Deconstructing Avida
| December 11, 2009 | Posted by William Dembski under Darwinism, Evolution, Informatics, Intelligent Design |
Back in 2003 NATURE (vol 423, pp 139-144) published an article by Richard Lenski, Charles Ofria, Robert Pennock, and Christoph Adami titled “The Evolutionary Origin of Complex Features.” The abstract reads:
A long-standing challenge to evolutionary theory has been whether it can explain the origin of complex organismal features. We examined this issue using digital organisms—computer programs that self-replicate, mutate, compete and evolve. Populations of digital organisms often evolved the ability to perform complex logic functions requiring the coordinated execution of many genomic instructions. Complex functions evolved by building on simpler functions that had evolved earlier, provided that these were also selectively favoured. However, no particular intermediate stage was essential for evolving complex functions. The first genotypes able to perform complex functions differed from their non-performing parents by only one or two mutations, but differed from the ancestor by many mutations that were also crucial to the new functions. In some cases, mutations that were deleterious when they appeared served as stepping-stones in the evolution of complex features. These findings show how complex functions can originate by random mutation and natural selection.
At no point in the paper is ID or any proponent of ID cited. Yet, when co-author Christoph Adami gave a PowerPoint presentation on Avida at a AAAS meeting some time back in Washington DC, his concluding slide showed Behe and his book DARWIN’S BLACK BOX. Moreover, Adami indicated that the whole point of this work on Avida was to refute Behe. Likewise, when co-author Rob Pennock wrote his expert witness report for the Kitzmiller v. Dover case, he claimed that his work on this NATURE article constituted a refutation of Behe.
The hypocrisy here is breathtaking. On the one hand, we are told that ID is not science. On the other hand, articles in places like NATURE appear that are clearly motivated by ID. And yet, the articles themselves are scrupulous to avoid referencing ID, its proponents, or published writings lest we gain an entry in the Science Citation Index and thus can further strengthen the case that ID is indeed science.
It was clear to the authors of the NATURE article that the shrill, illogical reviews of Behe that appeared early on would not silence him. But it was also clear to them that addressing him forthrightly in a prominent scientific venue could backfire, indicating that Behe was on to something important even if he was ultimately wrong. Some scientific mistakes are illuminating. If Behe were charged with committing an illuminating scientific mistake, then he would still be doing science (rather than pseudoscience or religion). Hence the subterfuge of not citing him at all the in NATURE article.
In any case, a thorough deconstruction of Lenski et al.’s article and of Adami’s Avida program has been long overdue. That deconstruction is now available:
Winston Ewert, William A. Dembski and R.J. Marks II, “Evolutionary Synthesis of Nand Logic: Dissecting a Digital Organism,” Proceedings of the 2009 IEEE International Conference on Systems, Man, and Cybernetics. San Antonio, TX, USA – October 2009, pp. 3047-3053.
Abstract: According to conservation of information theorems, performance of an arbitrarily chosen search, on average, does no better than blind search. Domain expertise and prior knowledge about search space structure or target location is therefore essential in crafting the search algorithm. The effectiveness of a given algorithm can be measured by the active information introduced to the search. We illustrate this by identifying sources of active information in Avida, a software program designed to search for logic functions using nand gates. Avida uses stair step active information by rewarding logic functions using a smaller number of nands to construct functions requiring more. Removing stair steps deteriorates Avida’s performance while removing deleterious instructions improves it. Some search algorithms use prior knowledge better than others. For the Avida digital organism, a simple evolutionary strategy generates the Avida target in far fewer instructions using only the prior knowledge available to Avida.
53 Responses to Deconstructing Avida
Leave a Reply
You must be logged in to post a comment.
Meaning the designer is not a designer in the sense we use the word, working in and with nature. Instead, it must be a magician employing magic methods. The way I see it, a god would be the best candidate for the job.
H’mm:
Seems the conclusion of the latest Dembski-Marks peer reviewed ID paper — congratulations — are also important:
_______________
>> III. CONCLUSIONS
A. Active Information
The Avida program uses numerous sources of active infor-
mation to guide its performance to successful discovery of the
EQU logic function. The sources include the following.
• Stair step active information. In the initial description of
Avida, the authors write [16]
“Some readers might suggest that we stacked the
deck by studying the evolution of a complex feature
that could be built on simpler functions that were
also useful.”
This, indeed, is what the writers of Avida software do
when using stair step active information. The importance
of stair step active information is evident from the inabil-ity to generate a single EQU in Avida without using it
[16].
• Active information from Avida’s initialization. The ini-
tialization in Avida recognizes the essential role of the
nop-C instruction in ?nding the EQU. Initializing using
all nonessential nop-A or nop-B instructions results in
the a decrease in NAIPI in Avida.
• Mutation, ?tness, and choosing the ?ttest of a number
of mutated offspring [5] are additional sources of active
information in Avida we have not explored in this paper.
B. Disclosure
According to the principle of conservation of information
[3], [4], [5], [7], [9], [10], [19], [20], [21], [22], [23], [28], [30],
all computer search algorithms of moderate to high dif?culty
require active information.
The conservation of information principle in computer
search, as manifest in the No Free Lunch Theorems, are [2]
“… very useful, especially in light of some of the sometimes outrageous claims that had been made of
speci?c optimization algorithms.”
To have integrity, computer simulations of evolutionary
search like Avida should make explicit
(1) a measure or assessment of the dif?culty, ?, of the
problem being solved,
(2) the prior knowledge that gives rise to the active infor-
mation in the search algorithm, and
(3) a measure or assessment of the active information,
either + or ?, introduced by the prior knowledge. >>
_______________
Have fun all.
GEM of TKI
PS: Been busy off-line and elsewhere, to those who inquired. (Some of it for not so happy reasons.) Cooking up some evil stuff. (have a look here at the functionally specific complexity of cellular metabolic reactions as a system. Compare to say the chemical reaction flows pathway of a petrochemicals plant. Then factor in our favourite 1,000 bit threshold and see where that gets ya.
Hi kairosfocus
I’ve just been having a look at your metabolic pathways link and the online presentation on compartmentalization by your (unnamed) biochemist who says he doesn’t believe in evolution. It’s pretty awesome stuff, and I must say that for sheer complexity, it leaves the petrochemicals plant in the dust.
They say that a picture is worth a thousand words, and your links are a perfect illustration of that saying.
Anyway, I’m glad to see you back.
Clive #35
That most assuredly was not your point, your point was to equate ID with astrology.
That was not my intention. I am sorry if I gave that impression. I chose astrology on a whim because of its links with Behe in the Dover trial. I could equally have chosen geology.
“Meaning the designer is not a designer in the sense we use the word, working in and with nature. Instead, it must be a magician employing magic methods. The way I see it, a god would be the best candidate for the job.”
Your entering the above text is an example of intelligent design. No natural laws imaginable would have produced this text. Are you a magician when you did this?
This is incorrect.
First, did you mean to say “random mutation plus selection” rather than what you actually wrote?
Second, ID theory accepts that it’s possible to do things like what the Dawkins WEASEL program does. Random variation and selection can indeed produce meaningful results. It is not true that “it is part of the theory of ID that this is not possible.”
Hi Clive,
I have two problems with your ID analysis of junk DNA:
1. It’s not clear to me how the prediction of “no junk DNA” follows from a ID hypothesis. In science, predictions are derived deductively from hypotheses. For example, we can correctly predict when the next solar eclipse will occur, based on the current positions of the heavenly bodies and the equations of Newton’s laws. How exactly does your “no junk DNA” prediction follow from what ID hypotheses? I am asking you to lay out the deductive reasoning.
2. It’s actually not true that there is no “junk DNA”, as far as I know. For example, human DNA is riddled with remains of transposons and retroviruses. That stuff is pure junk. Unless you know otherwise, the ID prediction is simply not true.
IrynaB,
If life is designed in whole or in part, then it is unlikely that you will find many totally useless aspects of the organism. For example, most organs are not appendices. But, ID could have predicted that the appendix does end up having a use. It is not merely “vestigal.” Same with DNA. It is more likely in an ID scenario that most, if not all, of DNA will have a useful function. In a neo-darwin scenario, you need lots of junk DNA to have enough raw materials for variation and a lot of vestigal DNA that was useful once, but is useless now. It is unlikely that that would be the case in a design scenario, so this is a prediction that ID makes.
Cabal:
If you have an argument for why the belief that a computer simulation is not teleological is a rational belief I’d love to hear it.
I daresay most readers here have at least some idea of what simulations are for.
hrun0815:
Compare the design portion of the following text with the evolution portion:
Design:
Evolution:
Here we have evolution meaning not evolution.
#39
First, did you mean to say “random mutation plus selection” rather than what you actually wrote?
Yes – thank you for correcting my error.
Second, ID theory accepts that it’s possible to do things like what the Dawkins WEASEL program does. Random variation and selection can indeed produce meaningful results. It is not true that “it is part of the theory of ID that this is not possible.”
OK. I need to spell it out a bit more explicitly. It is not possible without incorporating the target into the selection algorithm (or words to the that effect). The important thing is that there any many other groups that believe that evolution cannot explain the origin of complex features. Avida is an attempt to show how this can be done. It is about evolution, not about ID.
#41
If life is designed in whole or in part, then it is unlikely that you will find many totally useless aspects of the organism.
Useless for what? Are you making some assumptions about the objectives of the designer? If so, can we count any instances which do not support these objectives as evidence against the design hypothesis?
VJT:
Great to hear from you.
I see from the above that AVIDA has more than a passing resemblance to some of the problems with Weasel. (And that somehow it has not hit home that an intelligently designed computer simulation is not a really likely context for getting a good empirical demonstration of the claimed wonderful — I am tempted to say “magical” — design powers of chance variation plus natural selection.)
Anyway, I need to add that the author of the chart is the the International Union of Biochemistry and Molecular Biology. [Bottom left. That is, this is NOT anonymous, nor dubious. (Note the original source . . . I thought BU had linked that but realise now they copied it to their own server; sorry. The exchange at BU was set off by a med doctor and lecturer at med schools whose lecture heads the thread there.)]
Compared to the cluster of integrated functional pathways and associated enzyme catalysis, provision of intelligent substrates and clever compartments etc to support the process, the petrochems plant is put to shame. FSCI, and well beyond 1,000 bits.
And AVIDA simulations come up short too. I would like to see the scale of a program that could within the 10^150 state limit on our observed cosmos, set up a random search strategy with realistic rewards that would get to a structure significantly like the real world one.
I think it is well beyond the reach of the resources of our cosmos, unlike a toy example like AVIDA, which exploits bootstrapping mechanisms — as Dembski and Marks et al show — to ease the pathway to what is actually a fairly simple task: create an equivalent to an EX-NOR gate array in software.
True because it was predicted after function for some junk DNA had been found, no?
Does ID predict that all junk DNA has a function?
Since the concept of an unspecified designer could explain any observation it will accommodate both junk DNA and no junk DNA. The designer might have included mechanisms for utilizing or purging whatever was functionless DNA at any given time or he might have allow it to accumulate as long as it did not impose an intolerable burden on the organism. There is no way to decide without specifying the nature of the designer, something which ID resolutely refuses to do.
Vestigial doesn’t mean without use. It means without the primary function found in homologous structures of related organisms, but it may retain secondary fuctions.
The appendix is homologous to the end of the mammalian caecum, a pouch that holds a colony of bacteria that aid digestion of plant cellulose. In humans, cellulose digestion is not a significant aspect of diet, so this function is no longer as important as in its relatives. Secondary functions may include maintaining a shielded population of gut flora in case of bowel evacuation due to disease, and possible uses to the immune system.
Lack of strong positive selection in the appendix leads to wide variation in the size and structure of the appendix.
Ostrich wings are used for balance, but they’re still vestigial.
There is a cost associated with DNA replication, so a simplistic or first-order view of evolution would predict that excess DNA would tend to be weeded out by selection. Bacteria, which rely on rapid reproduction for success, have very streamlined genomes. That some DNA is clearly without function, means that this cost can be comparatively low in more complex organisms.
Zachriel,
I incorrectly conflated vestigal with useless.
It was my understanding that the appendix was evidence for unguided evolution because it was vestigal and useless. Since it was useless, it would never have been part of a design of human beings.
But a designer who is modifying one species with an appendix that performs its primary function, to a species that doesn’t need it for its primary function, may decide to keep the appendix for its secondary function. But any organ that has more cost than benefit will be discarded either by a designer or by evolution.
I said ” Collin: In a neo-darwin scenario, you need lots of junk DNA to have enough raw materials for variation and a lot of vestigal DNA that was useful once, but is useless now.”
You said:
“There is a cost associated with DNA replication, so a simplistic or first-order view of evolution would predict that excess DNA would tend to be weeded out by selection. Bacteria, which rely on rapid reproduction for success, have very streamlined genomes. That some DNA is clearly without function, means that this cost can be comparatively low in more complex organisms.”
Yet isn’t it true that the apparently high amount of junk DNA was used as evidence for evolution? Now the relative lack of truly junk DNA is evidence for evolution? Seems unfalsifiable.
The problem is that without positing specific characteristics of the designer, you could just say He had his inscrutable purposes.
Discarding or modifying an organ by evolution takes time. That is the key difference. (That, and the nested hierarchy of common descent.) Note again that vestigial organs tend to have high variability, which is consistent with relaxed selection.
There are countervailing forces. Pressure to streamline the genome is pit against the limitations of population and generation time. It takes time to eliminate junk. And with slow reproducers, little advantage.
I have no idea what a concept of an unspecified something (designer, rock, fantasy) means.
I fail to comprehend how concepts of unspecified whatevers can explain anything.
If “designer” specifies something, then “unspecified designer” is a contradiction in terms.
I am at a loss to understand how a contradiction in terms can explain anything.
Mung @ 52
It could mean proposing an intelligent agent as the primary cause of something but not specifying the nature of that agent since it is irrelevant, as Intelligent Design purports to do.
I agree. You should take it up with proponents of Intelligent Design.
No, there is no contradiction. It is quite possible to propose a designer as a cause without specifying its nature. All you are doing, though, is creating a label not an explanation.
It can’t but in this case there isn’t