In a lengthy comment to my post on probability and evolution, I pointed out that for scientists engaged in biological research, natural selection is not an abstract principle. It is not something that is invoked casually as a catch-all explanation for whatever complex biological system crosses their path. Rather, it is a tool that they use for generating testable hypotheses about the origin and history of whatever they are studying.
This point is highly significant, since anti-evolutionists tend to argue at a very abstract level. This is especially prominent in the work of William Dembski, who tries to use abstract mathematical arguments to argue in principle against the ability of natural selection to craft complexity. It is trivial to poke holes in his arguments on the merits. But even without that there would be reason to be skeptical, for the reason I have given in several recent posts. When the physical evidence strongly implies that something happened, but an abstract mathematical model says it cannot happen, there is probably something wrong with the model.
After posting my comment, I wandered over to the blog of the National Center for Science Education. You can imagine my surprise when I noticed that Glenn Branch had been writing about similar issues. In this post he discusses Dembski's idiosyncratic take on the subject of falsifiability. It is a continuation of this earlier post on the same subject. The part that caught my eye was this:
Scientists who actually are engaged in investigating the evolutionary history of the bacterial flagellum aren’t interested in what processes could or could not produce the system; they are interested in what processes in fact produced the system. And their hypotheses about those processes are falsifiable; for example, when Mark Pallen and Nicholas J. Matzke suggested in a 2006 paper in Nature Reviews Microbiology that the ur-flagellum arose from mergers between several modular subsystems, they were going out on a limb, empirically: their hypothesis would be falsified by, e.g., the discovery that the flagellum predated those subsystems.
Bingo! That is precisely the point I was making. Branch goes on to write:
The only argument, if it can be called that, on offer in Dembski’s books seems to be based on the observation that despite the fact that there is not presently a generally accepted and thoroughly detailed explanation in “Darwinian” terms of the origin of the bacterial flagellum, scientists are confident that such a explanation is in principle possible. (In principle, even if it is ultimately unavailable to us, e.g., if the evidence necessary for distinguishing between alternative scenarios is no longer available due to the passage of time—a common problem in the historical sciences). The observation is reasonable; the conclusion that Dembski bases on it is not. For Dembski regards such confidence as manifesting nothing more than a dogmatic commitment, come what may, to the unfalsifiable doctrine of “Darwinism”: “Darwinism is wonderfully adept at rationalizing its failures and therefore just keeps chugging along.”
But, again, scientists who actually are engaged in investigating the evolutionary history of the bacterial flagellum aren't interested in “Darwinism” as Dembski and Behe define it; they're interested in framing and testing hypotheses about what processes in fact produced the bacterial flagellum. Hypotheses that appeal to natural processes are, by and large, testable, so it's not surprising that those scientists are inclined to concentrate on them. If there were alternatives that were testable, then they would not be necessarily disregarded; there’s no dogma in place. Rather, the proponents of “intelligent design” have consistently failed to present any testable hypotheses of their own, instead hoping to batten on the supposed failures of evolution.
The Pallen/Matzke paper referred to by Branch is just one of a very long list of possible examples. George C. Williams discusses others in his book Plan and Purpose in Nature, as does Ken Miller in Finding Darwin's God. Several of the contributions to the Edis/Young anthology Why Intelligent Design Fails provide still further instances.
The bottom line is that scientists are not going to abandon a highly successful research program just because some creationist waves his hands and slings some jargon. The critics will have to do better.
"scientists are confident that such a explanation is in principle possible"
If I understood correctly the problem is that there are several hypotheses but not enough empirical data to decide which one is correct (ie to falsify the others). Compare with superconductivity at relatively high temperature. No single hypothesis seems to describe it correctly. Still I don't know any creationist who argues for divine intervention here. That's inconsistent.
"the proponents of “intelligent design” have consistently failed to present any testable hypotheses of their own."
That's why Adam Lee asked his two questions. It's totally unsurprising that Sean P neglects them.
It is a somewhat interesting semantic question whether "Darwinism" could be or has been falsified. Some will say that Charles Darwin got a heck of a lot right, enough that it makes complete sense to refer to modern evolutionary models as "Darwinian". Others may argue that he made key mistakes or simply had an incomplete model (one that lacked, eg, genetics) and so we are no longer "Darwinian". Still others could argue that it's a mistake in science to use someone's name like that.
Similar but distinct issues arise on the question of whether our universe is "Newtonian". Answer that with "yes" or with "no", and either way you seem to mss something important (Newton's laws are in one sense "wrong", but they still hold as excellent human-scale approximations of relativity).
But ID tends to define "Darwinism" as broadly including any naturalistic account of biological origins. Whether that could be scientifically falsified requires a better definition of "non-naturalistic". So while they might naturally assume that determining its falsifiability is the job of "methodological naturalists", the ball is really in their court; they're the ones trying (and I think, failing) to make scientific sense of "non-natural" explanations.
When the biochemist J. B. S. Haldane was asked by an overeager Popperian what it would take to falsify evolution, he replied, "a rabbit in the pre-Cambrian strata". Needless to say, Dumbski & Co. have yet to find the pre-Cambrian rabbit.
Re. Lenoxus @ #2: "..._any_ naturalistic explanation..." But any theistic explanation is by definition "supernatural" (above or outside of nature), thus unfalsifiable. The attempt to "make scientific sense of" supernatural explanations is something like a non-sequitur, along the lines of trying to describe a sculpture as if it was a symphony.
I'm inclined to believe that creationism and ID are, at root, attempts to convince people that a deity exists, by using assertions about nature rather than by using scriptural authority. In effect these things are forms of evangelism.
Question: which term do the majority of working scientists generally favor: falsifiable, testable, or decidable? And/or are there subtle differences in the ways they use those words?
Here's how I use them; feel free to tear this to shreds and point out where the mistakes are:
I tend to use "falsifiable" when speaking with people who have the background to understand it without further explanation. (Empirical hypotheses are not "proven" ("proof" = logical proof e.g. in math), but "supported" or "falsified" by empirical facts (measurements).)
I tend to use "testable" when speaking with people who don't have the background to have a complete grasp of "falsifiability."
And I tend to use "decidable" when referring to propositions that are not usually thought of as hypotheses, for example, "the existence or nonexistence of a deity is empirically undecidable."
I understand falsifiability has been used in court to distinguish science from non-science. Therefore, it can and has been used in the ID is a science vs ID is a religion debate. Personally, I have some problem with this. Mathematics is considered a science. But pure math develops from postulates and forms conclusions based on the methods (also assumed) employed. Therefore, a math conclusion is not falsifiable. For example, the general theory of relativity is not falsifiable. Here we must distinguish between the math and the physics. The Big Bang model is falsifiable and has been falsified by observation. The Big Bang predominates today because it has predicted some observations (one criteria - prediction) and it is most descriptive of a wide range of observations (another criteria - usefulness). Competing physics models has also used general relativity. String theory is another example. It uses a math model to describe particles. But the particle must be found first. Hence, string theory has not yet predicted but can describe.
After this I’m sure another debate may follow.
Science in general has several models, each of which can pass a falsifiability test and has been falsified. After all, our science as of today cannot create a universe. So it becomes a matter of which to use in a given situation.
The scientists will abandon their research if ID is successful in swaying public opinion and influencing federal spending. This is why we must oppose ID in the public forum. This has been done in the past. For example, the Catholic Church punished Galileo and effectively stopped scientific research in catholic countries. Science move north, then west to the US. ID is not tolerant. Science requires tolerance.
Is lack of tolerance a talking point against ID?
I neglected to mention John Alcock's book The Triumph of Sociobiology, which also discusses these issues. In particular, Alcock is not too impressed with the “spandrels” paper of Gould and Lewontin.
Lenoxus #2: Darwin "simply had an incomplete model (one that lacked, eg, genetics)."
In fact, Darwin stated that cells had to contain a mechanism for passing on traits, with high accuracy, and with a mechanism for introducing small amounts of variation. In essence he predicted that genes exist. He didn't predict that genes would be made of DNA. Nonetheless, the discovery of DNA capable of carrying genetic information and passing it on to progeny with high accuracy and a non-zero but small rate of variation is a striking confirmation of Darwin's prediction.
I would not immediately accept a rabbit in a PreCambrian stratum as falsifying evolution, but rather I would think it strong evidence for the existence of time travel. ;-)
The finding of pre-existing parts that could be put together to make a more complex system says nothing as far as the minimum number of modifications required to get these parts to work together in a new way as part of a more complex machine.
This is the fundamental problem with Matzke's papers on flagellar evolution in particular. His proposed steppingstones, based on pre-existing systems, are simply too far apart in sequence space for random mutations to get from one to the other in a reasonable amount of time.
You see, these required modifications cannot be guided by natural selection because these required modifications are not sequentially selectable in a positive manner. And, we are talking about dozens of them, which creates an effectively uncrossable gap distance within sequence space.
That is why, as predicted by empirically-based models of sequence space, evolution stalls out at very low levels of functional complexity were the steppingstones are actually much closer together. At the level of something like a flagellar motility system, which requires several thousand specifically arranged residues at minimum, evolution simply stalls out this side of trillions of years of time (because the nearest steppingstones to what already exist are simply far far too far away).
And, this hypothesis is falsifiable. All that Matzke would have to do to neatly falsify my hypothesis (and confirm his own hypothesis) is to demonstrate, in the lab, evolution from any one of his proposed steppingstones in the pathway of flagellar evolution, to any other steppingstone in that pathway. The problem of course, is that this has never been done, and, according to the true nature of sequence space, is very unlikely to ever be demonstrated this side of an effective eternity of time (regardless of the size of the population of bacteria under observation).
Flagellum evolution has been observed in the lab:
On the evolution of multicellular complexity from unicellular:
Now let's see if Sean P is intellectually honest and sticks to his word:
"to neatly falsify my hypothesis"
My bet is though that he is either going to split straws or just to neglect Blaine's links. But I'll be the first to praise him if he will throw ID in the dustbin. He won't even be the first:
Initially CD accepted ID in 2005, but when he realized it had no future he rejected it.
Is there a translation into English of his debate with ID?
Flagellum evolution has been observed in the lab:
This isn't an example of evolving the genes for a flagellum. All that's happening here is that the bacteria start coding for multiple flagellar motors instead of just one. In other words, it is just the production of more of the same time, not the evolution of something qualitatively new.
On the evolution of multicellular complexity from unicellular:
This was the result of pre-programmed genetic potential. Nothing genetically new evolved here compared to the original ancestral population. In fact, no genetic sequencing was done at all on these "evolved" populations. And, it is for this reason that these traits were expressed so rapidly every single time the experiment was run (within 60 days!). It should be very very obvious that this isn't really evolution of anything new within the gene pool at all that wasn't already there within the ancestral gene pool.
Consider further that most fungi are multicellular and Saccharomyces cerevisiae (budding yeast) almost certainly devolved from ancestors that could form hyphae. In fact, wild-type diploid strains or Saccharomyces cerevisiae will form multicellular filaments (pseudohypha) in response to starvation for nitrogen (Liu et al., 1996).
Many of the common lab strains have simply lost the original ability to form multicellular pseudohyphae because they carry a nonsense mutation in the FLO8 gene (Liu et al., 1996). Presumably, those strains have been selected by bakers and brewers over the past several thousand years. This is, however, a genetic loss of information, not a gain of information. It's very similar to cavefish that have lost the ability to grow eyes due to small mutations that disrupt the eye genes, but still maintain the genes for eyes largely intact.
Even ardent evolutionists, like biochemist Lawrence A. Moran, agree with me on this one. Moran recently wrote of this experiment on his blog:
"I don't this this is quite fair since the yeast strain is just reverting to a primitive condition. This might only have required one or a few mutations. It's not a very good model for de novo evolution of multicellarity."
In short, if its not convincing to those on your own side of this issue, why would you think it's a good argument to use against my position?
It is a somewhat interesting semantic question whether “Darwinism” could be or has been falsified.
In my mind, if we found that natural selection did not play a significant role in determining the course of evolution, that would be a refutation of Darwinism. Not of evolution, but of the Darwinian concept of evolution. For example, let's say science definitively concludes that genetic drift is the overwhelminly dominant force/mechanism that determines the distribution of allelles we see in most life forms over most generations. That would (IMO) be a pretty strong refutation of one of Darwin's core ideas.
Sean Pittman @12:
[Regarding flagella] This isn’t an example of evolving the genes for a flagellum. All that’s happening here is that the bacteria start coding for multiple flagellar motors instead of just one. In other words, it is just the production of more of the same time, not the evolution of something qualitatively new.
Just to be clear, the paper reports that there were multiple, different, a single mutations in the ancestral genome that resulted in different populations of these critters producing multiple flagella whips instead of one. This is important because later, you say...
[About a different paper]This was the result of pre-programmed genetic potential. Nothing genetically new evolved here compared to the original ancestral population.
But something genetically new DID evolve in the first case of the flagella. The original ancester population did not have the mutation(s) that allow for multiple flagella; the new populations did have them. So the first paper provides an exact example of a process you claim the second paper does not show.
Let's get specific here. Figure 2A shows the ancestral FLeN residue 200* to be: DRFLDVALQY. One of the new critters with the multiple flagella has, instead: DRLLDVALQY
Please describe to me how you know that change is due to some "pre-programmed genetic potential," because it looks to me like it's a great example of the sort of event you seem to be arguing against: the production of genetically new information.
*I chose this one simply because it is difficult for me to read the text in the image, and this line was clearest to me.
[About a third paper]This is, however, a genetic loss of information, not a gain of information.
Please tell me how you calculated that it was a loss and not a gain of genetic information. What was the genetic sequence in both cases, how did you calculate information content of each sequence, and what was the result. You've already told us your conclusion, so clearly you must have gone through some scientific process to make that conclusion, yes?
Heck, if you don't want to go through that big example, you can use the very simple example I gave above: an F to L amino acid change in the 220 residue. There are actually several different point mutations that could do that. TTC -> CTC or TTT->TTG are merely two. Please tell me how you calculate the information content for the 220 residue and how you determine whether the above point mutations comprise an informational loss, gain, or "push." (And before you say push, keep in mind that it lead to more than one flagella being produced where only one was produced before. If this can result from a push, then what you're admitting is that a lot of novel developmental features can arise under evolution even when no additional genetic information is added)