Paul Nelson: Testing for Common Descent (Signal + Noise?)

Signal + noise = all the observations.

I understand why evolutionary biologists protect their core commitment of universal common descent (UCD). We don’t want our central theories bounced around like a beach ball every time anomalous observations pop up.

On the other hand, when UCD is protected from empirical challenge by moves such as “that’s just noise, and we expect noise,” or an indefinitely elastic ring of auxiliary or ad hoc hypotheses (e.g., incomplete lineage sorting, convergence, horizontal transfer, you name it), UCD looks dangerously like an axiom, not a testable proposition.

“Oh come on, Paul,” you say, “no responsible evolutionary biologist thinks that UCD is an axiom.”

Actually – you’d be surprised:

“As a first principle, we adopt the Darwinian idea that all life is related.”

From here, first sentence of the paper:

“First principle” sounds like an axiom. In my academic life (1980-2018), I’ve watched one prediction of UCD after another – e.g., the universality of the genetic code, the conservation of early metazoan development, the homology concept [i.e., homologous genes underlie homologous anatomy] – go belly-up, and yet UCD largely holds its place as central theory.

Can you see why I’d be troubled? (This has little to do, btw, with my opinions about ID. I’d have the same worries if I were a philosophical naturalist and accepted “evolution,” meaning the natural origin of living things from non-living starting materials, as the best explanation for organismal diversity.)

Here’s a thought experiment, or a poll. What would you expect a leading philosopher of biology and evolutionary theory to say about the widely-accepted prediction, from UCD, that the discovery of a fossil mammal in pre-Cambrian strata would overturn UCD?

Please vote and then I’ll post the answer. Polls close at midnight today (10/31), when my wife and I start to consume the leftover Halloween candy.


I disagree. The signal is the data. The noise is not data.

You seem to be arguing that, because there are natural grafts that form between adjacent trees, therefore there are no such things as trees. And because men and women have been known to sleep around, there are no such things as families. And I disagree. The distinction between signal and noise makes sense.

It is normal, both in science and in everyday life, to oversimplify. You are making too much of this.

As for that pre-Cambrian mammal? I have doubted falsificationism for a long time, so I won’t put any money on that argument.


Finding a single mammal fossil in pre-Cambrian strata wouldn’t overturn UCD. It would be treated as an outlier and alternate explanations for its location (i.e strata shift in an earthquake) would be investigated. To cast doubt on UCD would take a major multi-faceted discovery, say a population of creatures living in isolation with a different form of DNA completely incompatible with all other extant life. Or maybe a population with life based on silicon instead of carbon. The conclusion of UCD isn’t based on just a single datum, it’s based on a consilience of evidence from many different sources. A single datum won’t erase all the other consilient evidence.


How about this:

Signal + noise = all the observations.

May I express your opinion on the thought experiment as “present, not voting”?

1 Like

Explain why.

I ask because exactly the same argument was made for the counterfactual existence of variant genetic codes, until they were discovered in 1985.

1 Like

Well, maybe. I see “observation” as a poorly defined term.

And sure, I’m okay with “present, not voting.”

1 Like

He would say it’s nonsense. A Creationist canard. A complete misrepresentation of evolutionary theory. Anti-Science and Anti-Evolution. An expression of ignorance. etc etc

1 Like

I would agree with Mung. The more problematic the discovery is for evolutionary theory the stronger the reaction to try and invalidate it. Logical fallacies fully in play :wink:

Actually I disagree. The signal has information in it and so does the noise. So as Penzias and Wilson realized, there is value (and a Nobel Prize) in studying the noise also.

Also the single + noise = total only works in linear systems. For non-linear systems (all of nature), there is signal + noise AND signal*noise = total.


Not true. This is only true in linear systems. Real systems are nonlinear so there is mixing of the signal and the noise. So signal + noise + signal*noise = total observations

1 Like

Here’s a question before I answer - why are antievolutionists so wedded to obvious, tired, worn-out straw man arguments?

Here’s my answer - show me a life form that does not have, at its core, the ribosome, and I will gladly admit - heck, i will argue so vociferously that I may be mistaken for those in Red Sox Nation that loudly (disrespectfully?) invade ballparks across the country - that this life form does not, cannot share a common ancestry with any of the myriad of variations of the rRNA “kind” that we see on Earth.


10 posts were split to a new topic: The Elusive Pre-Cambrian Rabbit

Actually, it sounds in this case like someone laying out the starting point for a particular argument.

During that same period, you’ve also seen prediction after prediction of UCD validating in truly spectacular fashion as genome after genome has been sequenced and analyzed. Is there a reason you didn’t mention this rather larger aspect of the last 40 years of biology?

Honestly, no. You seem more gleeful than troubled. Someone who was interested in actually testing common descent would be weighing the strength of the signal for CD against the noise, and would be assessing each of the proposed reasons for the noise individually, determining whether there is independent evidence for these processes and whether the data are consistent with said processes.


I’ll see about adding an emoji for that. In the meantime, people can use a hashtag (#moregleefulthantroubled)


1 Like

I did. By completely different non-compatible DNA I mean something like the bases were completely different molecules than adenine, guanine, etc. Not just variations of the genetic code which we already know exist.

Another potential falsification would be the discovery “kinds” were real and there really is a genetic barrier which make it impossible for one “kind” to evolve into another “kind”. But we know that’s not going to happen.

1 Like

33 posts were split to a new topic: Side Comments on Nelson’s Signal + Noise

It depends on what you are looking for. The signal (and data) are what you are normally looking for. But if you start to analyze the noise, then you will be looking at a different kind of data, and you will start to have a different view of what you consider noise.

There’s something subjective about what we take to be signal vs. what we take to be noise. And @pnelson disagrees with how we make that distinction. That’s what the thread is really about.

You can check out the statistical methods and formulas for comparing phylogenies here:

There is also this quote:

Statistical analysis is how science works. If you reject the idea that statistics can be used in science then you are rejecting nearly all of science.


5 posts were split to a new topic: Signals, noise, and the discovery of CBR

Thanks, y’all. Keep the comments coming, and don’t forget to vote on the pre-Cambrian mammal test.

I’ll do an omnibus reply tonight.