Macroevolutionary Theory Evaporates

Evolutionists boisterously advertise their theory as a fact.
Looking under the hood, though, one finds hot air and broken promises.

 

The burden facing Darwinism is macroevolution—major changes—not microevolution—minor changes.

To really understand Darwinism, one must get out of the indoctrination centers, known as public schools. One must listen in on the smoke-filled rooms where the movers and shakers of the theory gather to wring their hands over the problems, fight each other’s solutions, and plan their strategies. Once in awhile, their deliberations are released in print where investigators can see what is really going on. It’s not that any of them are ready to junk the theory. They remain true believers, assuring themselves of eventual success. But the problems are ugly, nothing like what the public hears. It’s like watching the blood and guts in the sausage factory before the product gets washed, sanitized and wrapped for the public and the schools, stamped with USDA approval.

In our acquaintance with such backroom sessions (where Darwin skeptics are barred; see here and here) solutions usually come in future tense. Long-standing issues thought to have been solved long ago rear their ugly heads again, causing much consternation among the true believers coughing over each other’s smoke. The meetings end in a draw as the participants agree to disagree and take out promissory notes to let the public know that solutions are right around the corner. They come out smiling for the cameras. The public doesn’t realize they have been offered vaporware and futureware.

Matter Arising: OoL Shakeup

Sometimes they don’t get away with it so easy. An example can be seen now in what is happening with origin-of-life (OoL) studies. Around 2007, I heard Dr Steven Benner at JPL admit to a group of scientists and other employees that the problems in his field are so vexing, it was almost enough to make one become a creationist. The audience chuckled nervously. Such an admission must never be heard by the public. Benner listed problem after problem: chirality, genetic takeovers, accumulations of tar, and the difficulty of making ribose. When Benner is on camera, however, he smiles and talks about all the progress they’re making.

That was before Dr James Tour, a tenured senior chemist, decided to take him to task. Tour had issued a series of YouTube videos making public the many and severe problems with OoL research.

After the first series, Benner appeared with Darwin hack Dave Farina, a YouTuber, assuring him that everything was fine, and that Tour was just plain wrong. In a three-part pushback in his second series, Tour explains why Benner is bluffing, lying and disgracing his field of chemistry—laughably so. Apparently, Benner has never had to deal with a steel man like Tour. It remains to be seen if he will confess his lies and admit that the problems with OoL research still remain so vexing, he should become a creationist. Probably not. But the fireworks are amusing tens of thousands of viewers who are watching Benner in the ring with his usual escape routes closed off: presumptive authority, consensus, bandwagon, and bluffing. The public audience can see those tactics don’t shield him from Tour’s blowtorch of knowledge, and they are no longer satisfied with promissory notes.

Did Anyone Tell You? Macroevolutionary Theory Is Collapsing

I ran across a Commentary in PNAS dated 21 February 2023 by Matt Pennell that reads like the minutes of a smoke-filled board meeting of the Darwin Party, where Murphy’s Law is causing headaches for the theory of macroevolution. Here’s the upshot.

All the Darwinists had assumed that macroevolution was solved in theory, thanks to a popular equation in the early 1990s. But then, Pennell and a colleague realized that the solution was vacuous; the same equation could produce an infinite number of different evolutionary histories. There was nothing scientific about the solution! This “thorny problem” (a crisis) Pennell labels “nonparametric nonidentifiability.” That’s a synonym for unspecifiable nonsense!

Pardon the long quotes below, but most readers will not have access to these damning admissions by Pennell, and we want his own statements to be clear and in context. They bear careful reading.

Using lambda (λ) for speciation rate and mu (μ) for extinction rate, and λ_p for effective speciation rate, here’s what he says. For those unable to peer behind the paywall into the smoke-filled room, this is like watching Emperor Charley’s fancy clothes being removed one by one:

Once we figured this out, it was clear that there were many mappings of different functions for λ and μ that would result in the same λ_p and, thus, the same likelihood. We called these alternative functions “congruent models” (collectively, a “congruence class”), as in congruent triangles, because this set of functions form a geometric group. The models in a congruence class are completely nonidentifiable, meaning they cannot be distinguished with any amount of phylogenetic data. In other fields, this type of nonidentifiability of different types of functions has sometimes been referred to as “nonparametric nonidentifiability”. And these different congruent models could potentially tell completely different evolutionary stories; even if one found high support for some specified scenario, there are always alternatives that were equally likely, and one would never know which was closest to the truth. In a follow-up study, Louca and I have shown that nonidentifiability, at least partially, explains persistent yet obviously nonsensical reports that historical extinction rates had been nearly negligible in many taxonomic groups—highlighting that, far from a purely theoretical matter, this hidden identifiability problem had been afflicting empirical analyses for decades.

For decades. DECADES. This is not just a theoretical problem. It touches the heart of whatever empirical (observational, testable) evidence they have presented that macroevolution has occurred. If macroevolutionary histories are nonidentifiable, they cannot rule out histories that involve Creation, can they?

Be astounded at what they hid from you! They’ve been teaching nonsense for decades—a vacuous theory—and they knew it!

Theory Rescue Emergency

Pennell reaches for a ring buoy tossed by Kopperud et al. in the same issue of PNAS, which concludes,

Nonidentifiability of diversification rates is a real issue and should be addressed when studying temporal patterns in diversification histories. Yet, the congruence class may not be a problem when testing for macroevolutionary diversification hypotheses in a qualitative fashion. Based on our selection of datasets and alternative models, we argue that abrupt changes in speciation and extinction rates are robust to the congruence class. Conversely, periods of constant or flat speciation and extinction rates are not robust, and it is trivial to construct congruent but contradictory evolutionary hypotheses. Furthermore, the restriction of the congruence class that all models share the same speciation rate at the present λ₀ often induces diversification rate changes near the present. Rates are robustly estimated in the near present, however, rate shifts near the present are not and should be treated cautiously. As such, we argue that while quantitative rate estimates are difficult to infer, it is still possible to draw qualitative conclusions about the general pattern. Additionally, we conclude that the space of possible speciation rate functions is more restricted than the space of possible extinction rate functions.

What happened to the quantitative rigor that scientists aspire to? “Qualitative” conclusions sounds like a justification for storytelling. Pennell has reservations about their hopeful-sounding solution:

Investigating the nature of congruence classes is a very difficult problem to study in the general case, so Kopperud et al. examined 12 case studies, in addition to simulated datasets. In each one, they fit a series of models using Bayesian inference, identified the best model using conventional model selection techniques, and then explored the qualitative properties of the alternative, congruent models they sampled.

But alas, the ring buoy falls apart. Notice how often he says, “Unfortunately,….”

In my view, there are two major and important discoveries in this paper. First, when rates were inferred to be changing rapidly during the initial model-fitting stage, most of the congruent models also had this property. This should leave us feeling optimistic about inferences about speed-ups or slow-downs to the pace of diversification, which have long been of particular interest to macroevolutionary biologists. Unfortunately, the reverse was not true; all bets were off when the original model found rates to be more or less constant. Second, the uncertainty in inferences brought about by considering the suite of congruent models was generally less than the uncertainty introduced by sampling error alone. If this is found to be generally true across more studies, then perhaps one could consider nonidentifiability a second-order problem, at least for modestly sized datasets. Unfortunately, it seems likely that researchers will have to examine the potential impact of nonidentifiability on a case-by-case basis, but these authors have provided both software and a clear illustration to enable others to follow their strategy.

The software is buggy, too. It’s no better than the qualitative assumptions plugged into it. Pennell’s new solution? Go radical! Begrudge your losses proudly! Embrace the uncertainty!

However, I think that neither the above description nor the text of the Kopperud et al. paper itself conveys just how radical the idea is here. Macroevolutionary biologists are accustomed to discussing the uncertainty of the values of model parameters and are (begrudgingly) comfortable with admitting the limits of our knowledge about events that occurred deep in the past. However, here Kopperud et. al’s work suggests that we can, and perhaps should, embrace uncertainty in the description of the process itself: We cannot identify the true history of lineage diversification, but perhaps we can identify all of the histories that are consistent with our data.

But if there are an infinite number of histories consistent with his data, what good is that? The nonidentifiability remains. Macroevolutionary biologists are, in short, clueless. They cannot figure out any evolutionary process that can account for the history of life.

At this point, the backroom bull session has to adjourn. Pennell needs something to tell the press. And so, as defeated Darwinians are trained to do, he smiles, looks excited like a clapping seal, and appeals to futureware.

If we want to push further than that and gain deeper insights into the biological processes responsible for generating patterns of diversity throughout time [i.e., Darwinian theory], we will need different types of data and knowledge. A seemingly straightforward path forward would be to combine fossil observations and molecular data into a single diversification analysis; unfortunately, this will not, on its own, resolve the nonidentifiability problem if the sampling process also needs to be modeled and may even make estimates of diversification rates worse in some cases. But perhaps we can nonetheless use paleobiological knowledge in orthogonal ways to complement and inform phylogenetic estimates of diversification rates. In a recent perspective, Liow et al. argue that researchers have learned a great deal from decades of paleobiological studies (as well from those of other disciplines) about the processes of speciation and extinction—as well as how variation in these processes is structured temporally, geographically, and phylogenetically—and that we can leverage this information to reduce the set of alternate histories we ought to consider. I find Liow et al.’s vision of a more synthetic macroevolution to be exciting and am keen to see researchers explore their ideas more rigorously.

… which means, logically, that studies of macroevolution have not been rigorous. Can Pennell and the other Darwin Party theorists really turn fossils sideways to help? (One meaning of “orthogonal” outside of mathematics is “having no bearing on the matter at hand; independent of or irrelevant to another thing or each other” according to Dictionary.com.) What he implies is that fossil evidence, molecular evidence and theory have been at cross currents. For a brief second, Pennell hoped that fossils would solve the nonidentifiability crisis, but upon reflection, he realizes it will not. Despite his hope in paleobiological studies and “knowledge” from other disciplines, none of them can help settle on a probable history of life.

The answer, he tells his friends, is blowin’ in the wind—somewhere in futureware fantasyland.

Here we are, in the year 2023 A.D., 163 years and 4 months after Darwin flung his earth-shaking “mechanism” onto the public, and you thought Science had figured out macroevolution. Ouch.