Uncommon Descent Serving The Intelligent Design Community

Follow up to critics agreeing with Dembski re: NFL

Share
Facebook
Twitter
LinkedIn
Flipboard
Print
Email

Joe Felsenstein (Zoologist) at Panda’s Thumb responded [1] to my previous article [2] showing that a couple critics, Wolpert in particular who created the NFL, actually agree with Dembski.

He refers me to a paper he wrote [3] where he explains that the problem with Dembski’s argument is the relevant fitness landscape for evolution is not under the domain of the NFL.  While he may be right, I’m skeptical since Wolpert explicitly denies this in his paper.

Here is the quote again:

“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] [4]

Additionally, there are other reasons I am skeptical.  For one, while there are a few domains where the NFL doesn’t apply, the NFL applies to most.  It may be that evolution just happens to be concerned with the extremely small subset where the NFL doesn’t apply.  But even if it does, even within this select group, most of the group is mostly random.  While there are relevant cites within other papers by Wolpert and other authors on the NFL (which I can dig up if requested), it is easy to just see this result mathematically.  According to Kolmogrov complexity, the majority of the landscapes will be completely random in structure.  Of the rest, only a small subset has any significant amount of structure, and this subset shrinks at an exponential rate as your landscape size increases.  So, given the timescales and number of creatures and variety of environments that evolution posits for its effectivity, it is highly unlikely the landscape is suitably structured for any manner of effective search.

Second, Dr. Felsenstein’s big promise that obviates any concerns about the NFL is:

“They have overlooked the NFL theorem’s unrealistic assumptions about the random way that fitnesses are associated with genotypes, which in effect assumes mutations to have disastrously bad fitness. ”

As far as I know, the current consensus of population geneticists is that mutations do indeed have disastrously bad fitness.

Finally, perhaps Dr. Felsenstein is right after all and evolution does happen to possess the extremely rare and valuable fitness landscape whereby algorithmic search is significantly effective to be worthwhile.  In this case Dembski would be indeed wrong about the applicability of the NFL.  However, given the high specificity of such a landscape, this would mean that evolution itself is intelligently designed to an extraordinary degree.

[1] http://www.pandasthumb.org/archives/2011/08/criticisms-of-d.html#comment-panels

[2] https://uncommondescent.com/intelligent-design/critics-agree-with-dembski-the-no-free-lunch-theorem-applies-to-evolution/

[3] http://ncse.com/rncse/27/3-4/has-natural-selection-been-refuted-arguments-william-dembski

[4] http://cs.calstatela.edu/wiki/images/1/15/Wolpert-Coevolution.pdf

Comments
Sorry, Eric. I'll see if I can rephrase. There are two questions at issue: 1. Is coevolution, as we observe in biology, better than blind search when applied to all conceivable fitness functions? 2. Is coevolution, as we observe in biology, better than blind search when applied to the actual fitness functions in nature? Wolpert answers "no" to question 1, and Felsenstein answers "yes" to question 2, and neither disputes the other, so there is no disagreement. (Wolpert's answer does negate something he said earlier in his "written in jello" note, so Wolpert's only dispute is with his own earlier thinking.) Felsenstein's point has been made many times in varying degrees of technical detail for at least the past dozen years, in particular with regards to the subset of functions with low enough Kolmogorov complexity to fit in the universe.R0bb
August 30, 2011
August
08
Aug
30
30
2011
11:03 PM
11
11
03
PM
PDT
Wolpert shows that this statement is not true for his self-play model, but it is true “in general in biological scenarios”. This is what he means when he says that the NFL applies. Felsenstein does not dispute this. His point is that the blind-search-equivalence of every algorithm “averaged over all f’s” does not tell us whether evolution is effective for fitness landscapes found in the real world. It seems like you're saying that since Wolpert claims the NFL is true in general for biological scenarios then this means that the NFL may not apply in general for biological scenarios? I'm confused, can you clarify?Eric Holloway
August 30, 2011
August
08
Aug
30
30
2011
09:01 AM
9
09
01
AM
PDT
Thanks, that is an excellent response. I've received another along the same lines, so will be formulating another article to address your points.Eric Holloway
August 29, 2011
August
08
Aug
29
29
2011
10:53 PM
10
10
53
PM
PDT
Eric:
While he may be right, I’m skeptical since Wolpert explicitly denies this in his paper.
It appears to me that Felsenstein and Wolpert are talking about two separate issues. Wolpert is talking about whether "any two choices of a – any two sets of strategy-making rules {a_i} – perform just as well as one another, averaged over all f’s." (Emphasis mine. See two sentences beyond your quoted passage.) Wolpert shows that this statement is not true for his self-play model, but it is true "in general in biological scenarios". This is what he means when he says that the NFL applies. Felsenstein does not dispute this. His point is that the blind-search-equivalence of every algorithm "averaged over all f's" does not tell us whether evolution is effective for fitness landscapes found in the real world.
It may be that evolution just happens to be concerned with the extremely small subset where the NFL doesn’t apply. But even if it does, even within this select group, most of the group is mostly random.
The NFL applies only to permutationally closed sets, which the set of real-world fitness landscapes certainly is not. And the fact that these functions are actually implemented in our universe tells us without a doubt that they cannot be mostly random. As Tom English has pointed out several times, it takes a lot of storage to implement incompressible functions. Take, for example, an incompressible fitness function that maps 300-bit sequences to a fitness with a precision of just 1 bit. According to Seth Lloyd, such a function exceeds the informational capacity of our universe. And that's just one function.
According to Kolmogrov complexity, the majority of the landscapes will be completely random in structure. Of the rest, only a small subset has any significant amount of structure, and this subset shrinks at an exponential rate as your landscape size increases. So, given the timescales and number of creatures and variety of environments that evolution posits for its effectivity, it is highly unlikely the landscape is suitably structured for any manner of effective search.
We can take the probability of a subset as being proportional to its size only if all outcomes are equally probable. Without a justification for assuming equiprobability, we can't say that the rarity of structured functions in the Y^X renders them improbable. And as pointed out above, only the structured functions fit within the informational capacity of the universe.
As far as I know, the current consensus of population geneticists is that mutations do indeed have disastrously bad fitness.
As Felsenstein points out, NFL addresses the effectiveness of algorithms over the average of f ∈ Y^X, which is equivalent to requiring that a point mutation have the same effect as regenerating the entire genetic sequence randomly from scratch. Is it your position that point mutations do indeed have that effect?
However, given the high specificity of such a landscape, this would mean that evolution itself is intelligently designed to an extraordinary degree.
And that, of course, is the argument of Dembski and Marks. But as Dembski used to remind us, specificity is not sufficient for a design inference -- there must also be improbability under all relevant non-design hypotheses. It's not enough to simply conceive of an abstract space in which such laws comprise a tiny subset and then proclaim them improbable. The structured nature of fitness landscapes are a reflection of the regularity and continuity of the laws of nature. How do we even even begin to form null hypotheses for the origin of those laws when we know nothing about the meta-reality behind them?R0bb
August 29, 2011
August
08
Aug
29
29
2011
09:28 PM
9
09
28
PM
PDT
For most species it in fact very hard to measure the distribution of fitness effect (DFE) for mutations.
This is a refreshing admission that the process of evolution by mutations is a mystery even to those who accept it. The generalities don't translate to specifics very well. I must admire the technique of reducing "distribution of fitness effect" to an abbreviation, DFE. It makes it sounds so routine, giving the impression that someone, somehow, has a clue how fitness effects are distributed, and says it so much that he needs a shorter way to say it. I'm not directing this at you, markf, because I'm sure you didn't invent it, but that really lights up my BS detector.ScottAndrews
August 29, 2011
August
08
Aug
29
29
2011
03:16 PM
3
03
16
PM
PDT
mike1962,
“but if it was DESIGNED, then there is a DESIGNER and that designer truly is a moral monster.
"If a Brute and Blackguard made the world, then he also made our minds. If he made our minds, he also made that very standard in them whereby we judge him to be a Brute and a Blackguard. And how can we trust a standard which comes from such a brutal and blackguardly source?" http://books.google.com/books?id=e19zlwlOVwUC&lpg=PA66&ots=ICabyxjb9b&dq=if%20a%20brute%20and%20blackguard%20made%20the%20world%20cs%20lewis&pg=PA66#v=onepage&q&f=falseClive Hayden
August 29, 2011
August
08
Aug
29
29
2011
09:27 AM
9
09
27
AM
PDT
"Our objections are based on the universe as we find it. It’s about what we’d expect if materialism is true" Of course, you may speak for yourself all you want, but the universe as it is, is not at all what I would expect if materialism were true. "but if it was DESIGNED, then there is a DESIGNER and that designer truly is a moral monster. It depends on what the purposes of the designer are. And who we are. And why we are. Something tells me your imagination isn't very exercised in this area so as to think of plausible scenarios that wouldn't violate your tender sensibilities. I have. Your utterances are not impressive.mike1962
August 29, 2011
August
08
Aug
29
29
2011
08:55 AM
8
08
55
AM
PDT
MarkF- Strange that no one can produce any evidence of genetic accidents accumulating in such a way as to construct new and useful multi-part systems. Why doesn't that count against your position?Joseph
August 29, 2011
August
08
Aug
29
29
2011
04:21 AM
4
04
21
AM
PDT
Attack ID with your ignorance all you want it still doesn't provide positive evidence for your position.Joseph
August 29, 2011
August
08
Aug
29
29
2011
04:19 AM
4
04
19
AM
PDT
It looks like clicking on "Reply" in the box where your formatted output is displayed as you type and then clicking on "POST COMMENT" sends your comment to the most recent topic, whatever that is.dmullenix
August 29, 2011
August
08
Aug
29
29
2011
04:03 AM
4
04
03
AM
PDT
Sorry, that last comment was intended for the "The universe is Too Big, Too Old and Too Cruel" thread.dmullenix
August 29, 2011
August
08
Aug
29
29
2011
03:30 AM
3
03
30
AM
PDT
Sorry, my last two replies wound up attached to the wrong postings. Just making sure this time. My goodness, where to start? “1. If you don’t like the universe that we live in, then the onus is on you to show that a better universe is physically possible, given a different set of laws and/or a different fundamental theory of physics. Only when you have done this are you entitled to make the argument that our universe is so poorly designed that no Intelligent Being could possibly have made it.” This is an ID classic! May I remind everybody that it’s pretty standard Christian theology that God has already built a much better universe – a perfect universe in fact. It’s called “Heaven”. Consider that a “Proof by Demonstration” that a better universe can be built. With respect to point 4: Since it is ID Dogma that ONLY an intelligence can create information, and particularly the information found in the DNA of living organisms, then the Intelligent Designer logically must have also designed the DNA in malaria parasites (Behe admits this), typhus, influenza, bacterial meningitis, the black plague, small pox, diphtheria and all the other thousands of diseases that have been causing misery and death through all of human history. (Not to mention all the animals that fall victim every day.) Would it be asking too much of the Intelligent Designer to stop making them? Or to at least spare the children? And what’s all this predation? According to Christian and Jewish orthodoxy, the lions used to lay down with the lambs until two people ate one fruit and thus learned the difference between right and wrong. Was it really necessary for an omni-potent being to condemn half the animals on earth to become prey for the other half because of this? What did mice ever do to deserve cats? “5. Objections to fine-tuning are of no avail unless they are even more powerful than arguments for fine-tuning.” An extremely large or infinite number of universes, each with randomly generated laws, completely answers the fine-tuning argument. If you say the odds of any particular universe are 99999999999999999 to 1 against, all it takes is a few gazillion times more universes to make it a dead certainty. This works for any odds. 7: “(iii) but it is impossible that the God of the Bible could exist, because He is a “moral monster”; (iv) hence, the cosmological fine-tuning argument is not true. This is a pathological form of reasoning, since it is emotionally driven by a visceral dislike of the God of the Bible. Nevertheless, I believe this form of reasoning is quite common among atheists.” Not really. The Bible does make God out to be a moral monster – dooming humanity because of the sins of the first two humans, killing everybody on earth save eight because of the Hitlerian claim that they were all evil, hardening Pharaoh’s heart so He could continue to torment Egypt in order to show off His powers, punishing David for seducing Bathsheba and murdering her husband by killing their baby, finding Himself somehow unable to forgive the human race until He’d had His own Son tortured to death – but atheists tend not to believe the Bible. Personally, I think the only thing God could really be blamed for in the above list is killing the baby. The Fall, the Flood and the Exodus are mythical. Our objections are based on the universe as we find it. It’s about what we’d expect if materialism is true – but if it was DESIGNED, then there is a DESIGNER and that designer truly is a moral monster. Why Christians want to blame this universe on your God is beyond me.dmullenix
August 29, 2011
August
08
Aug
29
29
2011
03:16 AM
3
03
16
AM
PDT
G'day - AssieID (I hope this doesn't become a duplication. I though I entered this comment once - but it disappeared). I am confused by your first paragraph. The first sentence suggests you think the average mutation rate is 1 per generation. The second sentence that it is 300 per generation. The second sentence is the kind of rate I have read and roughly accords with my first comment. I glanced through the interesting paper from Eyre-Walker and Keightley but could not find any discussion of "how the unyielding genome decay is not being assuaged by natural selection: NS cannot remove the damage being done." It is a fairly long paper and I did not read it all in detail so maybe I missed it. However, I did search on "natural selection". There were only 5 occurrences - none of them relevant. It would be a surprising conclusion as most species clearly not suffer from a disasterous accumulation of deleterious mutations. They carry on just fine! What E & K do say is: (1) For most species it in fact very hard to measure the distribution of fitness effect (DFE) for mutations. (2) To the extent they can measure it most mutations are mildly deleterious not disasterous. To quote from their conclusion:
Although we have made some progress in understanding the properties of the DFE, there is still much to be learned. This raises a crucial question: can we ever really know what the DFE is? For a simple organism like a virus, this does seem to be possible, because most mutations have large effects that can be assayed in the laboratory. For most other organisms, and particularly for multicellular organisms, quite the opposite is the case; most mutations, even if they are deleterious, have such small effects that one cannot measure their fitness consequences.
markf
August 29, 2011
August
08
Aug
29
29
2011
02:59 AM
2
02
59
AM
PDT
G’day Mark, In my reading the average mutational rate per person per gen is about 1 so we’re all mutants! (X-men here we come). Our rate of mutation is much greater than 1, and the number per individual is closer to 300. In Eyre-Walker and Keightley’s “The distribution of fitness effects of new mutations” in Nature Reviews Genetics (2007) they write about eucaryotic mutational rates and how the unyielding genome decay is not being assuaged by natural selection: NS cannot remove the damage being done. All multicellular life is genomically decaying since the detrimental mutational rates are so significant, the individual mutations’ effects are so limited and finally there are no conducive beneficial mutations. NS does not remove the corrupted information so we continue to harbour all the problems and pass them on (with interest) to the next generation. The accumulated genetic damage to cells that is being wrought on us through the years and the significant mutational load on the genome that we pass on shows that we do indeed have “disasterously bad fitness". How long does a species have with such a genetic burden? Hmmm!AussieID
August 29, 2011
August
08
Aug
29
29
2011
01:36 AM
1
01
36
AM
PDT
As far as I know, the current consensus of population geneticists is that mutations do indeed have disastrously bad fitness. Are you sure? After all each of us has between 100 and 200 new mutations.markf
August 29, 2011
August
08
Aug
29
29
2011
12:26 AM
12
12
26
AM
PDT
“They have overlooked the NFL theorem’s unrealistic assumptions about the random way that fitnesses are associated with genotypes, which in effect assumes mutations to have disastrously bad fitness. ” As far as I know, the current consensus of population geneticists is that mutations do indeed have disastrously bad fitness.
You're missing Felsenstein's point; the word "random" is important here. He expands on this in his PT post:
But notice the averaging part. It is critical to Wolpert and Macready’s theorem. In effect, it says that we are dealing with an infinitely rough fitness surface. If we change a genotype by making one mutation — changing a single position in its DNA — we arrive at a genotype whose fitness is randomly chosen from the whole set of possible fitnesses. In effect, a single mutation has the same effect as mutating every site in the genome simultaneously. (I apologize for shouting, but the point is not being noticed over at UD). Of course real biology doesn’t work like this. Mutations are on average worse, but they mostly don’t instantly reduce the organism to rubble. In the real world, nearby genotypes are usually similar in fitness — often a bit worse but sometimes a bit better. In the NFLT world essentially all mutations are disastrous, and evolution would not work. So the No Free Lunch Theorem does not model real biology, not even in a simple model of evolution searching for genotypes of higher fitness on a fitness surface.
So yes, we know that there can be mutations with bad effects, but we assume that they are not randomly distributed through sequence space.Heinrich
August 29, 2011
August
08
Aug
29
29
2011
12:24 AM
12
12
24
AM
PDT

Leave a Reply