Tuesday, March 26, 2013

Doug Axe Doesn't Understand Information Theory

]

Here we have the Discovery Institute's favorite biologist, Doug Axe, demonstrating his ignorance of information theory:

"... So, really, you put all that together, we now understand something about digitally-encoded information in cells, encoded in the genome. We understand why it's there: to encode proteins. And we understand how the proteins function to do the chemistry of life. And we also have the ability to measure, to some degree, how much information is there. If you put all that together, we now see something that looks very much like human designs, where we use digitally-encoded information to accomplish things, and we know that it's impossible to get information on that scale through a chance process that Darwinism employed."

This is false. We don't "know" any such thing. Axe cannot point to a single paper in the peer-reviewed literature that correctly explains why one can't "get information on that scale through a chance process that Darwinism employed". This is just something that creationists repeat over and over again without real justification.

In fact, just the opposite is true. Ironically, I am lecturing about Kolmogorov's theory of information today in my class CS 462. In that class we show that it is, in fact, while it is possible to produce information through a deterministic process (for example, by iterating the map xxx), it is even easier to produce as much information as you like through a random process -- precisely the opposite of what Axe is claiming.

"I remember thinking at the time that this looks like something, not just the product of engineering but the product of brilliant engineering. And that was the point where it occurred to me that someone needed to do the experiments to test whether that was really the case or not."

No experiment that Axe has done has tested the question of whether life occurs through the process of "brilliant engineering" or not. No one has a testable definition of "brilliant engineering" and no one has a procedure to test whether something is "brilliant engineering". Wes Elsberry and I gave eight challenges related to this kind of claim back in 2003. Ten years later, and not a single creationist has taken up our challenges.

We recognize human engineering because we are good at recognizing artifacts: the characteristic products of human activity.

"It's strange how your preconceptions really color the way you process data. And some people just went along with what they were taught, and I never tended to do that. I was always questioning what I was taught, including Darwinism."

And of course, creationists are miraculously free of preconceptions. That's what they're known for!

Here are a few other conventional ideas Axe has rejected:

  1. It's not a great idea to publish your papers in a vanity journal where you yourself are the managing editor.
  2. If you're a scientist, it's not a great career move to work for a "scientific" institute that gets most of its funding from the Discovery Institute --- a group with a documented history of misrepresentations, and driven by religious and political goals.
  3. It's not a great idea to have your colleagues extol the brilliance of your work, especially when referring to papers that have received few, if any, citations.
But hey, just go right ahead and ignore those conventions. You're a questioner, right?

"If you believe that everything was cobbled together through random processes, then there would be a lot of junk, there'd be the residue of cobbling sitting there and that's why people jumped to this junk DNA hypothesis. They found out that a very small fraction of the genome actually encodes proteins --- that was the one aspect of genomes that we understood well, is that they encode proteins --- so they assumed all the rest of it is junk. Well, the truth is, we didn't know what the rest of it was doing, but that doesn't mean it's junk. And it's becoming increasing clear that it isn't junk, and that's a significant prediction. It's not a prediction that Darwin himself made, but it follows very readily and naturally from Darwinism, and it turns out not to be correct. And that's becoming increasingly clear."

Axe misrepresents the history. Junk certainly could arise from an evolutionary algorithm, but it need not. It's logically possible that junk could have such a high evolutionary cost that it would tend to be weeded out. Acceptance of junk DNA came from data, not just theory. If you maintain that there is little or no junk in the genome, you have to explain exactly why different species of Allium have such wildly different genome sizes.

Axe likes to claim that he questions everything. But he hasn't questioned the ENCODE claims, even though they've been widely criticized. I guess that's due to his miraculous lack of preconceptions.

21 comments:

Diogenes said...

What a lying piece of shit! Are all IDers pathological liars?

Axe: "They found out that a very small fraction of the genome actually encodes proteins --- that was the one aspect of genomes that we understood well, is that they encode proteins --- so they assumed all the rest of it is junk."

Lying piece of shit! Molecular biologists have known since the 1950's that non-coding DNA can be functional! The Nobel Committee has handed out a shelf-full of prizes to scientists (none of them creationists) for finding functions in non-coding DNA!

How many times will these pathological liars repeat this lie?

Do we need to remind this lying piece of shit of the WHOLE HISTORY of molecular fucking biology!?

Nobel Prize for Jacques Monod and co-workers, 1965, for finding functions in non-coding DNA (regulatory elements).

Nobel Prize for Barbara McClintock in 1983 for her discovery of new functions in non-coding RNA (mobile genetic elements.)

Nobel Prize for Tom Cech and Sidney Altman in 1989, for discovery of catalytic functions resulting from non-coding DNA (catalytic RNA= ribozymes).

Nobel Prize for Jack Szostak and co-workers in 2009, for research in 1980’s on function in non-coding DNA (telomeres).

Nobel Prize for Richard Roberts and Phillip Sharp in 1993 for discovering introns (in non-coding DNA). [http://www.nobelprize.org/nobel_prizes/medicine/laureates/1993/press.html]

The structure of tRNA was known by 1964, crystal structure solved in 1974. tRNA is made from non-coding RNA.

The ribosome was known to be largely nucleic acid in the 1950's, general molecular structure known since the early 1970s, by the 1980's it was known the ribosome was a ribozyme-- based on functions residing in non-coding DNA.

Explain to me how scientists did not pay enough attention to function in non-coding DNA?

Meanwhile, the budge for Axe's Biologic Institute is about $300,000 a year. In the last four years, that's $1.2 million. For that much, how many nucleotides of non-coding DNA did they discover to have a novel function?

ZERO.

Diogenes said...

Axe: "And it's becoming increasing clear that it isn't junk, and that's a significant prediction. It's not a prediction that Darwin himself made, but it follows very readily and naturally from Darwinism, and it turns out not to be correct. And that's becoming increasingly clear."

BULLSHIT. What's "increasingly clear" is that after spending $400 million dollars, ENCODE could not find actual function in more than 10% of the genome-- if by function you mean contributes to the fitness of the host genome.

Here are some references to ENCODE's "death of Junk DNA" being debunked in the peer-reviewed literature:

1. The C-value paradox, junk DNA and ENCODE. Sean Eddy. Current Biology, Volume 22, Issue 21, R898-R899, 6 November 2012. doi:10.1016/j.cub.2012.10.002. http://www.cell.com/current-biology/abstract/S0960-9822%2812%2901154-2. Preprint: http://selab.janelia.org/publications/Eddy12/Eddy12-preprint.pdf

2. Can ENCODE tell us how much junk DNA we carry in our genome? Niu DK, Jiang L. Biochem Biophys Res Commun. 2013 Jan 25;430(4):1340-3. doi: 10.1016/j.bbrc.2012.12.074. Epub 2012 Dec 22. http://www.ncbi.nlm.nih.gov/pubmed/23268340

3. “On the immortality of television sets: “function” in the human genome according to the evolution-free gospel of ENCODE.” Dan Graur, Yichen Zheng, Nicholas Price, Ricardo B. R. Azevedo, Rebecca A. Zufall and Eran Elhaik. Genome Biology and Evolution Advance Access. February 20, 2013 doi:10.1093/gbe/evt028. http://gbe.oxfordjournals.org/content/early/2013/02/20/gbe.evt028.short?rss=1

4. Is junk DNA bunk? A critique of ENCODE. Doolittle, W.F. (2013) Proc. Natl. Acad. Sci. (USA) published online March 11, 2013. [PubMed] [doi: 10.1073/pnas.1221376110]

Curt Cameron said...

There's a new post over at Panda's Thumb:

Stephen Meyer needs your help

Meyer, in his last book, said that the best explanation for the information content of DNA is intelligence. However, in listening to his interviews, I was never sure whether he was referring to the DNA sequence of any life form (its information content not being able to grow with time), or maybe he meant the methodology of the DNA replication and how it uses the four nucleotides.

It sounds like for his new book, it's the latter - he is trying to use his information theory misunderstanding to cast doubt on the natural explanation for the Cambrian explosion.

Joe Felsenstein said...

Meyer, in his previous statements, has always left it carefully ambiguous whether he is talking about the origin of the DNA machinery or the subsequent change of the message in the DNA.

I once saw him debate against an evolutionary biologist. He announced that Digital Information has been found in the genome. The audience, mostly creationists, was wowed. Now they had been informed that what was in the genome was not just protein-coding genes, RNA-coding genes, control sequences and junk, but Something Else. Namely Digital Information! Clearly that was a sign of Design!

Of course if he had been pressed on it he would have to admit that what he meant by the Digitial Information was the protein coding genes, etc. Or else he would have said he was talking about the DNA machinery itself. He leaves that unclear.

kereng said...

"Ironically, I am lecturing about Kolmogorov's theory of information today in my class..."
The Kolmogorov complexity is not the only way to define Information.
Hazen and Szostak suggested to measure information concerning a function in living beings by the likelihood of acquiring that function.
I'm afraid, Axe uses Werner Gitt's definition of information, which requires a purposeful sender.

Jeffrey Shallit said...

Werner Gitt has no coherent definition of information.

Where can we find Axe's proof that information in the Hazen-Szostak sense cannot be generated by an evolutionary algorithm?

John Stockwell said...

Non linear systems that create information are as close as the musical sound of a dripping faucet. Perhaps intelligent design theorists will invoke the God of the Drips to explain that.

John Pieret said...

I remember thinking at the time that this looks like something, not just the product of engineering but the product of brilliant engineering. And that was the point where it occurred to me that someone needed to do the experiments to test whether that was really the case or not.

Except, of course, when it isn't, and poor Casey Luskin is reduced to mewling that the bad design of the Ford Pinto was still design:

http://scienceblogs.com/tfk/2006/11/20/nomination-for-stupidest-man-a/

It's strange how your preconceptions really color the way you process data.

Which is just the "presuppositions" "argument" every YEC uses to explain away inconvenient facts:

http://dododreams.blogspot.com/2009/02/err-presumptive.html

Walks like a duck ...

Unknown said...

" Axe cannot point to a single paper in the peer-reviewed literature that correctly explains why one can't get information on that scale through a chance process that Darwinism employed".
The problem is searching the domain of possible amino acid sequences. To find a specific protein of modest size (150 amino acids long) is beyond plausibility.
For example: As Stephen Meyer shows in his old book, the maximum number of events that could have occurred since the big bang is about 10 to the power 150.
(planck time (10pow43), times seconds elapsed (10pow17), times particles in known universe(10pow89). This is an absurdly generous upper limit. Compare this to the number of configurations of the domain of our modest length protein. There are 20 amino acids used in life. The possible arrangements for a protein of this length is thus 20 to the power 150, or ten to the 195. Thus this space would be unsearchable even if all sub-atomic particles in the known universe were exchanged for amino acids, each binding with another amino acid once per planck unit of time, from the big bang till now. (Also assumes the same amino acid chain never gets tried twice, and that they grow to 150 in length and no further!).
Please note this example doesn't account for all amino acids in life being of the left-handed form (2 to the power 150), or that they have to form peptide bonds rather than non-peptide bonds.
Now, as said above an 150 length amino-acid chain is modest, some are far bigger. Polymerase is three thousand amino-acids long, and is used in the ribosome during protein synthesis, so important to life. Note were dealing with exponents, 20 to the power 3000 is a huge number!
Obviously this is not the final word and nobody knows the composition of the domain space in terms of potentially functional proteins (that fold and are stable), nor how many different ways there are to much some molecular machine that will perform a given function etc. It does however show that the scale of the problem is likely to be too big for chance to be plausible.

Jeffrey Shallit said...

The problem is searching the domain of possible amino acid sequences.

Sorry, but that is not relevant to the question, which is how information is generated through random processes.

To find a specific protein of modest size (150 amino acids long) is beyond plausibility.

This is a basic error of reasoning. Who says that one needs to find "a specific protein"? It's like computing the probability that you would be born based on the probability that your parents, their parents, and their grandparents, etc., back 30 generations happen to meet and the particular sperm met the particular egg. Obviously this probability is astronomically small; yet here you are.

Jeffrey Shallit said...

Please note this example doesn't account for all amino acids in life being of the left-handed form (2 to the power 150

Oh, come on. Why do you think these events have to be independent, as they would have to be for you to multiply probabilities like that?

Unknown said...

“Sorry, but that is not relevant to the question, which is how information is generated through random processes.”

I think it’s indirectly relevant because if you can’t plausibly get there by luck then you would need to be guided by information. Also, a goal of Axe’s work was to provide estimates of the percentage of sequences in the domain that would fold into stable, potentially functional proteins. Though that isn’t stated in the video.

“This is a basic error of reasoning. Who says that one needs to find "a specific protein"? It's like computing the probability that you would be born based on the probability that your parents, their parents, and their grandparents, etc., back 30 generations happen to meet and the particular sperm met the particular egg. Obviously this probability is astronomically small; yet here you are.”

I think the truth is somewhere in the middle. In the sperm example any sperm will do. It’s not the same here. Firstly, as above, not all amino acid chains will fold into a stable form. Secondly, all sperm do the same job whereas proteins have to do a variety of jobs, requiring the protein to have certain properties (shape, charge, etc). For multiple of proteins to work together to do a job you will clearly rule out many more proteins than can combine in the right way to build a structure or perform a function.

“Oh, come on. Why do you think these events have to be independent, as they would have to be for you to multiply probabilities like that?”

There is no chemical reason for one over the other as far as I know. For arguments sake, say all the amino acids in the prebiotic soup were left handed, you still have a collective domain space of the range of lengths of proteins that’s vastly bigger than the largest conceivable physical referent.

Thank you for replying, I realize this is quite an old thread!

mns2012 said...

Jeffrey,

You are right in saying that information can be generated by random processes. However, how fast in practice is mathematically fast, is another question. This raises the question of statistical plausibility. Hazen information i.e. non-trivial function cannot be plausibly generated by stochastic and law like (selection) factors alone in the time bounds of natural terrestrial history.

In order to disprove that practical observation, you need to demonstrate the opposite, i.e. you need to demonstrate that a halting program can plausibly emerge from chaos in pretty much an order of 10^17 seconds given all natural processes and constraints.

I think that every time you want to prove somebody else is ignorant, you have to be extremely careful.

Jeffrey Shallit said...

you need to demonstrate that a halting program can plausibly emerge

You have no idea what you're talking about, do you? A "halting program", indeed. Try more comprehension and less babble.

Go read J. R. Koza, "Artificial life: spontaneous emergence of self-replicating and evolutionary self-improving computer programs", in Artificial Life III, C. G. Langton, ed., 1994, pp. 225-262, and come back when you've understood it.

mns2012 said...

Thank you for this reference and especially for your kindness.

Can I also suggest you read David Abel "The First Gene".

mns2012 said...

Jeffrey,

Thank you very much for the reference and for your exceptional politeness.

I am familiar with the works of Stuart Kauffman and Gregory Chaitin on this. Kauffman's work predated the work you recommended. Kauffman ended up questioning whether there is any law governing the emergence of life. Chaitin ended up deciding himself what program will and what won't work.

Mathematics is a language: whatever you assume will determine your results. Whether or not the assumptions are ludicrously unrealistic is immaterial to it. But in order to stay practical you need to always question your assumptions.

I maintain that your models guys need to be realistic. You guys need to learn to listen.

Jeffrey Shallit said...

I've read David Abel. His work is utter nonsense and completely without any value. I cannot think of a single serious scientist who thinks he has anything to say. There is a reason why his work is, for all practical purposes, only cited by creationists. You can check the literature yourself.

The fact that you would proffer Abel strongly suggests certain lack of discernment on your part. You need to be able to distinguish between real science published in real journals by real scientists, and transparent bullshit.

mns2012 said...

Who is then a serious scientist in your estimation? What about Niels Bohr or Albert Einstein or Max Born or Kurt Goedel? Is this real science?

Or maybe Mike Ruse who says life originated on the back of crystals?

You see, I am a practitioner and I did some research in the past. Science as such is just a formalization of human reasoning and therefore it is completely neutral to metaphysical questions. If somebody comes and says to me: "I can prove or disprove scientifically that your worldview is worthless", I will treat it as nonsense. No claims of this sort can be done on purely scientific grounds. This follows from the definition of science.

On the other hand, whatever scientific assumptions you start out with determines your results to a great extent.

Anyway, thanks a lot again for the reference. It looks like its not free access. I will try to get it when I am back to work from vacation.

Jeffrey Shallit said...

"Or maybe Mike Ruse who says life originated on the back of crystals?"

You seem quite confused. Michael Ruse is a philosopher, not a scientist, and he is not an expert on the origin of life. You probably meant Cairns-Smith, although how you could confuse Ruse with Cairns-Smith is beyond me.

Cairns-Smith does not assert that "life originated on the back of crystals". Rather, he put forth a speculative hypothesis about this idea. If you cannot distinguish between a speculative hypothesis and a positive assertion, I feel like your powers of discernment (again) need some improvement.

Yes, Niels Bohr and Albert Einstein and Max Born were all serious scientists. Goedel was not; he was a mathematician.

David Abel is a retired veterinarian who publishes the same incoherent drivel over and over again in (generally-speaking) venues of very very low quality. The fact that you cannot distinguish between Abel and people like Bohr, Einstein, and Born, suggests a certain lack of discernment on your part.

mns2012 said...

I admit the possibility of me lacking discernment in this. Thanks for agreeing Niels Bohr was a scientist. That is a relief ) But why discriminate between a scientist and a mathematician is beyond me )

In my opinion, Abel raises important questions regarding function, order and randomness.

Charles Darwin was a geologist as far as I remember (and again as far as I remember he did not hold any scientific degrees at the time of publishing his magnum opus, correct me if I am wrong), which did not stop him doing work in biology. By the way, his work at first raised lots of mockery on him by fellow scientists.

Anyway, it's getting late here. Thanks for your answers.

Jeffrey Shallit said...

why discriminate between a scientist and a mathematician is beyond me

They study different things and use different methods. At our university, mathematics is not in the science faculty. Of course, this classification is not universal and some disagree.

In my opinion, Abel raises important questions regarding function, order and randomness.

But, judging from the number of citations to his work, actual biologists and physicists and mathematicians don't seem to agree. Frankly, from reading his work, it seems like he doesn't really understand the subject very well. I teach Kolmogorov complexity in my 4th year course here at Waterloo, and I doubt Abel could solve the homework assignments I give.

his work at first raised lots of mockery on him by fellow scientists

It did, but Darwin was also praised and well-studied and cited immediately. Not true for Abel, who submits to venues of very low quality, writes in an obscure and bafflegab manner, doesn't present his results at peer-reviewed conferences, publishes his own books, and doesn't get cited by anyone.

Again, more discernment is needed.