuttal leaves open the question.

Is it possible you missed something here @dsterncardinale ?

He claims I misunderstand the substitution rate (per year) vs the mutation rate (per generation).

I like the fact that the wording of his heading confirms that he in fact does not understand the distinction. That’s a difficult thing to do accidentally, yet he managed it.

1 Like

Absolutely possible I’m missing something, but he was specifically talking about two mutations reaching fixation in sequence, NOT the baseline rate at which they occur.

Can you give time codes in the video and a transcript of what he said please?

If we’re talking about humans, the single-base-pair mutation rate is about 1.4x10-8 per base-pair per generation.

so here’s what they did they
said we’re going to start with a
population of 10 000 individuals
and we’re going to wait for a single
letter to change
there’s an a and we want to change to a
g how long will that take
well you don’t actually need a
mathematical model to do that or
computer model to do that because
we know the basic mutation rate under
a lot of simplifying assumptions the
mutation rate is they usually say it’s
1 times 10 to the negative 8. negative 8
what does that even mean okay math
10 to the sixth is the million 10 to the
eighth is 100 million
10 to the negative 8 is 1 over 100
in other words in any stretch of 100
million letters there’ll be about one
there per year or sometimes per
generation depending on how much you’re
talking about
so let’s talk in generations because
evolution doesn’t care about years it
only cares about


how long does that take well given
a stretch of a hundred thousand letters
one out of one hundred thousand
generations that’s when that will appear
generation times about 30 years that’s
about 3 million years

And later still…

then you have the waiting time to that
lucky one that doesn’t go away
but that’s not the waiting time the
waiting time is how long does it take
that mutation to spread throughout the
population and replace the original
that’s the waiting time population
geneticists talk about something called
fixation or fixity and this is hard
because we don’t use that word fix
like we used to use it in fact
we use it as repair what they mean is
stuck as in fixed in place immovable
can’t change anymore


So he’s talking about fixation rate, not the rate at which the mutations occur.


This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.