Saturday, May 26, 2007

The Blind Watchmaker - chapter 7

This chapter starts with a misrepresentation of a common creationist argument: "People sometimes think that natural selection is a purely negative force ... ". The real issue isn't the natural selection, but the mutation. We don't try to induce mutations into our offspring for obvious reasons: We assume that they are overwhelmingly bad. As I mentioned earlier, Dawkins presumes a monotonic property to the fitness design space which is actually built into evolution theory. Given a constant slope and an equal number of uphill and downhill directions, evolution is guaranteed to head the direction of maximum fitness almost no matter how you work the survival statistics out. The problem here is that there aren't an equal number of uphill and downhill directions. At a summit, all directions are downhill. From the Dead Sea, all directions are uphill. In general, as you approach an optimum the number of increasing fitness directions decreases and the number of decreasing fitness directions increases. Eventually the bad mutations swamp the good ones. Evolution isn't all that efficient, so there is a basic problem that remains glossed over. Moving into X-dimensional design space where X is a large number, this problem is compounded with each new significant design degree-of-freedom. In the multi-objective problem, some of the objectives won't show up every generation. How often is there a plague? Again, the key issue has been completely evaded by the one who presumably has the superior knowledge and can address the issue head on.

The next few pages are appear to be an intron, because I don't know why this matters to the subject at hand. Then we get to a discussion of exon and introns which is worthy of some attention. The assertion is that there are unused portions of the genome - introns - and useful portions - exons. Dawkins claims that this is evidence against design, because it shows that random perturbations acted on the genes. Unfortunately for Dawkins, he explained this using an analogy of the text for The Blind Watchmaker. Various drafts together with the nature of DOS acting on a floppy left fragments all over the disk. So do we conclude that The Blind Watchmaker and/or the various fragments weren't intelligently designed? Perhaps we have evidence that Richard Dawkins doesn't exist!

Of course I am being a little unfair. Dawkins' point is that God wouldn't have been so untidy. This, however, is one of those little points where it is nice to bring out what biologists know, but won't tell you. The floppy disk, for example, is most cost effectively produced in sizes that exceed a majority of the programs out there. What is most efficient for DNA? What is the biological/fitness cost of keeping around a huge DNA molecule that is 90% junk in every cell of an organism? I am not a biology expert, but anyone who picks up a molecular biology text will immediately see that this junk is going to have a big impact on the rate that individual genes are transcribed and many other cell processes, even if the code is completely junk. My software development experience also tells me that unless you know 100% of what a program does, it is foolhardy to declare a portion as junk. Another issue (which might be quite naive) is that these junk portions might be quite useful in some sense related to disease. Some sort of decoy code? Regardless, I have exactly zero confidence in biologists honestly discussing this issue at a college level.

Most of the rest of the chapter is devoted to the notion of a biological arms race. This is the predator-prey paradigm, but again, Dawkins compares this to an arms race such as occurred between the Soviet Union and the US. Is this evidence that neither the Soviet Union nor the US employed Intelligent Designers? A key problem in evolution theory is that technology and intelligent design exist.

Links: Chapter 1 & 2, Chapter 3, Chapter 4, Chapter 4b, Chapter 5, Chapter 6,


Bunc said...

Hi Looney, havn't had much time to get back to you. I see that you continue to be unimpressed by Dawkins.

I havn't much time but just wanted to raise a couple of points;

1)It seems to me that in your discussions of fitness and mutation that there is an implicit assumption that somehow a species is maximally fitted to its environment and that therfore any mutational change would by definition produce reduced fitness.

But of course there is no reason to believe that any species is maximally fitted for survival. there is therefore plenty of scope on the "fitness landscape" for a species to evolve in ways which do not reduce fitness. That mutations are in mots cases deliterious is of course trivially obvious and not at all damaging to evolutionary theory. All that matters is that there is scope for mutational change which is advantageous and which results in greater frequency of that changed gene within the gene pool.

2) I have yet to hear from you what predictions you understand the "theory" of intelligent design makes which are in any way testable and falsifiable and which would qualify the "theory" as science rather than faith?

3) By the way does your fundamentalism extend to a belief in the "young earth"; "vegetarian dinosuars" and so on as propounded by the Creationist Museum?

4) I would also be very interested in how you explain the fossil record from an ID perspective.

Looney said...

Thanks again for the comments. I should say that I am a creationist who doesn't like creationists! Creationists are sloppy, don't check their facts and latch onto any rumour. It is embarrasing.

Most agnostics take the position that Darwin was 100% correct, but don't know what to make of God. You should probably view me as an agnostic creationist! The data looks puzzling, from a creationist viewpoint, but no less puzzling from a Darwinist viewpoint. Given the incentives of researchers to bury data they don't like, together with my first hand experience in the human fallibility of the best of the best in research, I don't see why this is unreasonable. We simply aren't going to have accurate data until the fanatics are no longer pushing the results one way or another. When will that happen?

You mentioned earlier that there are about 100 peaks in Scotland higher than 2,000 meters? If Scotland were a Dawkins design space, then there would be exactly one peak in all of Scotland higher than sea level. Only then is evolution possible.

The primary prediction of ID is that design features will on the one hand represent a tree or heirarchy, but the designer would periodically do something strange and out-of-order with respect to the hierarchy. Thus, pigs are preferred in medical research over chimpanzees because much of their physiology is closer to humans. (I got that first hand from a medical researcher.) The problem is alluded to in Dawkins' chapter on taxonomy, but I think he is precluded from discussing the full extent of the difficulties.

Bunc said...

1) "You mentioned earlier that there are about 100 peaks in Scotland higher than 2,000 meters? If Scotland were a Dawkins design space, then there would be exactly one peak in all of Scotland higher than sea level. Only then is evolution possible."

Please explain - I'ts been a long day and I am not getting your analogy.

2) "The primary prediction of ID is that design features will on the one hand represent a tree or heirarchy, but the designer would periodically do something strange and out-of-order with respect to the hierarchy."

How on earth does this follow from the central idea of intelligent design? First the Theory presupposes that we actually know what intelligence is and as I am sure you know there is considerable debate about that.

Second if there were an "intelligent designer" then by definition the designer, being capable of deisgning life - and giving it all the appearance of evolving - must have abilities beyond our comprehension. How can ID theory then assume to predict anything that the designer would do - other than post hoc point to things and say ah well thats what we see so thats what the designer must do!

There is no compulsion in an ID theory to have reproduction with heredity for example. An "intelligent designer" would be perfectly free presumably to have created a process that involved reproduction but without heredity.

Evolutionary theory however cannot exist without heredity and in fact predicts that there must be specific mechanisms for this. The theory was propounded before the specific mechanisms for heredity were known.

I agree that "design" tends to produce a hierarchy of designs with more complex forms at the tips of the "tree". (Talking very generally here of course). eg the first designed motor leading over time to a complex design space which includes aeropplanes, cars etc etc.

However there is a difference. There is a tendency for people to assume that evolution presupposes a trend towards more complex design. It does this only in a limited sense.

Given a need for some baseline complexity where things start out there is a tendency for the distribution of complexity to smear out in the direction of more complexity. However the bulk of living organisms remain relatively uncomplex in comparison to the forms which have evolved in the direction of complexity.

I am probably not explaining this clearly. Think about the number of known species of bacteria, viruses etc. These far outweigh and outnumber the more complex life forms like us, apes and elephants. What is predictable about this from an ID point of view? Nothing.

From an evolutionary point of view it is predictable though. There are many many niches for the more simple life forms so one would expect the bulk of life forms to remain adapted primarily to these niches.

3) The problem with reasoning from Genetic Algorithms or mathematical modelling is that this presupposes that any of our modelling is currently able to adequately model the massively parallel process involved in reproduction and selection for life forms on earth.

The complex interactions between the environment, organisms within species and between species is what produces selective pressures.

Lets not get caught up with our evolution/ID debate at the moment - wouldnt you agree that even the best models at present barely scratch the surface of being able to model such complex interactions?

One problem that strikes me is that with models we create we always need to predfine fitness artificially in advance. The world of course isn't like this (as far as we know). There is no reason to assume that there is any a priori "best fit" or best design for a life form. Indeed many life forms have features which appear sub-optimal. (and of course from my perspective this seems good evidence of the evolutionary "tail" - if you will excuse the pun!)

Nice talking to you.