Tuesday, January 17, 2012

Critics agree with Dembski, the No Free Lunch theorem applies to evolution

Copied over from Uncommon Descent:


We’ve all noticed the ID critics all speak outside of their realm of expertise. Biologists expound their expert opinions on mathematics, mathematicians make claims about computer science, and computer scientists think they know it all when it comes to evolution.
So, I thought, what happens if I only listen to their opinions in their actual realms of expertise?
Here’s a mathematician, MarkCC, author of the blog “Good Math, Bad Math.”
What’s his expertise? Math. What does he say about Dembski’s mathematics?
“he’s actually a decent mathematician”
What is not his expertise? Computer science. What does he say in the domain of computer science?
“But his only argument for making those modifications have nothing to do with evolution: he’s carefully picking search spaces[competitive agent (co-evolutionary) algorithms] that have the properties he want, even though they have fundamentally different properties from evolution.” [formatting mine]
Here MarkCC misunderstands the point of said paper, which is to define the how fitness of agents in co-evolutionary algorithms should be measured in general, regardless of the search space. (As an aside, he also doesn’t realize the triangle inequality can apply to evolutionary scenarios as well: B outbreeds and eliminates A, C outbreeds and eliminates B; but A could have outbred C given the chance.)
But, MarkCC is excused since both of these issues are outside of his realm of expertise.
Alright, let’s look at what the computer science experts have to say, namely Wolpert. Wolpert responds to Dembski’s earlier work on the NFL, which didn’t address co-evolution.
Let’s remind ourselves that Wolpert’s expertise lies in algorithms, not in biology. Does he detect any problem with Dembski’s understanding of the NFLT? Well, if Wolpert does, he says nothing. Instead, the supposed problem is:
“…throughout there is a marked elision of the formal details of the biological processes under consideration. Perhaps the most glaring example of this is that neo-Darwinian evolution of ecosystems does not involve a set of genomes all searching the same, fixed fitness function, the situation considered by the NFL theorems. Rather it is a co-evolutionary process.” [formatting mine]
So, within Wolpert’s domain of expertise he detects no problem with Dembski’s work, just like MarkCC, or at least is silent. Wolpert’s only complaint is outside his field, whether Dembski correctly formalizes evolutionary processes within his argument, not that Wolpert has much sympathy for Darwinists either.*  He does indicate that he believes the NFL does not apply to co-evolution**:
“Roughly speaking, as each genome changes from one generation to the next, it modifies the surfaces that the other genomes are searching. And recent results indicate that NFL results do not hold in co-evolution.
Now for the punch line: but what happens when Wolpert does examine the evolutionary details and whether the NFL applies to them?
“In general in biological coevolution scenarios (e.g., evolutionary game theory), there is no notion of a champion being produced by the search and subsequently pitted against an antagonist in a “bake-off”. Accordingly, there is no particular signifcance to results for C’s that depend on f.
This means that so long as we make the approximation, reasonable in real biological systems, that x’s are never revisited, all of the requirements of Ex. 1 are met. This means that NFL applies.” [formatting mine]
It is commonly noted that when smart people achieve expertise in a certain area, they suddenly think they are experts in many others, even when lacking the necessary knowledge. When listening to smart people, it is always wise to take this into consideration, and listen most closely to their opinions about what they’re carefully studied.
The ID debates are no exception.
* “First, biologists in particular and scientists in general are horribly confused defenders of their field. When responding to attacks from non-scientists, rather than attempt the rigor that the geometry of induction and similar bodies of statistics provide, they fall back on Popperian incantations, trying to browbeat their opponents into acceding to the homily that if one follows certain magic rituals—the vaunted “scientific method”—then one is rewarded with The Truth. No mathematically precise derivation of these rituals from first principles is provided. The “scientific method” is treated as a first-category topic, opening it up to all kinds of attack. In particular, in defending neo-Darwinism, no admission is allowed that different scientific disciplines simply cannot reach the same level of certainty in their conclusions due to intrinsic differences in the accessibility of the domains they study.”
** From the comments regarding how exactly the NFL applies to co-evolution:
What Wolpert is saying here is that co-evolution can produce fitter competitors, but it still cannot produce complex functionality:
“For example, say the problem is to design a value y that
maximizes a provided function g(y), e.g., design a biological
organ that can function well as an optical sensor. Then, even
if we are in the general coevolutionary scenario of interacting
populations, we can still cast the problem as a special case
of Example 1….
Due to the fact that they’re a special case of Example 1, the
NFL theorems hold in such scenarios. The extra details of the
dynamics introduced by the general biological coevolutionary
process do not affect the validity of those theorems, which is
independent of such details.”
However, it can possibly produce a better survivor:
“On the other hand, say the problem is to design an organism that is likely to avoid extinction (i.e., have a non-zero
population frequency) in the years after a major change to the
ecosystem. More precisely, say that our problem is to design
that organism, and then, after we’re done its ecosystem is
subjected to that change, a change we know nothing about a
priori. For this problem the coevolution scenario is a variant of
self-play; the “years after the major change to the ecosystem”
constitute the “subsequent game against an antagonist”. Now
it may be quite appropriate to choose a C that depends directly
on f. In this situation NFL may not hold.”
Note that this is consistent with ID’s claim that evolution cannot produce complex functionality.

No comments:

Post a Comment