Saturday, August 6, 2011

Fallacy of the Best-in-Field Fallacy

~ In which the blogger gives with one hand (for most of the posting) and takes with other.

I first heard of the best-in-field fallacy in a talk by William DiMichele, the Smithsonian’s curator of fossil plants.  It piqued my interest because at first blush it seems so powerful (and damning if you can claim an opponent has fallen victim to it).  DiMichele has written about it as well and I’ll use one of his articles to explain what the fallacy is all about.

Analyzing the interplay between climate change and changes in flora during two intervals in the Paleozoic, DiMichele warns against what he calls the “best-of-field” (sic) fallacy (DiMichele, et al., Climate and Vegetational Regime Shifts in the Late Paleozoic Ice Age Earth, Geobiology, Volume 7, 2009, p. 200-226).  (As explored below, the person who apparently first identified this fallacy named it “best-in-field” fallacy.)  In this paper, DiMichele cautions that looking well back in time at vegetational responses to climate change may take us where our modern analogues fail to capture the causes and effects that were then in play.  That’s where the fallacy kicks in.  He writes,
In historical science, our experience and imagination constrains what we choose as explanations for the patterns we see.  In what was described as the “best of field fallacy” (Macbeth, 1971), we choose between explanations ‘a’, ‘b’, and ‘c’, when, in fact, the correct answer is ‘d’. (p. 202)
But in casting about for that possibly elusive (and possibly true) ‘d’ choice, we do not put everything on the table.  For instance, in the paleobiology context of his research, DiMichele asserts, “We must . . . assume that the underlying physical principles do not change.”

He cites Macbeth, 1971, as his source for the identification of the fallacy.  It’s an interesting source.  Norman Macbeth, a lawyer, published a slim volume in 1971 titled Darwin Retried:  An Appeal to Reason, in which he approached his study of the Darwinian theory of evolution as a lawyer would and concluded that it had collapsed under the weight of hopeless contradictions and impossibilities.  Though he acknowledged the fact of evolution, he found no evidence supporting the ability of natural selection to answer the how and why questions of evolution.

It’s fairly readable tract but one that raised my blood pressure and prompted hasty scribbling of marginalia (ranging from “what a crock!” to whole paragraphs sprawling into wherever there was white space).  My primary task in this posting is to show how Macbeth presents the best-in-field fallacy which is a key part of his attack on Darwinian evolution; but then, in an effort to reduce my hypertension, I indulge in a bit of an attack of my own.

The context for his discussion of the “best-in-field” fallacy is what he labels adaptation.  Macbeth first disparagingly describes the Darwinians’ efforts to explain how the accumulation of many small changes accomplishes the creation of complex organs as the “wave-the-wand method” of scientific exposition.  He then quotes Darwin on the centrality to his theory of the effects of this accumulation of small changes:
If it could be demonstrated that any complex organ existed, which could not possibly have been formed by numerous, successive, slight modifications, my theory would absolutely break down.  (On the Origin of Species)
At this, Macbeth declares
Since this fact [the fatal flaw identified by Darwin] seems to have been demonstrated, if only by default, the reader will ask whether the modern Darwinians concede that the theory has broken down.  The answer is a strange one – they are not greatly troubled by their failure to explain the adaptation because they are sustained and soothed by the best-in-field fallacy.  (p. 76-77)
The Darwinians, according to Macbeth, fall to the fallacy when they compare their theory to the competing ones that been proffered.
The Darwinians have shown that none of these [other] theories are any good. . . .  Thus the Darwinians are able to say that Darwin made a better try than anyone else, and they find real comfort in this. . . . [But,] [i]s there any glory in outrunning a cripple in a foot race?  Being best-in-field means nothing if the field is made up of fumblers.  (p. 77)
In other words, ‘c’ may be the best answer we have among the choices before us of ‘a’, ‘b’, and ‘c’, but that doesn’t make it correct, or, indeed, much better than ‘a’ or ‘b’.

Macbeth lays claim to being the first to identify this fallacy and, as is his usual approach in this book, digs at evolutionary scientists with a bit of sarcasm.
The best-in-field fallacy seems to be my own discovery.  It does not appear in books on fallacies and I have not seen it clearly expressed anywhere else.  Perhaps it appears with unusual frequency among the evolutionary theorists, who seem to have a special weakness for it.  (p. 78)
Macbeth irritates me for a variety of reasons, trivial and significant.  He’s particularly annoying when he labels natural selection a tautology which he is able to do by defining it as an outcome – “differential reproduction.”  As a result, he posits that these scientists enter into the following circular reasoning:
. . . Question:  Why do some [species] multiply, while others remain stable, dwindle, or die out?  To which is offered as Answer:  Because some multiply, while others remain stable, dwindle, or die out.  (p. 47)
The “Answer” Macbeth attributes to evolutionary scientists is not the one they would be likely to give because, as I understand it, natural selection is a process not an outcome.  The “differential reproduction” of which Macbeth speaks is the product of the forces of nature acting on variations among organisms within populations (not species), favoring the survival (and reproduction) of some but not of others, thereby propagating certain variations.  Darwin, in the first edition of On the Origin of Species, wrote, “This preservation of favourable variations and the rejection of injurious variations, I call Natural Selection.”

As a result, the “Answer” more properly, more logically, and more likely to be given is that some populations are more likely to survive because they are better equipped for the circumstances in which they find themselves; those less well equipped may be less likely to survive.  In my mind, that answer to the question hardly constitutes a tautology and isn’t meaningless.

Among the other aspects of Macbeth’s book that annoy me are the following:
1) His cleverness in portraying himself as a disinterested lawyer trying Darwinian evolution in a “court of law.”
2) His related claim that, as a critic, he’s under no obligation to offer alternative explanations, just poke holes in the arguments and evidence for evolution.
 3) His marshaling of highly selected quotations from an idiosyncratic group of scientists as well as authors of popular pieces to “prove” terminal uncertainty and contradiction among them regarding evolution, thereby damning the theory.
4) His ploy of reproducing “imaginary conversations” among scientists to prove his points (hardly acceptable in a court of law).
5) His serious misconstruing of the fossil record of his time (and for clearly not being around to try to argue away the past three productive decades of paleontological work).
6) His sleight of hand in leading the reader to believe that it has been proven that there are complex organs that could not have been created through the accumulation of small changes over time.
(On this last point, where and when was this proven?  And, what does Macbeth mean when he writes, “Since this fact seems to have been demonstrated, if only by default”?  I interpret this comment to suggest that he believes Darwinians have the obligation to prove that each complex organ could be derived from the process Darwin describes.  Macbeth should have read the passage from Darwin which he quotes (see above) more closely.  Darwin puts the onus on his critics to find a complex organ that could not have been produced by the agent of natural selection over long periods of time.)

In comments on the book, Stephen Jay Gould proved somewhat more complimentary (Macbeth “raised some disturbing points”) but believed the book was fundamentally flawed because Macbeth came to it applying an “inappropriate” standard:  “the defendant (an opponent of evolution) is accused by the scientific establishment and must be acquitted if the faintest shadow of doubt can be raised against Darwinism. (As science is not a discipline that claims to establish certainty, all its conclusions would fall by this inappropriate procedure.)”  (Impeaching a Self-Appointed Judge, as found in The Unofficial Stephen Jay Gould Archive)

I am left after all of this with a number of questions, but the overarching one is whether the best-in-field fallacy really has any substance as a fallacy.  Two elements are critical in my judgment.

Deliberate manipulations of the options.  On a somewhat trivial level, one potentially valuable aspect of recognizing such a fallacy is that it alerts us to the possibility that what is presented to us as our choices is a biased sample and there are other known or conceivable options, perhaps better ones, which aren’t offered to us.  Nearly any debate in the current political environment in this country offers an example of this skewing of options.  When a politician says, “We have two choices . . . ,” my reaction is to look for the third or fourth choice.

Best of the available options.  In science, which is the venue of this posting, of what validity is the best-in-field fallacy?  DiMichele considers it real and important.  But, frankly, the more I’ve thought about it, the more insignificant I think it is, particularly within science.  Indeed, DiMichele in his article has a sentence immediately following his explanation of the best-in-field fallacy (see quotation above) that seems to me to prove this point.  He writes,
In the best-of-field context, we should expect that our causal explanations of patterns will be, potentially, incorrect, and subject to improvement as new data or new ideas and insights intrude into the explanatory framework.  (p. 202)
If I understand DiMichele, that's what we need to acknowledge to avoid the best-in-field fallacy.  Wait . . . isn’t that what scientists do and what science is about – the testing and retesting of hypotheses, the discovering of new findings that refute or buttress positions, the changing of minds and theories?  So, does the best-in-field fallacy occur simply because the critic fails to understand that science works this way or doesn't find scientists making that clear at every juncture?  To be a bit more generous, perhaps it is a matter of scientists using some restraint and not implying that a prevailing theory is the best there can or ever will be, rather that it's the best of the options we can conceive of at this moment.  Not much of a fallacy in my opinion.


  1. I think "best-in-field" is closely related to uniformitarianism. As such it plays a role in constraining scientists, and that's not necessarily a bad thing. I often quote Carl Sagan to my students: "Extraordinary claims require extraordinary evidence." But, as you say in your last paragraph, the only way to improve the field is to increase the data.

    Repurposing the race analogy, imagine a 400 m race. Before the starting gun sounds, all participants are equally likely to win (there's no data). When the race starts, after 20 meters (limited data) one runner has already fallen way behind (the crackpot/pseudoscience ideas). Perhaps after 100 meters, it's clear half our runners are not going to win (reasonable ideas, but demonstrably wrong), and by 300 meters there's a definite frontrunner.

    Now maybe at this point someone drives a race car onto the track, and they pass the leader with 50 meters to go. We'll then change our favored prediction and instead say the race car is going to win, but only because we have the new data that the car is present and leading.

    This is essentially the "best-in-field" argument; that if we constrain ourselves to the runners, we might not predict the race car. But science plays the odds, and assumes that it's very unlikely that the race car will show up. If you're watching the race and predict that the race car will win, when there's no race car on the track, your suggestion would be laughable.

    Note also that, as soon as the race car appears, we would almost always pick it as the winner. I described the race car joining the race when the front runner was 50 meters from the finish, but suppose instead the race car started at the same time as everyone else. We would still pick the race car as the winner; we just have to have the data at hand (the presence of the race car).

    Finally, to completely beat the analogy to death, imagine one more scenario. The race car only starts the race when the front runner is 1 meter from the finish line. In this case our extraordinary data (the presence of the race car) is found, but it doesn't affect the outcome; the frontrunner wins anyway, because he's supported by mountains of data (a 399 meter lead).

    This would be like piecing together tons of data that showed that the Chicxulub impact occurred, but finding out that it happened 10 million years after the dinosaurs went extinct (an extraordinary event that has no bearing on the question at hand).

    Or tying it to evolution, even if somehow someone could prove the existence of God, it would still not disprove evolution; it would simply prove that there was a God.

  2. Anton:
    A thoughtful and thought-provoking comment. I considered some wrinkles to the racing analogy (e.g., a starting line partly shrouded in fog), but the critical point, which you make, is that, in this race, we must acknowledge that don't have complete knowledge about the field. We take our best shot at picking the winner with that understanding.

    Perhaps some scientists are guilty of writing about evolution with a touch of hubris, as though all is known, all questions are answered. Macbeth may well have been reacting to that and, because scientific theories are always being tested, he could cherry pick a bunch of criticisms of evolution. But I don't think he identified a logical fallacy damning the whole enterprise. It's science at work, knowing it's using the best data available at the moment.

    I like your Sagan quotation and would respond that I think with regard to evolution and natural selection we do have extraordinary evidence.

    Finally, since this particular race is being run under the rules of science, one thing we do know is that no supernatural runner will materialize in this field. As you note, that's a different race entirely.


Nature Blog Network