Towards A Good Samaritan World

Tuesday, September 12, 2006

DARWIN AND FALSIFIABILITY

The comments of my recent post "A Challenge" have taken on a life of their own. (What provoked the debate was my "challenge" to Darwinists to show how the theory of evolution meets the Popperian test of falsifiability, i.e. there is some conceivably observable fact which, if true, would disprove the theory; I gave a floating rock as an example of a falsifying counter-example to the theory of gravity.) The leading contributor is Nato, who blogs (occasionally) here. I'll just bring out one excerpt here, since I responded in the comment thread:

The hypothesis that all life on earth was generated by such design is not al all a priori, though, so it needs to be disconfirmable empirically. If we find one example of a creature that's designed by an intelligence, then we've disconfirmed that hypothesis, but not in an interesting way. What one would really need is an understanding of what algorithmic design can do and what it can't, then look for the latter.


Now, as far as I'm concerned, Nato has just conceded the point. "What one would need is an understanding of what algorithmic design can do and what it can't..." Exactly. That's what we would need in order to make the theory of evolution falsifiable. We don't have it yet, or at any rate it's controversial whether we do or not. (The Intelligent Design people, as far as I understand, think that they have understood what algorithmic design can do and what it can't, thus rendering Darwinism falsifiable, and they have, in fact falsified it, by showing that things we observed can't be accounted for by algorithmic design. I haven't read their case and have no opinion on whether they're right, though my priors are to be skeptical.)

Anyway, as long as we don't know what algorithmic design can and can't do, then we can't look for the latter (what it can't do) because we don't know what to look for, hence evolution is not falsifiable, and in Karl Popper's sense is not a scientific theory. I'm sure Nato doesn't agree. But I'm not sure how he can avoid agreeing. His arguments in the comment thread seem to be of two types: (a) even gravity is not really falsifiable, and (b) if we keep working at it, we might understand what a falsifying counter-example to evolution would be, someday. If true-- as long as we're accepting Popperian "demarcation" of science from non-science-- then (a) implies that gravity is not a scientific theory either, and (b) implies that evolution is not a scientific theory yet, but has the potential to become one. Either way, evolution doesn't pass the Popperian test now.

But I'll focus on Tom Reasoner's comments, since I didn't engage them in the earlier comment thread. Tom begins:

To falsify anything requires experimental evidence. So, to falsify "Darwinism" of the sort that claims that we're descended from lower life-forms would require a controlled and repeatable experiment that shows (or does not show) that 1) living organisms can occur "naturally" at all (ie without divine intervention), and that 2) living organisms can produce things/patterns with reproductive capabilities themselves that are more complex and complicated.


Tom seems to set the bar pretty high here. How can we show, by experiment, that living organisms can occur "naturally?" Of course, we could cruise around the solar system to a lot of young planets and watch to see if life emerges. But it would take millions of years and millions of light-years of travel to catch just the right moment, and the whole matter would be very hit-and-miss, particularly given the challenges of defining "life" (there might be different possible chemical bases for "life" in the sense of self-replicating structure) and the fact that at the time of its genesis it would likely be very small. So this isn't really experiment. Part (2) is hard for a different reason: how do you define "more complex and complicated?" We have an intuitive sense for what complex means, but can we define it in a rigorous way? There are, it would seem, insuperable obstacles to implementing the kind of "controlled and repeatable experiment" that Tom is describing.

But there's a more serious problem. The experiment, says Tom, "shows (or does not show)" the truth of his propositions. But to meet the Popperian test, for the experiment to be able to "not show" his propositions (on a given occasion) is not enough. It needs to be able to falsify them. Can it?

If I mix carbon, hydrogen, nitrogen and oxygen in a totally sterile test tube, leave it in the sun for a while, and lo, we find living things in there a few months later, that surely shows that "living organisms can occur naturally." But if there are not living things there, does that show that living organisms can't occur naturally? Of course not. Everyone agrees that the chances of life emerging spontaneously are infinitesimal even in the optimal conditions. So this is just another example of the "heads, I win, tails, we flip again" reasoning that is typical of Darwinists.

Tom goes on:

[I]f we found some way to observe experimentally a transistion from an organism with X amount of genetic information to an organism with X+Y amount of genetic information, then presumeably we would have data by which we could produce a theory that would either confirm or disconfirm Darwinism.


What is "genetic information?" If an organism mutates to sprout a bit of new, useless gobbledygook at the end of its genetic code, does that count as an increase in genetic information? Surely not, which underlines the point I made earlier about the difficulty of rigorously defining, as opposed to intuiting, "complexity." Moreover, my other complaint still holds: this "experiment" can't disconfirm Darwinism, only confirm it. If an organism fails to transition to greater complexity, that doesn't prove that it wouldn't sometimes happen.

That said, in Tom's defense I don't think experiment is even needed to show that organisms can become more complex. We know mutations occur. We understand how the principle of natural selection operates by mere logic. We can infer that organisms could increase in (what we would recognize intuitively as) complexity through the interaction of mutation and natural selection-- certainly, at any rate, we could infer that this is at least possible, though the probability of it occurring is harder to appraise-- using mere logic. So I think that talk of experiments is a red herring but he's nonetheless on firm ground so far.

Tom goes on:

Now, I would say that most Darwinists believe we already have such experimental evidence through fossil records and experiments with fruit flies and hereditary analyses and so on.


One reason that the fossil record doesn't qualify as "experimental" evidence is that experiments are-- as Tom said-- "controlled and repeatable," and the fossil record is neither. Perhaps that's splitting hairs. A more substantive point is that there are many gaps in the fossil record, countless "missing links," which Steven Jay Gould has tried to explain (or excuse) through his (very plausible) idea of "punctuated equilibrium;" also there are numerous extinctions, which again are actually innocuous to the theory of evolution, but nonetheless certainly "don't support" evolution in that currently extant species did not evolve from those animals at least, and then it's interesting that the most famous fossils of all, the dinosaurs, are quite non-evolution-confirming in that they went extinct (at least that seems to be the most common account); another problem is that the fossil record provides only bones, which have to be interpreted, and the interpretation assumes the truth of Darwinian evolution and uses it as the basis for inferences about what the animals really looked like...

But the real reason that none of this qualifies as experiment is the same complaint as before: none of this could falsify the Darwinian theory. Nothing in the fossil record could, for example: an organism with no apparent extant descendants is simply assumed to have gone extinct, and "missing links" in evolutionary chains simply mean that none of the "links" happened to get fossilized.

As for "hereditary analyses," they can certainly prove (a) Mendelian genetics, and maybe (b) mutation, and (c) natural selection. But, as Tom goes on to say:

Of course, none of that evidence is *sufficient* even taken altogether to prove without a doubt evolution's veracity as a whole, but each experiment does prove, in the sense that gravity is proven, bits and pieces of evolutionary theory.


Exactly. Except that "bits and pieces" of the theory is not good enough. Nothing is more common than for a theory to have many correct pieces and yet to be as a whole, false. Or even meaningless.

Honestly, it's not that hard to envision experiments that would falsify Darwinism.


I don't think this claim is true. I'm still skeptical whether an experimental to falsify Darwinism is conceptually possible. The "heads I win, tails we flip again" reasoning seems quite pervasive.

Basically, almost every single observation we make has some implication or other for Darwinist theory.


One could say, just as truly, that "basically, almost every single observation we make has some implication or other for creationism." Clouds in the sky, with their strange and ever-changing shapes, are a sign of God's infinite creativity. The light of the sun represents God's providence. A blazing sunset fills us with a sense of glory, which is God helping us to conceive of His ineffable glory. Etcetera. "Has some implication for" is not the relationship between observation and theory which we are interested in. Once you have a theory, you look for confirmation of it everywhere.

Like, I notice that whales have these strange floating bones in their blubber that sort of resemble bones used for walking in other land mamals.


Yes, and tomorrow some biologist might discover some reason that these are useful to a whale, and suggest that they are not vestigial, but evolved specially for some aquatic purpose. And that would not be a challenge to the Darwinian theory.

Or I notice that my baby boy has eyes just like his daddy, but his nose is more like his mother's. Or the cross-polenation between my orchids and hydrangeas just produced a flower I'd never seen before.


This is not a demonstration of Darwinism, but of Mendelian genetics.

All of these little observations add up into a cohesive whole.


But the whole has a great many holes in it: "missing links" in evolutionary chains, how consciousness can be accounted for in physicalist terms (on which there is certainly no consensus even if some claim to have resolved it), why human beings have so many abilities of no apparent evolutionary utility, etc., etc. The cohesive whole is what people want to see-- it's so comforting to know, instead of not knowing, so they see it-- but it cannot (I submit) stand up to a rigorous philosophical examination.

And what's the alternative to gravity and evolution (at the moment)? Divine intervention and intelligent design, otherwise known as "I don't know, it's a mystery".


But there are cases when we don't know, and in that case, we should firmly say so, and not pretend that we do. Anyway, I object to the conflation of (a) intelligent design with (b) "I don't know, it's a mystery." Intelligent design people, if I understand them, think that they do know, that they have proof that life was intelligently designed. I, on the other hand, am a member of the "I don't know, it's a mystery" school, and I do not buy into intelligent design theory. (If I believe vaguely in intelligent design of the world-- that's not my preferred framing of the question, anyway-- it's on grounds that I would not describe as scientific.)

If that sounds like a lazy cop-out though, let me offer a bit of substance. What if the Darwinian theory-- the modern synthesis of Darwin, Mendel, Crick/DNA, etc., to be more precise-- is offered as the answer to a different question? Forget the Darwinian theory of the origin of life; try the Darwinian theory of imperfect ecosystemic homeostasis.

Homeostasis is the characteristic of life-forms that they maintain a consistent shape even though their substance is changing. In this sense life-forms are like waterfalls. A waterfall is an unusual "object" in that it is forever in motion, yet stands still; the substance-- the actual molecules-- that it is composed of are changing from second to second, yet its shape remains identifiably the same. Animals and people (to a lesser extent plants) are the same way. I breathe in, I breathe out, I eat, I excrete, the carbon and hydrogen and oxygen and nitrogen atoms of which, in one sense, I consist, are constantly passing through me, coming in, going out, yet "I" remain. That's homeostasis.

Well, just as homeostasis is a property of organisms, if molecules/atoms are taken to be the underlying substance, homeostasis is a property of ecosystems, if organisms are taken to be the underlying substance. The organisms which comprise an ecosystem are themselves mortal, they are born, give birth, age, die, yet the ecosystem as a whole maintains a certain constancy despite the perpetual change that characterizes its constituent parts. This is made even more mysterious because the mechanics of reproduction in the underlying organisms do not ensure that successive generations of organisms remain the same; on the contrary, there is "descent with modification" and "natural selection," both of which have an obvious potential to cause organisms to change across generations.

How this happens is one aspect of biology, not especially Darwinian; where the Darwinian theory comes into its own is explaining why homeostasis occurs, not just in this ecosystem or that ecosystem, but in ecosystems generally. The Darwinian theory explains why species are adapted to their environments; and why, if they were not, they soon would be. It would be very improbable for us to observe a non-homeostatic ecosystem, since that ecosystem would exist only briefly and would move towards some homeostatic equilibrium.

And yet here's where the theory becomes interesting: there is no reason to think that ecosystemic homeostasis is perfect. It is difficult to say, when observing any particular ecosystem for a short period of time, whether that ecosystem is homeostatic or not, just as it is difficult to tell whether a person's character is changing by observing him for two weeks. (Probably his mood is changing, is all.) But the tensions inherent in an ecosystem may lead to a steady honing, and perhaps even occasional dramatic changes. Moreover-- here it gets really exciting-- the tensions within ecosystemic homeostasis suggest a mechanism by which, over long spans of time, dramatic changes in species and ecosystems could occur, in response to climactic changes or even without them, and this may even account for why life on Earth in past times seems to have been dramatically different than it is now.

So what's the difference between the Darwinian theory of the origin of life, and the Darwinian theory of imperfect ecosystemic homeostasis? The latter cuts loose the non-scientific claim that Darwinian is an adequate or complete account of all life on earth. The latter, that is, doesn't claim to know more than we actually know. It leaves the uncharted corners of the map blank, instead of writing "Here there be dragons."

A final subplot. Max Borders, the managing editor of Tech Central Station, and author of this moving article about 9/11 among many other things, as well as a person with formidable philosophical credentials, questions the Popperian premise of the discussion:

Why should the criterion of falsifiability be the test of science? Popper was taken to task on this. It's useful, but not absolute.

Popper also referred to Darwinism as a "metaphysical research program" versus a scientific theory. Why not?


Though Nato comes to Popper's defense, perhaps Max has offered a way out for evolutionists? If we reject Popper's epistemology, we can reject a rejection of Darwinism as scientific on Popperian grounds. But in that case, we need some other way to do what Popper was trying to do when he proposed his falsifiability test, namely to distinguish between science and non-science/pseudo-science.

2 Comments:

  • I'm not going to respond to your whole post, but I'd just like to point out a sort of post-modernist corner that you've painted yourself into that undergirds your entire argument.

    You wrote: "If I mix carbon, hydrogen, nitrogen and oxygen in a totally sterile test tube, leave it in the sun for a while, and lo, we find living things in there a few months later, that surely shows that "living organisms can occur naturally." But if there are not living things there, does that show that living organisms can't occur naturally? Of course not."

    Using your gravity example, if I drop a rock and it does obey our current understanding of gravity, does that mean that it can't ever *not* obey our understanding of gravity? How do we falsify that claim? Quite obviously, the emphasis on experimentation is on something that is *repeatable*. If we do a particular experiment a million times, we should expect the outcome to asymptotically approach a certain result, which hopefully represents a truth that we can use to create theories around (or use to refute other theories). If life couldn't occur naturally, would we ever know for sure? We could only be as sure as the number and variety of experiments performed concerning that issue, the same as with any other piece of "scientific knowledge". You've quoted Hume's refutation of inductive reasoning before, and this is just the same thing in a different context. By being skeptical of inductive reasoning, you call into question any and all scientific claims, because they inherently rely on induction. In fact, I might go so far as to say that science is tautologically induction. So, you're right: if we can't use induction, then Darwinism isn't falsifiable; it wouldn't be confirmable or disconfirmable either way in this case.

    Of course, induction isn't always right, but most of the time, as with the theory of gravity, it at least approaches being right. Science isn't like philosophy. It doesn't try to be absolutely right 100% of the time. Science is about getting closer and closer to the truth. It's more statistical and probabilistic. As long as a theory is in the ballpark, it's good enough, usually. And if it ever becomes not good enough, then it's refined to where it is again. And Newton's theory of gravity is a perfect example of that fact. Now, it's possible for a scientific theory to be 100% correct, but it still requires induction to support it, so that we're sure it's correct. And if it's wrong, induction will show that too. Induction is the only way you can falsify anything.

    By Blogger Thomas Reasoner, at 1:44 AM  

  • I feel a little boneheaded not remembering this really obvious disconfirmation case earlier - I think it's so obvious it's hard to remember. If we found a fossil (or better yet, a set of fossils, to avoid the anomaly issue) that was laid down in an age incontrovertably too early for its lineage. Here I'm thinking of something like cetacean fossils in rock dated solidly prior to the evolution of vertibrates. That would either disprove evolutionary history or prove time travel (anomaly?), anyway.

    As for an understanding of what algorithmic design can and can't do, well, we can put certain limits, like designs can't transport themselves back in time, and there are certain obvious limits in how similar convergent features will be. The issue that I intended to address was that if one could prove in an a priori (mathematical) way that certain designs can only reasonably be reached through foresightful engineering, then one can immediately trash whole ranges of supposed evolutionary products instead of finding one that might have been specially created amongst a crowd of natural species. We, after all, can edit organisms through genetic engineering (including adding a copyright message to DNA), but we can't claim more than a tiny bit of the design work for ourselves.

    A few other bits:
    A) My position is that gravity is falsifiable, but single-experiment falsifications aren't as easy to come up with as you seem to suppose.
    B) That we can come up with some promising falsification cases (now that we've thought about it a little more) for evolution is a bit of luck, really. I still can't think of any really good ones for gravity. Probably physicists could.
    C)It would seem that if there weren't holes in evolutionary theory, then no one would be researching any more.
    D)Punctuated equilibrium is either a fairly obvious result from population genetics implying that rate of evolution in populations will vary wildly by population size and selection pressure or a weird, apparently unsupportable hypothesis of saltational changes at almost the individual level. The latter view is retarded, if one is a physicalist, but that doesn't keep some folks from wanting to see things that way. I think folks want to impute "evolution" with mysterious properties. It's like the folks who think evolution has "momentum" and the human trajectory from smart ape to full sentience must therefor lead soon to supersmart super sentience or something. It's painful.

    By Blogger Nato, at 2:20 AM  

Post a Comment

<< Home