Aplologies straight away for clogging up such an excellent site with this old chestnut but in the light of GEM (Kairosfocus) having apparently directing a long OP at me over some exchanges in an earlier thread, and as Kairosfocus has closed comments I feel I ought to take an opportunity to respond here. I’ll put everything else below the fold.
The original point at issue cropped up when I posted a link to Elsberry and Shallit’s list of eight challenges in a previous thread, having had the list, which I had forgotten about (it was published in 2003, after all) drawn to my attention by a comment by Jeff Shallit himself, pointing out the list was still unanswered. Kairosfocus responded with:
>>1 Publish a mathematically rigorous definition of CSI>>
It has long since been shown, objections and censorship games notwithstanding, that reasonable quantitative metrics for FSCO/I and so for CSI, can be built and have been built. Indeed Durston et al have used such to provide a published list of values for 15 protein families.
Since then, I have been trying to clarify Kairosfocus’ claim that a “reasonable metric” for CSI [has] “been built”. As Durston’s FSC (functional sequence complexity) does not seem to exactly correspond to Dembski’s CSI, I suggested concentrating on Dembki’s concept. For the same reason I ruled out Kairosfocus’ own personal invention which currently appears to be FSCI/O (a sort of acronymic amalgam of FSC and CSI).
(However, I was sufficiently motivated to find these two threads on UD where Kirk Durston posts quite a few comments before mysteriously withdrawing. It seems from some of the comments that Kairosfocus was impressed by Durston. Larry Moran – Moran is a professor of biochemistry at Guelph and Durston was a Ph D student there in 2009 – was a little less impressed and there have been inferences that academic disapproval may have prompted Durston’s abrupt departure from UD comment columns.)
In the back-and-forth of the thread on Elsberry and Shallit’s challenges, Kaiosfocus finally responds:
25 –> AF then introduces a novel case, demanding that I address it (it was in fact already taken up by Joe):
Try the nine residue polypeptide, oxytocin . . .
26 –> AF knows or should know that 9 AA’s are represented by 27 mRNA bases, and come form a space of 20^9 possibilities for relevant protein space, where the maximum number of bits per 20-state element (disregarding redundancies) is 4.322, yielding ~ 39 bits as carrying capacity; well within the 500-bit limit for FSCO/I. He also knows that it is bio-functional. So, on a simple model of informational capacity adequate for this purpose, we already know that oxytocin is not beyond the threshold where design would be inferred. Chi_500 = 39 – 500; i.e. we are 461 bits short of the relevant threshold of exceeding the search capacity of the solar system’s 10^57 atoms.
So, Kairosfocus seems to be agreeing with my take that a CSI calculation for a protein sequence merely depends on a count of nucleotides residues. This is an analysis that results in two sets of proteins; shorter or longer than an arbitrary threshold. This would seem to be trivial and useless, merely telling us one thing we already know. And as Elsberry has emphasized with the now apparently redundant explanatory filter, you can’t rule out unknown possibilities and default to “design”.
So to Kairosfocus, the challenge remains unanswered. I agree with Elsberry and Shallit. There is no valid concept known as complex specification that can be used to distinguish designed from undesigned entities.
That would be at the limit. The actual calculation would be the ratio of the number of sequences with the given function and the number of possible sequences. This can be ambiguous, as it depends on the *degree* of function.
Your basic point is well-taken, though. They often seem confused as to whether CSI is a continuous measure or a Boolean result. Clearly, there is attempt to measure something, and there’s no reason why it can’t be applied to shorter sequences, even if the so-called design inference is not as cosmically certain.
Whatever became of Durston? His argument was a favorite of gpuccio’s.
Apparently you need the Boolean result before you can calculate the continuous measure.
If the sequence can be reached by natural processes, the calculation is irrelevant.
Hence the claim that function is too sparse to support a natural progression.
Hence Douglas Axe’s work and argument.
I believe this is the Durston you are referring to?
http://powertochange.com/organization/staff/kirk-durston/
Hi Zachriel,
Long time no speak 🙂
Well, my formal maths education stopped around 1970 🙁 But my issue has been, no matter how you manipulate the numbers, if your only input is a residue count, what can it tell you about the properties of any particular sequence. Durston tries to attribute a functionality measure but he can only do that for a protein where the function is known.
I am prepared to concede that Durston would not endorse KF’s inflation of his FSC into a tool that can genuinely detect or predict design in unknown protein sequences. It always comes back to how rare functional proteins are and whether mutations at known rates helped by duplications etc are capable of finding the functional intermediates.
@ Petrushka
Durston, KF and gpuccio all get to chat in the two threads I came across. Did you miss them. I note you didn’t comment.
Just stating that something is total bullshit is often not a very fulfilling argument. Even so, how can this sort of thing be taken seriously AT ALL? Much more is now known about how variation in genomic information arises than was known even ten or 20 years ago, and lots of what was previously seen as “random” doesn’t look so random anymore. But progressive change, in the sense of accumulation, chemical/metabolic, etc., and emergent properties and opportunistic complimentary niche matching and refinement within a complex physical, chemical, biological context that approaches infinity really defies such simplistic mathematical models or calculations. Sometimes bullshit is just bullshit.
I’ve commented on just about every thread where gpuccio has been, including some at UD before I was banned.
Gpuccio argued that protein domains are isolated in sequence space — that is they have no relatives. He therefore argues that they have no common ancestor and are presumably poofed into existence.
Pressed on this he argued that a designer might have created them by intelligent selection. I argued that natural selection is much better than artificial selection at navigating multidimensional spaces, and he seemed not to follow that line of reasoning at all. He seemed stuck on the idea that evolution was all about optimizing proteins for a specific function.
He could not accept the possibility that the history of a sequence can be erased by drift. I offered my word making program as an example of how drift can erase history, and his response was that I had designed the program.
I posted this over at Panda’s thumb in the discussion going on over there as well.
The entire CSI obfuscation can be boiled down to the product Np.
All Dembski has done by breaking down N into a bunch of factors is to allow the inclusion of a bunch of excuses to make the number of trials bigger so that no events in the building of the molecules of life can have a probability large enough to overcome it.
Take away the logarithms and the labels called “information,” because those completely obscure what is being discussed.
If Dembski, or anyone else in the ID movement, wants to tell us that there is a maximum number of trials possible in the universe to produce a given event, and he grabs a number like 10^120, then the minimum probability required for Np ≥ 1 is p = 10^(-120).
So Dembski has to tell us why p for the occurrence of a given event in the universe is less than that.
Either he has to tell us that the probability has to be jacked up by the intrusion of intelligence in order for the event to occur, or that the number of trials to produce that event is not as large as he claims. He is not about to admit the latter.
What he has done instead is lard-up and obfuscate this simple calculation with hundreds of pages of “philosophy.”
Thanks Mike. I’m reassured it’s not just me.
I have looked in on your exchanges with that kairosfocus character, and I also looked at his extended rants as well as that closed thread.
I have known for decades now that the concept of CSI was just larded-up probability calculations attempting to make assertions of insufficient probability look scientific.
I have also been interested in how those characters over at UD attempt to defend them as well as force their opponents into adopting their jargon. This has always been a tactic of ID/creationism going all the way back to Morris and Gish – another unmistakable genetic linkage that can be seen in their tactical methods.
One should ALWAYS assume that when an ID/creationist proponent makes assertions and calculations, they will try to drag the “debate” onto their territory using their definitions, misconceptions, and misrepresentations. That is exactly what I am seeing with those characters over at UD; especially with kairosfocus.
Look at it as a socio/political tactic. I choose not to engage it but to study it instead. In something like 50 years I have never seen an ID/creationist debater behave otherwise. Kairosfocus’s anger and threats come from a socio/political agenda that is being thwarted by better educated people who understand the ruse.
My general feeling is that ID/creationists like that need to be taken down by nobodies coming out of nowhere and disappearing back into nowhere after the take-down.
Let me add my 2 cents, by repeating what I have said before. Leslie Orgel’s concept of specified information is not bogus. Hazen et al.’s calculation of “functional information” is also meaningful, and makes Orgel’s notion specific. For that matter in 1978, in American Naturalist, I introduced a calculation (in a narrow context) of “adaptive information” which is the same basic idea.
But whatever uses these concepts have, they do not “break the mo[u]ld” of “Darwinistic evolution”, (or even of “Darwinistical evolution” or of “Darwinisticaloid evolution”). They calculate how much useful information has been built into the genome — by natural selection.
So I disagree with all those who feel that SI is bogus. But I agree with all those who say that it does not tell us anything that causes us to reject evolutionary mechanisms.
You have no disagreement from me on that.
Information theory has its roots in physical phenomena. All the developers of this field knew that it takes energy to flip bits; so at its heart, there are legitimate connections between information and entropy as it is understood in thermodynamics.
Biologists are usually talking about “states” that can ultimately be traced back to environmental effects; and that underlying assumption connects bits with energy because biological organisms interact with the environment and exchange matter and energy.
Where things start going off the rails is when the tools developed for bits that have a correspondence to units of energy start being applied to objects that do not interact with each other or with an environment; i.e., there are no energy exchanges whatsoever among the objects representing the “patterns” perceived by a human.
Now there is nothing wrong with finding ways to characterize patterns in terms of things like “information”. It is an important area of research; but if there are no energy exchanges involved in those patterns and how they are characterized, it is a huge mistake to start using such patterns as representations of the constituents of a real, physical system.
Strings of ASCII characters don’t interact among themselves or with any environment; there are no energy transfers involved in the permutations of those strings. Using them as representations of molecules is inappropriate because the energies of binding sites are not considered, nor are arrangements that are physically not possible.
Dembski’s “Law of Conservation of Information,” or “Conservation of CSI,” or whatever variations the ID/creationist community of “theorists” come up with, are not descriptions of things that exchange energy among themselves or with an environment. Such a system in physics does not appear to be possible; constituents in an isolated system interact among themselves and entropy goes to a maximum and remains there.
Biologists using the ideas in information theory are on safe ground as long as they are dealing with patterns and arrangements that are energetically connected among themselves and with an environment. As long as a biologist knows about natural selection, there is a very high probability that individual will choose wisely. Imitate what nature actually does and you are pretty safe.
Here is a nice thought experiment that makes the point about the difference between “information” describing non-interacting constituents of a pattern and “information” that connects to entropy.
Consider a steel box with perfectly elastic ball bearings in outer space, away from gravity, and all in motion like an ideal gas.
Suppose at some instant in time we can quickly spray-paint the ball bearings in one half of the container red and, at the same instant, spray-paint the ball bearings in the other half blue. The “paint” is effectively massless, dries instantly, and does not effect the collisions among the ball bearings and with the walls of the container. (This is a thought experiment.)
This doesn’t change the thermodynamic entropy even as the blue and red become completely intermixed. The color in this thought experiment has nothing to do with the energy and energy exchanges taking place among the ball bearings.
One could use an expression like the thermodynamic entropy to attempt to describe what is happening to the colors; but that expression has nothing to do with the physics of the system. It is a description of a changing color pattern. There is nothing wrong with trying to characterize that change in some mathematical fashion; but it has nothing to do with the energy. There is no trend toward equilibrium of energy states taking place.
@ Joe
That’s why I tried to concentrate on CSI as claimed by Dembski and developed by other ID proponents at UD. Hazen makes sense and even Durston limits his claims to existing sequences. I guess he had to do that to get published 😉
My simple point was, however you manipulate the figures, to pretend to be able to predict functionality of unknown sequences is the bogus element. On the other hand, if that is not the claim, then the process becomes trivial
Eric Anderson in a comment to me seems to accept that the CSI calculation is not yet done and dusted.
He writes:
@ Eric,
To your questions regarding a piano, computer and ribosome, in the ordinary sense of the word, they all have functions. As for a mathematical formula that can be used to equate function as a general property of any entity. I’m sure it hasn’t been done and my scepticism says it can’t be done. The piano, especially, makes me think of evolution. A bow is a weapon but that twangy sound has possibilities. Try different tensions, try a few extra strings. A lyre. Attach a gourd for resonance. A harp. Try tapping those strings. Try two bows, speciation – the violin! Sorry, I digress.
Now I am sure human inventors found sketches and maths very useful tools in designing and developing all sorts of things but there’s no more powerful tool than trial and error.
Re Shannon information. It is undoubtedly a powerful tool in telecommunications. And as you, too, realise, it has nothing to do with information content, merely* its compressibility, We get to the point of bogus CSI claims. The argument that I perceive from ID proponents is that there is some mysterious property of biological entities that prevents certain processes over a threshold of complexity from changing over time under the effects of mutation and selection. Some seem to claim they can demonstrate this mathematically.
Depending on how you define “function”, I make no such denial. Though I think “functionality” is as vague a descriptive as “intelligence” unless rigorously defined. My doubt is that there is a meaningful metric for some universal property called “function”. Similarly, with intelligence; even a reliable method of assessing human intelligence is not yet established and a universal measure of intelligence in the abstract is a pipe-dream.
Well, I have been told on numerous occasions that this has been done. Now you confirm that it is a goal – an aim – which is fine. Would it not be better to be honest about what has and what hasn’t been achieved? I don’t accuse you of dishonesty but to say “it’s been done” when it’s patently obvious that it has not makes one question the objectivity of those who say it.
*Not to belittle Claude Shannon’s foundational work in information theory.
My favorite comment on that thread is by Eric:
Best. Viola joke. Ever.
More Eric Anderson:
Eric, you seem to misunderstand the core of evolutionary theory which is reiterative steps of variation and selection. Even for the purpose of rubbishing it, understanding how evolution works would stand you in good stead. Dawkins might be the devil incarnate, but his explanations are very clear. “Climbing Mount Improbable” might be worth a read. I am sure other commenters here will have other suggestions of background reading.
I thought your, and franklin’s posts about chemistry were absolutely on point. It sometimes seems as though some ID proponents don’t just think that a designer was needed to design and tweak cells from time time, but actually drives the cellular machinery itself, the chemistry somehow being inadequate (despite being Designed!) to do the job.
A kind of vitalism, in other words. It’s not universal amongs IDists, but another striking characteristic of the ID movement is its lack of internal consistency. Some people think that an ID is needed only at specific points – Cambrian Explosion, OOL, bacterial flagella, Big Bang – while others seem to think that that all mutations are Intelligently Guided – and others still seem to think that tRNA is Intelligently Guided (or it wouldn’t find the right codon).
I noticed that too! Made me chuckle. The designer on the sidelines offering a musical accompaniment. Prune a few syllables and you’d have a great haiku.
It is the answer! A bare-bones reduced replicator would occupy every bit of its sequence with the task of replicating. By extending that sequence with bells and whistles, the core replication task is enhanced – necessarily more complex than the Mk l version.
If the world was infinite, this would not gain anything. If there was room and resources for everything, replication would remain a simple task, as there would be no driver increasing the representation of the complex versions. As soon as you put a lid on, replication itself starts to build the pressure. The finite world introduces competition, and replicators replicating at different rates start to uncover reward and penalty for their differences – there is a payoff for ‘bells and whistles’, if they increase market share. Without replication, nothing would happen.
That example, if you make the steel box be two boxes welded together with a small hole connecting them, and let the “left half” and the “right half” be the two boxes respectively, is the case of “Poincaré cycles”. I think we also have to assume that the “gas” is very dilute so the molecules do not collide with each other, just with the walls of the box.
In it, if we make an “entropy” (in quotation marks, note), it is not maximized but fluctuates up and down, occasionally returning to the its starting value. The expectation of the “entropy” does increase.
But I digress …
*** quibble mode on ***
This is often said, but I think is not correct. Darwin argued from finite resources, but natural selection can happen even with infinite resources. If you have, say, two asexual genotypes, and one replicates faster than the other, then over time both will increase, but the faster one will increase more, and its frequency among all individuals will increase. And that too is evolutionary change.
*** quibble mode off ***
‘Viola’ is an Eric Anderson specialty.
The alternative being poof.
My favorite there is:
C# maybe better? Maybe a whole dominant seventh on A?
The famous population geneticist James F. Crow (one of my original mentors) was a lifelong violist. Here you can hear him play, at a concert on celebrating his 93rd birthday, when he and some musician friends decided that they should honor the occasion by performing for the attendees.
One anecdote also involved his viola. Jim, who was at the University of Wisconsin, regularly played with the Madison Symphony, which often played in the orchestra pit at the Madison Opera. One time the members of his fruit fly lab played a prank on him — they put hundreds of chilled fruit flies into his viola. As he played in the orchestra pit, the fruit flies started swarming out. Jim was the only one who noticed this; he said that he immediately realized what had happened and just kept on playing.
That’s true, but complexity must have some impact on the replicative process. I’m not suggesting that there would be no selection, because there is always the possibility of differential reproductive rate, but that it would be less likely to lead to complexity. If becoming more complex could speed up replication, then I’d agree, but I don’t quite see how it could.
In the situation I had in mind, each organism is essentially in a ‘perfect bubble’. It has no competitors, because it has everything it needs – all the space and raw materials it could wish for. It has no need to over-elaborate. The equivalent is found in parasites – as soon as you fulfil the basic needs, the native capacity atrophies, and organisms become more simple.
That’s hilarious!
Well, it might depend on whether the organism that “has everything that it needs” could nevertheless get even more resources and grow even faster if it had some complicated new structure.
I love this. Eric makes a typo and Lizzie is all over it. And then the rest bob and weave to Lizzie’s street dancing!!
What a kick ass argument Lizzie!!! You’re the Tizzie!!!
Welcome Steve.
That’s the downside of commenting in public. Some people may point out glaring grammatical errors rather than address the points
theyyou are trying to make. What can you do? (Perhaps make a mental note to not confuse the musical instrument – the viola, with the exclamation borrowed from French- voilà. Or, alternatively use another word).On the other hand, we do also point out that Eric Anderson seems rather uninformed on matters of biology and chemistry.
And a little harmless banter?
Presumably you are familiar with the website, Uncommon Descent. That’s a very sooty pot you have there.
Given that Lizzie’s actual arguments are rarely addressed by anyone at UD what difference does it make?
Do *you* actually have any comment on the OP or can I just assume your position is that “KF is right, Lizzie is wrong” by default?
Steve – it is not as if the typo is the sole point made in relation to the arguments. And some typos are funny. Best Behaviour rules are reasonably suspended when it’s funny, provided it’s not cruel. I’ve made my share, and seen the funny side.
Steve, I’ve made a number of arguments in response to some of Eric’s posts, but he doesn’t seem to read TSZ, and hasn’t responded to any of them so far.
And that little strawman that I cited there did come with a rather sweet typo attached, and I couldn’t resist.
I am a viola da gamba player, after all.
Since the other organisms may as well not be there (as they don’t impinge on each other’s resources) it would be equivalent to a setup where every time an E Coli split, the daughters were placed in new jars. Each jar contains just one E Coli. New jars would be required at a faster rate for some variants than others (so there would be differential reproduction) but with infinite space to hold jars, and infinite medium to fill them with, could there be any work done by this NS?
Of course, in an infinite setup, some lineages could become infinitely complex. But what would drive that, other than stochasticity? Infinite resources turn off the ecological, competitive aspect of evolution. Differential reproduction among ecological competitors is a different thing from ecologically isolated “differential reproduction”, which (maybe wrongly) infinite resources implies to me.
Vaguely related, an interesting paper investigating complexity in Avida populations, taking some of the common themes re: information and complexity, though making a conflationary howler between thermodynamics and informational entropy: here
Allan,
Yes, assuming that resources are not just infinite but also infinitely dense. Otherwise there could be local shortages even though total resources are infinite.
Well, only those mutants that were incapable of reproduction would suffer extinction, which is quite different from the real-life case. But there would still be extinction, and the population would come to be dominated by the fastest-reproducing strains. So selection would still be operating, albeit in a weaker form than usual.
There are really two questions to be asked:
1) What would drive the emergence of complexity?
2) Would a complex variant come to dominate the population?
The answer to #1 is “stochasticity”, but the answer to #2 is “it depends on its reproductive rate relative to the other variants.”
The bottom line is that complexity would be favored to the extent that it leads to increased reproductive success.
But where are my manners?
Thanks, Steve, for coming over! I do appreciate direct communication.
Steve,
If you run into Eric, invite him over here. We’d love to engage him directly on matters more serious than spelling errors.
Hi Alan. Joe’s had a go at a rebuttal on his blog.
He asks the question (as a title):
“How Can CSI be a Bogus Concept When the World Depends on It?”
and in point 2 (point 1 is a grumble that “CSI pertains to ORIGINS and Darwinistic evolution takes place AFTER the origins”) he tells us why:
“2- The world depends on CSI so I doubt it is a a bogus concept”
So there you go!
How Can CSI be a Bogus Concept When the World Depends on It? because “The world depends on CSI so I doubt it is a a bogus concept”.
It’s a very sophisticated line of reasoning using S4 modal logic.
How many entities do we have a (C)SI calculation for that returns an actual value?
NONE. You can follow all of KF’s links and find NONE.
Indeed. The only example of someone on UD attempting to calculate CSI according to Dembski’s definition to my knowledge is vjtorley. He concluded that a gene duplication does, in fact, generate CSI.
Later in the thread he attempts to squirm away from his conclusion and still later wrote Why there is no such thing as a CSI scanner…, applying apologetics to Dembski’s math.
Even the most enthusiastic intelligent design creationists recognize that CSI is a bogus concept.
‘Recognize’ and ‘admit’ appear to be very different things. CSI is now simply a placeholder for ‘looks designed to me’ all wrapped up in sciencey-sounding words.
Well, I think the question gets conflated with “how do we tell if something is designed?”
People assume that by rejecting CSI as bogus, we also reject the possibility of detecting design. Well, I don’t. I think CSI is a bogus (KF, if you are listening, I don’t mean “fraudulent” I just mean “doesn’t do what it says on the tin”) concept not because design detection is impossible, but because CSI neither reliably detects design nor reliably rejects it. It has neither Positive Predictive Validity nor Negative Predictive Validity. And that’s because whether it works at all depends on whether you can construct a non-design null. For some things (like detecting a biased die) it works fine, because you have a clear non-design null (and you don’t need the BIGNUM part even – most people would be perfectly happy with a much less stringent alpha).
But for biology, it’s absolutely hopeless.
The only hope for ID, IMO, is to demonstrate that some feature of the simplest possible common ancestor of modern living things is still too complicated to have arisen through chemistry. CSI won’t tell you that, though.
But I do wish they’d start calling us OOLian materialists instead of Darwinian materialists.