Dembski and Felsenstein: Part II

This is part 2 of my series (first part here) on William Dembski’s talk given to the Computations in Science Seminar at the University of Chicago. In this part I want to look at Joe Felsentein’s comments about this video. Below I follow my usual practice of quote followed by my comment.

Felsenstein: Did he (Dembski) present new evidence? There was no new evidence…. the arguments are not new.

My Comment:  True, but at this stage I don’t think Dembski should present anything new as it seems to me quite hard work just getting people to take on board the significance of what Dembski is saying. In fact as far as I could tell Dembski was simply doing his best to convey the gist of his one main thesis; namely, that for evolution to work, the fitness surface needs to have a very particular and rare form; that is, fitness space must be front loaded with the right information. In this sense evolution is a kind of linear time “decompression” operation revealing information already present.

Felsenstein: They then argue that choosing a smooth enough fitness surface out of all possible ways of associating the fitnesses with the genotypes requires a Designer.

But I argue that the ordinary laws of physics actually imply a surface a lot smoother than a random map of sequences to fitnesses. In particular if gene expression is separated in time and space, the genes are much less likely to interact strongly, and the fitness surface will be much smoother than the “white noise” surface.

Dembski and Marks implicitly acknowledge, though perhaps just for the sake of argument, that natural selection can create adaptation. Their argument does not require design to occur once the fitness surface is chosen. It is thus a Theistic Evolution argument rather than one that argues for Design Intervention.

My Comment: My reading of this situation is that both Dembski and Felsenstein would agree that it is the nature of the fitness landscape (or “surface”) which determines whether or not evolution can happen. In fact Felsenstein says that for a randomly scrambled “white noise” fitness landscape the evolutionary process will frequently get stuck. True! But Felsenstein believes that the right kind of fitness surface is implicit in the laws of physics; he might be right, but I’m not sufficiently familiar with Dembski’s writings to know whether or not Dembski also thinks that these laws imply an evolution friendly fitness surface. Notice, however, Felsentein’s implicit deism: For him “Theistic Evolution” means that once the surface is set up the designer can withdraw and no longer needs to “intervene” in the natural process. As far as his theological concepts are concerned Felsentein appears to be a dualist in as much as he demarks a clear distinction between “Design Intervention” and the workings of the natural order once the design has been laid down (unlike myself who sees cosmic activity as the very process of design taking place).

Felsenstein: Dembski and Marks’s argument involves defining a new form of information, showing that it is conserved. Evolution can succeed only if this information is already present, so therefore evolution does not bring about new information. In Dembski’s case he goes on from that to make a theological argument (in his recent book), which I gather is basically “In the Beginning is the Information”.

My Comment: Obviously, I can’t speak for Dembski, but Felsenstein appears to be describing the kind of theology which motivates believers to find a logical hiatus and then declare that “This is the bit God did”. It is a theology that is very difficult for Westerners to think round and it is part of the mind set of both theists and atheists alike. But Dembski’s mathematical conclusion, a conclusion that is based on the standard classical physics understanding of evolution, remains true: From a classical evolution perspective the universe has a mysterious burden of information and this information is found in the implicit fitness surface.

Felsenstein: Dembski and Marks have a simple model with genotypes and fitnesses. Of course it is overly simple, but all models are. It is worth examining, because if evolution is in trouble in such a model, we need to know why.

My Comment: I’m not sure whether or not Demsbki and Marks are saying that their work implies evolution has a problem or is in trouble. In my reading of Dembski all he seems to be saying is that the fitness surface required for evolution to work is a very rare object in the huge space of mathematically possible fitness surfaces. Given its rarity and assuming equal a-priori probabilities it follows that the required fitness surface has a huge amount of information. The simplicity of Dembski’s mathematical model is not necessarily a limitation of his model but an intended feature, in fact a sign of its great generality rather than limitations:  Very general mathematical models of this kind intentionally abstract out large amounts of detail that are immaterial to the final mathematical conclusion. The sort of mathematics Dembski is dealing with stipulates very general conditions that any classical computation is logically bound by regardless of many otherwise irrelevant details. In effect Dembski is giving us the mathematical bounds or performance envelop that any classical computation is logically constrained to work within.

Felsenstein: What a mutation does: What if, instead of changing one base, we took the drastic step of mutating all of the bases in the genotype at the same time? If the Bernoulli Principle applied, we would get to a genotype whose fitness was also chosen at random. So in that case, on average, that would be no better and no worse than changing just one base. In other words, when fitnesses are randomly assigned to genotypes making a single typographical error is exactly as bad as changing every letter in the text.

Real biology doesn't work anything like that. Making one mutation in one of my genes will on average make it worse, though sometimes not. If it produces a protein, a single amino acid change often leaves the protein still functioning. But making changes in every site of its DNA is the same as replacing every protein by a random string of amino acids. Which will be a complete disaster.

Similarly, in statements in English, one typographical error might change “to be or not to be that is the question” into “to be or not to de that is the question”. Changing all letters would give something like “bdglvwujzib lxmoxg rjdg a ohlowugrbl owj”. It should be obvious that the latter is far less functional. The comprehensibility of English sentences is more like the actual fitness of organisms, and not like the fitness of the organisms Dembski and Marks imagine.

My Comment: The foregoing is part of the thesis that Felsenstein develops: His thesis is that in real biology the fitness landscape is smooth – that is, small incremental changes don’t drastically change the fitness of an organism. Felsenstein puts this down to the laws of physics which are based on localized interactions implying continuity of change rather than sudden jumps; this in turn leads to smoothness of the fitness surface and this smoothness is one of the conditions needed for a surface that is transversable by classical evolution. In this respect I believe Felsenstein is right – in fact, I believe he is as right about this as Dembski is about his conclusions. So what are Felsenstein and Dembski disagreeing about?

As I said in part I Dembski and Felsenstein are talking past one another. In spite of Felsentstein’s (valid) inference about the nature of fitness space Dembski's thesis still holds good because it is a general theorem about the mathematical envelope of total possibility, an envelope which encloses real biological space as one of the possibilities but says very little about the specifics of that space. The mathematical fact is that smooth fitness spaces are extremely rare beasts indeed when measured up against the totality of what is possible. It is this rarity (taken together with the assumption of equal a-priori probabilities) which implies that a fitness spaces favouring evolution pops out of Dembski’s equations as objects of very high information. But as Felsenstein tells us that information traces back to the givens and contingencies of our very particular set of physical laws. Of course, real biology doesn't work with a white noise fitness space! But that is precisely the point; given that white noise is the overwhelmingly typical fitness surface, Dembski’s thesis is simply telling us that in the mathematical platonic realm of possibility the real (smooth) fitness surface of a workable evolution is highly a-typical; it doesn't follow that this means evolution is necessarily in trouble. I certainly don’t read Dembski as necessarily implying that the fitness surface has got to be a white noise surface!

Felsenstein: It is notable that Dembski and Marks’s argument is not actually an Intelligent Design argument. It argues that a Designer is needed to explain the shape of the fitness surface, but once that surface is smooth enough, natural selection and other evolutionary forces do the rest. So there is no Design Intervention needed.

My Comment: Here we have a very clear statement by Felsenstein of dualistic, deistic, God of the Gaps theology: According to Felsenstein, once the right fitness landscape has been set up God can then back-off and let “natural forces” do their stuff. As I have already said I can’t say that I’ve read enough Demsbki and Marks to say whether or not they are God of Gaps theologians but the fact is the North American ID movement has a susceptibility for an implicit “God of the Gaps” theology. If according to this theology physical determinism means that once that determinism has been set up God can withdraw and let natural processes “do it” then this view becomes problematic in the light of nature’s chaotic regime which is constantly perturbed by apparently random quantum events; down at the microscopic level some kind of  “new” information is being created all the time.

Felsenstein: In the No Free Lunch argument the performance of the search that moves uphill on the fitness surface is extremely poor if averaged over all possible fitness functions. This is the same as its behavior on a typical randomly-chosen fitness function. At least seven major criticisms of Dembski’s No Free Lunch argument have objected that white noise fitness functions are not realistic (links to their articles and posts are given in my 2007 article and in a summary I wrote here at Panda’s Thumb). The criticism goes back to 2002 and has been voiced by all these authors.…….But whether or not that theorem is proven, the point remains that evolution will do badly almost all the time on a white noise fitness function. So a smoother fitness function is required. But, as we have seen, the laws of physics make a white noise fitness function unlikely. This is true whether or not the HNFL theorem can be proven rigorously.

My Comment: I have no argument with this except to say that if Dembski’s theorems are correct then they shouldn’t be interpreted as necessarily contrary to the biological facts about the nature of fitness space; the white noise  model is not a challenge to biological realities but tells us something about the huge mathematical envelope of possibility within which real biology is just one possibility. From these theorems we begin to appreciate that actual biology (which is constrained by physical laws in the way Felsenstein has identified) is an extremely rare case. Although some might over interpret it as if Dembski’s theorem is an anti-evolutionary gambit it is in fact just telling us that evolution is a rarity in the relevant mathematical regions of platonic space; although Dembski's ideas could be used as an anti-evolutionary gambit if someone reasoned fallaciously along these lines: The conditions needed for Evolution to work  are such an improbable rarity that it couldn't have happened that way. Wrong.

Felsenstein: The point about physics and the unlikelihood of white noise fitness functions is also true however we define information, and it is true whether natural selection “creates” information or whether it takes existing information that is implicit in the smoothness of the fitness surface and repackages it in the genome. I suspect that Dembski and Marks’s “active information” will end up not being a helpful concept, but for the purposes of my present critique that issue is not central.

My Comment: If evolution works in the way experts like Felsenstein say it does then I would certainly go along with the foregoing. Felsenstein accepts that, depending on one’s definition of information, evolution appears to be “converting” the up-front information implicit in the fitness surface into genomic information. In this sense Felsenstein appears not to be necessarily at odds with Dembski’s mathematically abstract conclusions. However, I think that Felsenstein is probably right in anticipating that “Dembskyism” may not, in the long run, be that scientifically helpful, particularly to biologists. These biologists are concerned with the actual details of the fitness surface that makes or breaks evolution, rather than very general mathematics that simply abstracts away those vital details. But where Dembski does score is, I believe, is in the area of world view synthesis where questions of the enigmatic and very particular contingency (= information) of our world  leads to strenuous and sometimes bizarre attempts to explain it. (See Max Tegmark in this blog post)

Felsenstein: Is Dembski’s theology of information central to his argument about evolution?

No, because he’s got to end up arguing that, for the laws of physics to be the way they are, requires some active Design. But once the laws of physics are admitted, how they got that way is just not part of any argument about evolution. Biologists will certainly decide not to waste time on the issue and to leave it to cosmologists.

My Comment: Yes, the given burden of information that evolution requires, if physics is accepted as the depository of that information, is more than just biology. But this just moves the problem on! As Pauk Nelson has said "Filling one hole by digging another!".

Final Comments and Reservations.

I don’t want to fall out with either Dembski or Felsenstein; both of them earn my respect; tough luck that they inhabit respective sub-cultures that have come to loath one another. Felsenstein might be right about the laws of physics providing the up-front information needed to define the fitness landscape; after all, he is a biologist and in what is probably the computationally irreducible world of natural history no amount of analytical reflection is likely to compensate for hands on biological and paleontological experience. Someone like Felsenstein will have a good feel for natural history and his practical knowledge will (and perhaps should) trump theoretically reflected opinion (But then perhaps I ought to mention someone like biologist Cornelius Hunter of the North American ID community!)

But allow me a little doubt about Felsenstein’s convictions. Mere continuity of the fitness landscape is not enough; continuity doesn’t guarantee two further essential mathematical features that are needed to get classical evolution to work: I call them connectedness and linkage. If the landscape at a fitness peak plunges continuously into unfitness on all sides evolution will not go any further than the peak; If the fitness peaks are isolated islands surrounded by unfitness then evolution will not go far. What are required in the landscape are not peaks of stability but connected ridges of stability that provides pathways for diffusional migration, a migration that effectively defines evolution in its generalised sense. In fact if OOL is to work there must be a network of connected ridges of stability in fitness space that run all the way from inorganic matter to human beings. But something else on top of this ridging is needed. If the ridges are too thin (that is, if the linkage is too tenuous) then the number of ways an organism can plunge into unfitness (and extinction) will be far too great to give evolution a realistic probability. So, all in all, the question of whether standard evolution is viable could still go either way from a theoretical perspective.  Biologist Cornelius Hunter seems to have an intuition for the huge exponentiating space of possibility evolution has to  traverse, even given our own very restrictive laws of physics:

(See here for context o fthis quote )

And of course these designs are observed by us only because they were the evolutionary winners. They are the proverbial tip of the iceberg. For every winner there are untold myriad losers. The designs that produced some other chemical rather than benzyl acetone. The designs that detected chemicals that the caterpillars don't secret. The designs that didn't couple with the detection system. The designs that produced secretions that had no effect on the caterpillars. The designs that wreaked havoc on the flowering process rather than merely altering the flowering time. And so on, and so forth. The plant must have been a veritable idea factory, churning out all manner of mostly useless Rube Goldberg devices.

I have my doubts that the fitness surface is sufficiently connected and linked to allow classical evolution to work. Also, it seems strange to me that a cosmos where neither purpose nor teleology are thought by some to be meaningful,  we should find the solution to life encoded in the implicit fitness surface from the outset, a surface that in effect provides the front-loaded information for what is tantamount to a linear time “decompression” that reveals life. In response to my doubts my current avenue of probing/speculation can be seen in my Melencolia I series..

We don’t know in advance if physics effectively front loads evolution friendly information in the fitness surface. We certainly don’t know how many suites of physical laws define an evolution friendly fitness surface, so we don’t know the chances of happening upon a physical regime that is evolution friendly. If we were to start from this position of not knowing we would be unable to front load a “life-solution” into the cosmos from the outset. Instead we would have to search for the solution from scratch; trouble, is an ordinary linear time search is simply unable to traverse such a huge exponential space in a realistic time. What we would need is a system with sufficient processing parallelism to match the exponentiating vistas of configuration space. That kind of computational potential is hinted at in quantum mechanics. Quantum physics, I’ll hazard, is constraint, search mechanism and construction set all rolled into one package. In the light of this potential power it seems strange that evolution, as currently conceived, is an entirely (slow) classical process that doesn't exploit Quantum Mechanics 

But to think along these lines one has to entertain the idea that one’s anthropic gut feelings about teleology, purpose and aversion to pointlessness might be a reliable guide after all. To do otherwise is a bit like trying to decode some enciphered text without attempting to make a connection with the culture and thinking of the encoder, thus denying oneself possible access to valuable information. If one can’t make the assumption that we have some a priori anthropic insight into the nature of the cosmos there is no reason to think one is going to go anywhere very fast; what intellectual hope is there of making sense of insentient meaninglessness? The temptations of nihilism rear their ugly heads if a meaningless pointless cosmos is our starting point. However, one might have a chance of decoding and untangling the nature of nature if there is an anthropic agenda behind it. Atheists, of course, would eschew any entertainment of anthropic ideas. Fair enough; I’m not expecting anyone to follow suit – they are more than welcome to try and press ahead with their world view and see where it takes them.