Nelson's Test of Common Descent Comprehension

Who is willing to take @pnelson up on this? Please do!

Once a few of you have had a chance to get a peak at it from him, I’d like to explain this:

(Full disclosure, I’m also sure that @pnelson and I are using different answer keys.)

I’m very curious to see how everyone else does, and who else stumbles on what I am sure is the right answer. I’m curious how some of @pnelson’s students would respond to hearing the right answer.

Also, testing a theory by testing student comprehension of it is usually not a good idea. I’m not sure what it proves except that most students do not understand the implications of common descent. A rigorous test of the sort @pnelson is making here was already done with phylogenomics: A Test of Common Descent vs. Common Function. I’ve discussed this with @pnelson before, and I’m looking forward to when he incorporates this into his lesson plans at the Discovery Institute.


It’s unclear to me what’s being talked about here. Is this anything more than an exercise in the maximum parsimony prediction of missing data? What is the purpose behind it?

1 Like

I think the purpose is to demonstrate that common descent is not useful in biology and leads to wrong predictions. However, if you understand common descent correctly in these cases, you don’t make poor predictions, even if the answers are blinded.

Come on @John_Harshman, I hope you take the test. I want to compare answer sheets.

Given that you passed the test, doesn’t that undermine its intended purpose for disproving common descent? Does @pnelson agree that you passed the test?

1 Like

If I recall correctly, he stated that I answered the questions in a different sort of way than anyone else. He did not say I was wrong, but it was not what he expected. I did not cheat, and I’m certain that @pnelson scored no wrong answers. He just received answers in a different form that he expected.

And yes, this does undermine the teleology of the test.


Do i write him for the test? Or do i click on a link for the test?

1 Like

You email him.

1 Like

I did the test. I got different answers from Joshua, but I think we both agree on the central lesson: it’s a fake test, meaning that it’s set up specifically to make evolutionary predictions fail, and that this is done by cherry-picking the cases.


I don’t want to give our answers yet. It will be most fun for others to give it a try first. Then, those that have done it can compare notes here.

But how is that even possible?

If UCD or CD had empirical content of their own, it should not be possible to fake them out — i.e., to trick UCD or CD into giving the wrong answer. Compare:

Surgeon (to hospital technician): I used these surgical instruments today. Please sterilize them.

Technician: I’m sorry, sir, but sometimes bacterial cells self-organize on the tray after we put the instruments in the autoclave.

Surgeon: You idiot — the autoclave is broken. Call the repairman.

Theoretical background: spontaneous generation does not occur. Full stop. There are no exceptions to worry about, and the technician needs to re-learn cell biology.

UCD and CD, by contrast, travel around with enormous and ill-defined ceteris paribus clauses attached to them. These clauses are parasitic, and bleed both UCD and CD of their empirical content. Any damn thing can happen along the branches of a phylogenetic tree, and does. For example, from 1966 to 1985, the genetic code looked universal, and this confirmed UCD. From 1985 to 2018, we observed variant (non-universal codes), and, mirabile dictu, that also was consistent with UCD.

So the phylogenetic exercises aren’t fake. UCD and CD rule out nothing, as the phylogenetic exercises demonstrate.

As I said earlier, the phylogenetic exercises merely make vivid what systematic biologists have known for a long time. John Harshman declined to read Vogt 2007, which I provided to him, but this is relevant:

"Neither such background knowledge as for instance 'descent with modification’, nor any specific tree hypothesis prohibits the occurrence of convergent evolution. This allows for both apomorphy (throughout this paper, I will use the term ‘apomorphy’ to mean structural sameness due to shared common origin and not simply observational similarity) and homoplasy as possible explanations for the sameness of character states and their distribution patterns. A given tree hypothesis is logically congruent with any specific observable evidence of character state distribution. In other words, a given tree, in combination with decent with modification as background knowledge, does not prohibit any specific character state distribution pattern (Farris, 1983; Sober, 1983). As there is no deductive link between any tree hypothesis and any specific character state distribution there exists no direct empirical test of hypotheses of monophyly (i.e., clades) sensu Popper (Sober, 1988; Rieppel, 2003)—one cannot think of any observation, which, in case it would represent a true statement, would allow to conclude the falsity of a clade or a given cladogram through modus tollens.” (p. 4, emphasis added)

From here:

I hope others will try the exercises, because the bigger the sample of outcomes, the more interesting the moral we can draw. Please contact me at or, and I will send the form. I will also provide the source publications after you do the exercises.

Do you really believe that? What you are saying is simply that given UCD, both homoplasy and divergence should be impossible. Why would anyone think that? What you call “any damn thing” is just the inevitable consequence of any evolutionary process. It makes data sets imperfect, and we have to deal with that. But we can, as long as we consider the problem from a statistical perspective. It’s data in the aggregate that tell us about phylogeny, not single characters.

They’re fake because you have chosen only cases in which parsimony fails to predict the result. Do you deny that you chose your questions for exactly that reason? Thus your test is a test only of your ability to pre-select particular results. If you have chosen at random from all possible instances I suggest that evolutionary predictions woud have been correct much more often than not. Of course, as distance becomes too great, parsimony and, eventually, likelihood methods fail to produce results, as data become randomized. But you seem to be claiming a lack of common descent long before that point is reached.

Perhaps, but that isn’t what you think. You realize that I’m a systematic biologist, right? Cladistics, I’m afraid, is a philosopical sink. Now, while I can’t see anything but the abstract, it doesn’t seem as if Vogt is actually rejecting the idea that phylogeny exists or even that we can reconstruct it and find the true tree. Is that correct? So is it possibly that you’re quote-mining here?

So this isn’t a scientific exercise at all; it’s an attempt to teach a predetermined moral. The moral I draw here is one of cherry-picking and, perhaps most charitably, self-deception.

I used to believe that too, until I realized that none of the UCD and CD confirmations I had been taught as solid (such as the universal genetic code for UCD, conserved essential information transfer proteins for UCD, conserved early development of Metazoa and Vertebrata for CD) had survived in the face of new data. Instead, freshly-minted auxiliary or ad hoc hypotheses, such as “developmental systems drift,” were contrived to patch the fit between UCD, CD, and the observations. This happens every day in the evolutionary biology literature, mostly recently with the discovery of orphan and taxonomically restricted genes / proteins, which no one predicted (on the basis of UCD or CD), and in fact – see Jacob’s famous 1977 Science paper on “Evolution and Tinkering” – explicitly denied was possible.

This is the part of your comment that I absolutely refuse to credit. Have you not always been a young earth creationist?

Oh, come on, John. Grow up. Evolutionary biology was my minor at the University of Pittsburgh (1980-84), and evolutionary theory and the philosophy of science were my major field of study at Chicago (1985-1997). My dissertation, “Common Descent, Generative Entrenchment, and the Epistemology of Evolutionary Inference,” classified as Evolutionary Theory (QH call number) doesn’t have one word in it about creationism. You seem to think it is impossible to think (i.e., reason from evidence and logic) in more than one way at a time.

If that were so, education and scholarship would cease overnight.

I sent you the Vogt 2007 pdf last night. Read it. You will see that I am not quote-mining.

Interesting. So when did you become a YEC, and why?

“Are you now, or have you ever been, a member of the Communist Party?”

A mature scholar possesses the ability to do what lawyers call “assume arguendo” – namely, entertain a theoretical structure or system of postulates, to (a) learn how that system works, and (b) to find out what follows from it. Assume philosophical naturalism. OK, now do biology.

Simple as that.

If you want to know about my YEC opinions, read the chapter I co-wrote with John Mark Reynolds on the topic, in the book Three Views of Creation and Evolution (Zondervan, 1999). This thread, however, concerns UCD, CD, and their empirical content. Stay focused.

If I can get anything out of that reply, it seems to say that you were a YEC from the beginning of your college education. That suggests that I was right to doubt your original statement that “I used to believe that too”. Correct?

I’m unable to find that book in my local system, even through ILL.

No? What is your take-home lesson from that paper? The author’s seems to be that Popperian falsification is a poor model for phylogenetics, but that there are other, more suitable models. This doesn’t seem to be your take.

You seem to have this view that only perfectly accurate models are scientific, or that all physical processes only produce one result. As it turns out, nature is really complex and really messy. Our models are necessarily simple and don’t fully describe nature. That is true of nearly every single scientific theory. For example, we are very confident that smoking increases the risk of lung cancer. That is what the science shows. Are there people who smoke and don’t get lung cancer? Yes. Are there people who get lung cancer but have never smoked? Yes. That is the noise in the model. Even with that noise, we are still very confident that the empirical evidence supports the scientific conclusion that smoking increases the risk of lung cancer.

Biology is complex. Biology is messy. There is noise, and we expect to see it. Without absolute complete knowledge of every interaction between every organism in every population we can’t make 100% accurate predictions. That’s science. However, we can predict statistically significant correlations, and that is what we see.