No Signature in the Cell

 A review of Stephen C. Meyer’s Signature in the Cell

Pub 2009, Harper One (edition reviewed was the Kindle edition).


In late 2011, Stephen Meyer delivered a lecture in London. Organised by the UK’s very own Centre for Intelligent Design (C4ID), and hosted in Whitehall by Lord Mackay of Clashfern (a notable member of the Free Church of Scotland), I received an invitation in the post.  Circumstances surrounding this lecture coupled with some background reading I’d done on Meyer’s thinking and an awareness of how Intelligent Design creationists have in the past used academic attendees at events as some kind of litmus test of acceptance, I decided not to attend.  Instead, I was quite vocal (critical of ID creationism) in several internet fora, which attracted some criticism that I had not actually read the book in question (had I attended, I would have been given a copy).

I can summarise my opinion of the book quite succinctly. It is lengthy, tedious, overblown, very defensive at times, occasionally interesting, generally deceptive, but ultimately completely unconvincing to a practising biologist.  However, I did read this book with the intention of reviewing it, so here goes.

I reviewed the Kindle edition of this book.  This hasn’t made it terribly easy to work through as it’s quite difficult to move to and fro around the text, particularly if one wishes to take a quick look at a citation. HarperOne could have done a much better job.  Also the figures are pretty scrappy, resembling sketches drawn to assist a graphic designer: whether the figures in the paper version are better, I don’t know.

The Author

Stephen Meyer is a major figure in the American Intelligent Design creationism movement.  Along with William Dembski (who is extensively quoted in Signature), and Jonathan Wells (who’s on record as setting out to destroy Darwinism), he’s one of the intellectual leaders of this regrettable movement.  Meyer graduated with a degree in Physics and Earth Science from the christian Whitworth College, and worked for a while as a geophysicist before undertaking a PhD in the philosophy of science at the University of Cambridge.  The more interesting sections of Signature reflect his PhD: they are an account of hypotheses for the origin of life. Meyer then held teaching positions at Whitworth and at Palm Beach Atlantic University (another college with a distinctive christian ethos), before settling at the Discovery Institute, where he’s been a leading light in the Discovery Institute’s Center for Science and Culture.

Meyer’s background isn’t in biology, or molecular biology, which may explain some of the more startling asides in Signature.  His involvement in Intelligent Design can be traced back to around 1992.

The Discovery Institute and the Wedge Strategy

Somewhere along the line Meyer became convinced that life on Earth must have originated with a designer of some kind. According to Signature this was a realisation that followed his study of the many hypotheses of life’s origins that have been advanced over the years.  Meyer is credited as one of the authors of the Wedge Strategy document, which sets out the Discovery Institute’s strategy for intelligent design. To quote from the Wikipedia page:

The document sets forth the short-term and long-term goals with milestones for the intelligent design movement, with its governing goals stated in the opening paragraph:

  • “To defeat scientific materialism and its destructive moral, cultural, and political legacies”
  • “To replace materialistic explanations with the theistic understanding that nature and human beings are created by God”

There are three Wedge Projects, referred to in the strategy as three phases designed to reach a governing goal:

  • Scientific Research, Writing & Publicity
  • Publicity & Opinion-making
  • Cultural Confrontation & Renewal

Recognising the need for support, the institute affirms the strategy’s Christian, evangelistic orientation:

Alongside a focus on the influential opinion-makers, we also seek to build up a popular base of support among our natural constituency, namely, Christians. We will do this primarily through apologetics seminars. We intend these to encourage and equip believers with new scientific evidences that support the faith, as well as to popularize our ideas in the broader culture.

I’m including this outline of the position of ID creationism as seen by the Discovery Institute because I think it reveals an overall agenda in pushing a set of religiously motivated objectives, an agenda that initially was covert.  And the identity of the ‘Designer’ is clear from the Wedge Document (though proponents of ID creationism like to keep this vague). Looking into the history of Intelligent Design creationism, one sees that the Wedge Document (thought to date from around 1998) reflects a neo-creationist strategy that began in the early 1990s, and in specific response to successive legal defeats of co-called creation science in the USA, where the teaching of creationism falls foul of the constitutional separation of religion and state.

Intelligent Design is creationism

In 2005, an American court reached judgement on whether Intelligent Design could legitimately be described as science, in the context of the constitutional separation of church and state.  The case arose because a school board in Pennsylvania chose to present ID in schools.  In the end, the court (on the basis of very persuasive evidence) recognised that ID was a form of creationism, albeit one which had attempted to cover its tracks with a surface of science-y language.

Signature in the Cell

Signature is one of the main sources for Intelligent Design creationism.  It presents a version of ID creationism aimed at convincing the general public that not only has science failed to explain the origin of life (specifically the origins of genetic systems), but that it never will, and that ID creationism is (a) a scientific approach and (b) the best explanation for the origins of biological information.  Unfortunately for Meyer, he fails to persuade informed readers.

The Victimhood Card

Somewhat strangely, Signature begins with an account of the Sternberg affair, in which a journal editor abused his position and enabled publication of a paper by Meyer.  Meyer’s take is somewhat at odds with historical reality, making out that the whole affair was one in which Sternberg’s academic freedom was attacked.  However, detailed investigation revealed this to be far from the truth (see for example this report).  But I thought that beginning your popular science exposition of Intelligent Design creationism with a bid for martydom was more than a little odd.  It’s a theme that recurs occasionally as one ploughs through the tome.

Background to Molecular Biology

Next up is what’s essentially a primer on basic molecular biology.  This isn’t presented at a particularly heavy level, which is appropriate for the target general audience.  I thought the material was pretty clearly presented, though the reader would be well-advised to look at any one of the multitude of college-level texts that exist if they wish to get an even better feel for the complexity within the cell.  It’s while reading these chapters that the astute biologist gains the first inklings of the strategy to come, which is to argue that the complexity of cellular systems is just too much to have occurred through natural processes.  I think the biggest failing here is that Meyer’s background is insufficient to truly express how biological systems are systems in which ‘information’ (note that I use ‘information’ in a metaphorical way) can degrade, change and also expand and contract in quantity.  And what’s not made clear is that all the molecular processes that are required for evolutionary change are pretty well characterised and understood.

Strangely, Meyer has the hubris to believe he’s having major inspirational insights:

But notice too that there are no chemical bonds between the bases along the longitudinal axis in the center of the helix. Yet it is precisely along this axis of the DNA molecule that the genetic information is stored.

Amplified later as this revelatory insight:

There in the classroom this elementary fact of DNA chemistry leaped out at me. I realized that explaining DNA’s information-rich sequences by appealing to differential bonding affinities meant that there had to be chemical bonds of differing strength between the different bases along the information-bearing axis of the DNA molecule. Yet, as it turns out, there are no differential bonding affinities there. Indeed, there is not just an absence of differing bonding affinities; there are no bonds at all between the critical information-bearing bases in DNA. In the lecture hall the point suddenly struck me as embarrassingly simple: there are neither bonds nor bonding affinities—differing in strength or otherwise—that can explain the origin of the base sequencing that constitutes the information in the DNA molecule. A force has to exist before it can cause something. And the relevant kind of force in this case (differing chemical attractions between nucleotide bases) does not exist within the DNA molecule.

Well, I’m not so sure I find this an astonishing insight.  In the context of present-day DNA and RNA synthesis, the phosphate moeities are crucial in the formation of linkage between successive nucleotides.

Origins of Life

This is a genuinely interesting overview of how theories of the origins of life have arisen and been modified in light of increasing understanding of cellular molecular biology.  Unfortunately Meyer is clearly presenting a thesis that there are no successful hypotheses.  It’s here that the deeply unscientific nature of ID creationism becomes visible: “Science” is not some completed enterprise: indeed new hypotheses and findings are continually made – not least in the fields associated with the origins of life.  This is the more intellectually satisfying aspects in origins of life research: that novel hypotheses are continually framed and modified – and tested.  But hey, when you’re religiously motivated, why not cut and run and invoke a supernatural designer?

Design Detection, Information Theory & Specified (Functional) Information

There are two significant areas where Meyer draws on the work of colleagues, and in particular William Dembski, a mathematician who claims to have devised probability based methods for the detection of design.  Meyer relies heavily on William Dembski’s approach to detecting design.  I have several problems: essentially this becomes an exercise in probability, where the probability of events cannot really be calculated.

Yes, it may seem that the probability of life beginning was quite low, but Signature exaggerates this by seeming to insist on features of molecular replication being established all at once.  It’s as if Meyer is really keen to place obstacles in front of non-supernatural explanations.

Information Theory

Meyer’s use of information theory in Signature is difficult. As far as I can tell, Meyer doesn’t really have the background to adequately discuss this (and neither do I).  Meyer uses neither Shannon nor Kolomogorov information as the basis of his discussions, but a strange hybrid form in which not only the information is considered, but the message/meaning that is in the information. I find this completely uncompelling.  Meyer’s arguments require an understanding of which stretches of DNA sequence ultimately have functional significance, and he tends to avoid anything other than protein coding sequences.  He doesn’t consider, for example, the latitude there may be in protein sequence before a protein’s function is compromised.  I got the sense that for Meyer, the world is a rather binary place: stuff works or it doesn’t work.

Jeffrey Shallit, who has experience with information theory and mathematics has written several articles slamming Meyer’s treatment of information theory in Signature (for example Stephen Meyer’s Bogus Information Theory): much of this revolves around the undefined concept of functional specified information.

Meyer makes great play of the similarities between genetic information and human language – he gives many examples, even reiterating objections to Dawkins’ ‘Methinks it is a weasel’ computer programme, to which he imputes particular strategy in programming.  It’s interesting to note that the Weasel programme was a simple script to demonstrate how selection speeds up the attainment of an endpoint, in comparison to random letter choice.  This isn’t really a simulation of evolution, and of course the selection pressure here is to home in on a target phrase. I’m unsure quite what Meyer’s goal here is or was, but it’s instructive to read this take on the issue: Dembski weasels out.

Off the Fence: Meyer’s Intelligent Design

Finally, Meyer gets of the fence.  He believes that life originated as an act of Design. Unfortunately, Meyer refuses to come clean that he believes that God was that designer, at which is what I would be led to believe from his co-authorship of the Wedge Document (“To replace materialistic explanations with the theistic understanding that nature and human beings are created by God”).  Frankly, this is a hugely dishonest approach, but one in keeping with the strategy of Intelligent Design creationism proponents.

But who or what is the designer?

If we can take Signature at face value, no effort seems to be made to establish who or what the designer was.  There is no evidence in any shape or form for the existence of such a supernatural entity – though I believe that in the minds of ID proponents they have uncovered evidence OF a designer

Inference to Best Explanation

Meyer’s attempt to infer the existence of intelligent design are somewhat naive.  He summarises his strategy as “Inference to Best Explanation”.

Premise One: Despite a thorough search, no material causes have been discovered that demonstrate the power to produce large amounts of specified information.

The flaws with Premise One:  (1) There are plenty of chemical and biological mechanisms which can and do increase the quantity of biological information.  (2) ‘Specified Information’ is a bogus concept, and one which Meyer never actually defines. (3) A number of hypotheses have been advanced to explain the origins of biological information, and it’s only the straw man versions set up as easy targets by Meyer which fail.  (4) And finally, science will ultimately continue to generate origin of life hypotheses, some completely undreamt of as yet.

Premise Two: Intelligent causes have demonstrated the power to produce large amounts of specified information.

The flaws with Premise Two: (1) Specified Information continues to evade definition by Meyer.  (2) The only intelligent cause that Meyer can demonstrate is human cause, and (3) the complex information devised and generated by humans does not in fact correspond to biological information.

Conclusion: Intelligent design constitutes the best, most causally adequate, explanation for the information in the cell.

ID creationism fails.  Even were one to suppose intelligent design, for this form of creationism to gain traction, one would need at the very least to identify who or what this designer is (or was), and the means by which this intelligent designer undertook this major design effort (which actually exceeds the human ability at present).

Ultimately, Meyer’s concept remains a restatement of the classic ‘God of the Gaps’.  Because Meyer cannot conceive of a natural mechanism by which genetic information arose (or indeed believes that such a mechanism can never be uncovered) the role of a supernatural entity is the explanation.  Meyer rejects this, and rather unsuccessfully argues that his ‘Inference to Best Explanation’ is not such a ‘God of the Gaps’ argument.

Meyer restates his chain of illogic as follows:

Premise One: Causes A through X do not produce evidence E. Premise Two: Cause Y can and does produce E. Conclusion: Y explains E better than A through X.

For Meyer, the ‘E’ in the first premise is biological (genetic) information.  He also supposes that an exhaustive catalogue of explanations has been generated and rejected.  This reveals his unscientific approach: the list can not be exhaustive, and many explanations only fail for Meyer because of the way he’s set up straw men, and applied irrelevant statistical tests.  Meyer makes a direct connection between human generated information and biological information.  From my perspective as a research in molecular genetics, the premises must be restated as:

Premise One: Causes A through P do not explain (in Meyer’s view) the origin of evidence E (biological information), though we have yet to devise causes Q-X . Premise Two: Cause Y can and does produce F (human-generated information). Conclusion: Y explains E better than A through X.

This quite clearly illustrates the flaws in Meyer’s reasoning. I don’t believe that linguistic information or computer code are directly analogous to biological information.

Victimhood again

Meyer goes on at some length about how ID creationists in Institutes and Universities have been persecuted for holding heretical views.  An earlier example was the Sternberg affair, which Meyer misrepresents.  A similar accusation of individuals having their academic freedom trampled on is raised.  Unfortunately for Meyer (and fortunately for everyone else),  academic freedom allows academics to hold views, and conduct research that run counter to accepted wisdom.  This is entirely appropriate.  It can have quite serious effects: it’s why, for example, Peter Duesberg is able to continue claiming that AIDS is not caused by HIV – the failure to accept this has led to considerable mortality.  But what academic freedom does not (and should not) permit is the deliberate mis-education of students.  Frankly ID creationism is so far “out there” that it should not be represented in the undergraduate science curricula.

Meyer spends some time battling against the Dover trial, at which the presiding judge ruled that Intelligent Design was indeed a variety of creationism. But anyone reading the testimony could not fail to be convinced by the evidence that Intelligent Design was a deliberate strategy to make creationism look a bit less religious: particularly damning was the evidence from the iterative revision of the creationist text book Of Pandas and People. Also notable was that the Discovery Institute experts cut and ran from the trial when it became obvious early on that they would be involved in a big time courtroom defeat.

Is Intelligent Design creationism a testable theory?

Meyer wheels out the existence or not of junk DNA as an opportunity to portray ID creationism as a testable theory. He writes:

Consider the case of so-called junk DNA—the DNA that does not code for proteins found in the genomes of both one-celled organisms and multicellular plants and animals. The theory of intelligent design and materialistic evolutionary theories (both chemical and biological) differ in their interpretation of so-called junk DNA. Since neo-Darwinism holds that new biological information arises as the result of a process of mutational trial and error, it predicts that nonfunctional DNA would tend to accumulate in the genomes of eukaryotic organisms (organisms whose cells contain nuclei).

Evolutionary theory does not make the prediction that non-functional DNA would tend to accumulate. Whether mutational events lead to genome expansion would depend on selection pressures.  Within the eukaryotes, one can see huge variation in the extent to which ‘junk‘ DNA is present. Whether ‘junk’ DNA exists or not depends of course on how one interprets function: undoubtedly some DNA is present as a structural feature, some will be transcribed to yield functional RNA, and some will be transcribed to yield mRNA. P. Z. Myers gets it about right (Junk DNA is still junk) in a blog article on a peer-reviewed paper.

One assumes that the ‘predictions’ of ID creationism that ‘we expect DNA, as much as possible, to exhibit function’ (and that’s a pretty crappy prediction – as much as possible) are an attempt by ID creationists to read the mind of their creator designer.


While I’ve been preparing this review (it’s taken rather longer than I’d thought, due to pressure of work), I note that the Centre for Intelligent Design has posted an article about November’s Meyer lecture (C4ID’s Inaugural Lecture 2011: ‘Is there a Signature in the Cell?’).  Sadly, no video is presented.  It proclaims:

The content of Dr Meyer’s lecture was fascinating and his book, ‘Signature in the Cell’ is truly groundbreaking.  In his lecture, he outlined his initial interest in the problem of the emergence of first life and the absence, then and now, of any credible explanation for it.  He described how this had led him to Cambridge to undertake a PhD in the philosophy of science and to his discovery of the significance of the information content of DNA.  Using the method of inference to the best explanation, the very approach adopted by Charles Darwin in his elaboration of evolution by variation and natural selection, he described the scientific legitimacy of concluding that the information carried by DNA is best explained by intelligent mind.

This, frankly is nonsense.  The book is not groundbreaking but rather contains an insupportable claim not only that science has no credible explanation for the origins of biological ‘information’, but that it can never have such an explanation.  It uses outdated accounts of the origins of life, does not review more recent theories that have emerged since Meyer’s PhD thesis, and makes use of mangled information theory.  And as for ‘inference to best explanation’, the proposal that some unknown entity, with supernatural powers, was responsible for life’s appearance on Earth is not only ludicrous without evidence for the existence of such an entity but is intellectually hopeless.

There’s a rather succinct demolition of Meyer’s argument in a posting to Panda’s Thumb (ID: Intelligent Design as Imitatio Dei (report on the 2007 ‘Wistar Retrospective Symposium’)) which reviews a bogus conference staged by the Discovery Institute in 2007 (a bit like this one, but where actual scientists were invited – I presume the ID crew learnt from that mistake).  I somewhat regret not having read that article earlier, I might not have wasted the money on Meyer’s ‘groundbreaking’ book or more importantly the time it took to wade through it.


31 thoughts on “No Signature in the Cell

  1. I still have not read Meyer's Signature in the Cell, though I do have a copy. I must, however, take issue with the reviewer's comment that functional specified information is undefined. Nature had a brief article in 2003 by Jack Szostak arguing that biology must think in terms of functional information. That article proposed a simple mathematical definition of the term. This was followed up by an article by Hazen et al. in PNAS in 2007 that gave a more technical definition of functional information in biological applications. That same year, another paper by Durston et al was published in TBMM, mathematically defining functional complexity (an alternate term for functional information) with application to protein families.

    I must also take issue with the reviewer's belief that 'There are plenty of chemical and biological mechanisms which can and do increase the quantity of biological information.' Traditional concepts of information that do not concern themselves with functionality, yes certainly, but functional information as first defined by Szostak, no. There is no testable, verifiable method for chemistry and biology to actually produce statistically significant levels of functional information. Low level, statistically insignificant levels, yes, but nothing on the level of, say, being able to code for a novel protein family. Intelligence can produce functional information of course. Our recent success in producing artificial proteins is an example of intelligent design in action. Now we need a testable, verifiable model as to how chemistry and biology can do it, not toy computer examples, not creative stories, but real, testable and verifiable models. Until then, it is simply not the case that 'There are plenty of chemical and biological mechanisms which can and do increase the quantity of biological information.'

  2. Please supply the citations for the papers by Szostak, Hazen et al and Durston et al.

    I am a molecular geneticist, and I can see genetic information within a species increasing by segmental duplication. Within phylogenetic analyses, one can see the effects of duplication of chromosomes and whole genomes as a route to increasing the amount of genetic information. And if you think this is not a route to increasing 'functional' information you reveal your ignorance: duplicated genes diverge in sequence in function.

  3. Grumpy Bob, here are the citations:

    Szostak, J. W., 2003 Functional information: Molecular messages. Nature 423(6941):689.

    Hazen, R. M., et al., 2007 'Functional information and the emergence of biocomplexity'. Proc Natl Acad Sci U S A 104 Suppl 1:8574-81.

    Durston, K. K., et al., 2007 'Measuring the functional sequence complexity of proteins'. Theor Biol Med Model 4:47.

    I'm not sure how you 'can see genetic information within a species increasing by segmental duplication' if you are unaware of the papers defining functional information and, by inference, do not have the foggiest as to how it is mathematically defined or measured. As a first step, I suggest that you familiarize yourself with the literature cited above, especially on how functional information/functional complexity is actually measured, and then we can have an informed discussion on this.

  4. I see duplication of genes. In the literature I see generation of novel genes by duplication and gene fusion. This is increase in genetic information.

    I am a molecular geneticist, this is genetic information. Increasing.

    1. If I take a photocopy of some text, have I increased information? If your students plagiarize someone-else's research, have they increased information? If a gene duplicates, is there an increase in information?

      1. IDproponent: I think depends on the definition of information. In the case of genetic information, duplication of gene(s) gives rise to material upon which subsequent sequence divergence can be acted on by selection (and by drift) to lead to two distinct but related genes. One needs to think of events as chains of events. To take your analogy, if a student copies another student’s essay, then introduces modifications while the original student modifies his/her essay in different ways, then yes information has increased by any measure.

        But I think this misses Meyer’s claims, which are that science has not explained the origin of biological information bearing molecules, but that it can never do. He then invokes an unknown creator with supernatural powers to do the job. I think (though I am no specialist in origon of life) that science has formulated many hypotheses of processes that may have given rise to biological information, not all of which are given full consideration by Meyer, who also supposes that components of life had to leap into existence in a more or less fully functional state.

        1. The difference is, that if a student copies an essay and makes modifications, he/she does it intelligently. You are giving an example of intelligent design in action.

          But imagine a student who copies an essay, and then starts randomly altering the letters (maybe using a computer program), and then hands it in, spelling mistakes and all, would you credit the student for creating new information? Wouldn't such a process reduce the essay to gibberish?

          And yet you expect us to believe that gene duplication, followed by random genetic errors is the source of biological information.

        2. I'm sure you would say that Natural Selection would alleviate the problem. But Natural Selection can only select what is generated by mutation and drift. And if these genetic changes take functional sequences and turn them into gibberish, Natural Selection is impotent.

          1. Well, my modification to your analogy assumed you would realise that I was merely illustrating that duplication followed by divergence did yield additional information.

            In the case of duplicated genes, subsequent random sequence changes (mutation) coupled with natural selection (and random drift) leads to divergence between the genes and ultimately increase in genetic information. There is evidence for all these processes.

  5. "Meyer refuses to come clean that he believes that God was that designer, "

    Perhaps Meyer's entire book should be read before making such a claim, because Meyer does in fact state in that book that he believes that God is the designer.

    1. If Meyer comes clean and admits that God is the designer, he can be accused of having religious motivations. If Meyer refuses to come clean and admit that God is the designer, he can be accused of dishonesty. You see the beauty of it? Heads I win, tails you lose. The only problem is that none of this is relevant to the validity of his arguments.

  6. Looking at the three papers on functional information mentioned earlier, Szostak, Hazen and Durston seem to be providing a mathematical definition for what we might otherwise call ‘genetic’ or ‘biological’ information, or Meyers might call ‘functional specified’ information. The definitions provided by Szostack, Hazen and Durston are closely related and share their roots in a special case of Shannon information, could also apply to any other kind of functional information, such as ‘meaningful’ information for humans, though their focus is on biological information.


    Looking at Hazen’s definition of functional information (which they denote as I(Ex)), N is a constant and M(Ex) represents the total number of different sequences (known and unknown) that will meet or exceed the degree of function required (with ‘different’ being the key word here). Therefore, duplicating a gene produces no increase in functional information. As a sequence diverges, if the divergent sequences still code for the same function, no new functional information is produced, because M(Ex) is the total number of known and unknown sequences that will produce that function to the degree required by the cell, not merely the known or discovered ones.


    New functional information is produced if a duplicated, diverging sequence codes for a novel function. Durston’s paper presents a method to mathematically quantify the change in functional information/complexity in that case. If it is only a single amino acid change that produces the new function, the new functional information produced will be very modest.


    The big question, looking at Hazen’s equation and Durston’s results, is whether an evolutionary search engine is capable of producing higher levels of I(Ex). The ratio of M(Ex)/N is essentially a probability or uncertainty (as it always is in any form of Shannon information) so for higher levels of I(Ex), the probability M(Ex)/N is going to become vanishingly small. For example, Durston estimates (on the basis of Pfam data and their equations) that RS12 (121 aa) requires about 359 Fits (Functional bITs) to encode. Plugging that value into Hazen’s equation, we can solve for M(Ex)/N and find that the probability of finding a functional RS12 sequence in a single search is about 10^-108. The number of functional RS12 sequences M(Ex) appears to be about 10^49 but the size of 121aa sequence space (N) makes finding a functional RS12 sequence a needle-in-the-haystack problem. I don’t know if this is Meyer’s point or not, but it is a valid problem. We do know that it is trivially easy for an intelligent human to produce 359 Fits of information, so we already have a testable, verifiable candidate for genetic information. The question is whether an evolutionary search engine is effective in a needle-in-a-haystack problem.


    (To put my cards on the table, I’ve read very little of ID literature and would certainly not want to defend everything that they might argue (whatever it may be) but I am favorable to the idea that the ability to produce high-level, functional information is unique to intelligence. Thus, I think biological life shows empirical evidence of an intelligent origin and am quite fine with God as the candidate for both that, and the fine-tuning of the universe to be able to support biological life.)

    1. I read the Szostak (2003) article – I didn't find it terribly useful, quite brief and more of a small 'idea' article. His 2007 paper ( Hazen et al) is rather more thought provoking, and really codifies quite neatly how one might think of functionality in sequence. Unfortunately as I've never used Avida, I couldn't come to judgements about that specific discussion. I have yet to find time to read the Durston paper.

      To be honest, I didn't think that the Hazen concept of functional information corresponds with Meyer's (which isn't really defined in his book) – it's a lot more coherent for a start. But I'll look at the relevant sections of Signature again, once I've read Dunston.

      In any event, assertions that there are no known mechanisms where new genetic information can arise in biological systems are clearly false. Meyer's claim remains that natural processes can not have led to the appearance of replicating systems. One assumes that one version of an intelligent designer is that a supernatural entity established primitive replicating systems on Earth and then left it to run on.

  7. I got a chance to take a brief skim of a couple sections of Signature in the Cell where Meyer discusses biological information. That is one thick book! First, I must say that I could not see (in my quick skim) any mathematical definition for how he is using the term ‘functional information’, which does leave things a little vague. However, in his chapter of the same title as the book, he does make the same point (on why classical Shannon information is inadequate) that Szostak made in his brief article in Nature and suggests that that functional information is ‘Shannon Plus’. A revised form of Shannon Information is what Szostak, Hazen et al., and Durston et  al. all propose for the same reasons. So at least on that, there seems to be common ground with Meyer. Having said that, I think Szostak’s one page article is clearer for the lay person with little or no scientific background.


    The controversial claim Meyer makes is whether known natural processes can produce, say, enough functional information to code for a 150 aa protein. He does allow for sequence variability on the bottom of page 207. I skimmed over how he got the probability. The best it can do is give a very rough estimate that leaves a lot of room for questions.


    I think there is a way to test Meyer’s claim that is more accurate using actual sequence alignment data. In the Durston paper, there is a link to their software. I made a number of modifications/improvements to their software (e.g., I think a ground state that allows for the influence of the genetic code might be better than using the null state that they use). I also added a section to their code to estimate Hazen’s M(Ex)/N probability.


    Here are some results:


    RS7 domain: 148aa; 3,751 input sequences, 332 bits of func info, M(Ex)/N= 10^-100

    ABC2 Mbrn3: 343aa; 10,478 input sequences, 287 bits of func info, M(Ex)/N= 10^-86

    AA perm: 433aa; 16,267 input sequences, 466 bits of func info, M(Ex)/N= 10^-140


    (Note: input sequences are not Hazen’s M(Ex), they are only data to compute the probability distributions for Durston’s method of estimating the functional complexity/information required for various proteins.)


    Bottom Line: Hazen’s equation shows that the degree of functional information is inversely proportional to the -log of the probability M(Ex)/N. So for the degree of information required to code for the average protein, M(Ex)/N becomes so small that it raises the question of whether there is a known mechanism that will produce anything more than trivial levels of functional information (e.g., homologues). Genetic drift is a random walk (not a method one would want to use to code for a subset of sequences that only occupies 10^100 of sequence space), and selection only works for hill-climbing problems, not needle-in-the-haystack problems like novel stable folding functional protein families determined by physics. Testing Meyer’s claim suggests that he might have a point.

    1. Don't take Durston's paper as seriously as Szostak's or Hazen's. It was published in a low quality journal, not Nature or PNAS.

      Meyer focuses on long modern proteins because looking at short, old proteins or RNA hairpins would kill his argument. Recognition motifs can be functional at 5 AA long. Yes, you can build short functional things randomly, and then select for them.

  8. Comments are very interesting. That's unusual, usually they degrade very quickly! I thought the reviewer was too biased to be fair to the book, but the comments were informative.

  9. Re. David vun Kannon’s comments:


    One short article and two papers and van Kannon figures he doesn’t need to check out one of the papers because the impact factor of the journal was not as high as Nature or PNAS. Good thing he doesn’t work for me. Both Hazen’s and Durston’s papers compliment each other. Hazen’s method is not usable for proteins because M(Ex) is an unknown for protein families but Durston paper uses a more standard version of the Shannon equation that works for proteins.

    Regarding Kannon’s assertion that ‘old’ proteins are short and easy to obtain, I think we can test that belief. I used the modified software mentioned in my last post to test three universal protein families, thought to be among the oldest protein families we have available today and part of the Last Universal Common Ancestor (LUCA) genome. Here are the results:

    RS2: 211aa, 3,605 input sequences, 465 bits of func info, M(Ex)/N=10^-140

    RS12: 122aa, 3,249 input sequences, 346 bits of func info, M(Ex)/N=10^-104

    RecA: 320aa, 6,608 input sequences, 976 bits of func info, M(Ex)/N=10^-294

    Looking at M(Ex)/N for these most ancient protein families, I’d say that Kannon’s assertion looks like it may be falsified. These ancient protein families are not easy to locate in an evolutionary search.

    Kannon’s contention that recognition motifs can be functional at 5aa doesn’t really mean anything. Very short functional sequences are trivially easy to obtain. What we are interested in are sequences that will produce stable, repeatable 3D structures that are functional.

    1. Censored, I agree we might not have the best relationship if we worked together.

      Sorry, I've read all three papers. You are wrong about these proteins being difficult to locate in an evolutionary search, because Durston's made up calculation has little bearing on the problem of evolutionary search.

      "The number of Fits quantifies the degree of algorithmic challenge, in terms of probability, in achieving needed metabolic function."

      No, to calculate that you'd need to examine all possible sequences at each length and test them for degree of functionality. Durston's method of summing the probabilities of each AA in the sequences doesn't come close to that. He'll never examine a possible functional sequence that Nature hasn't already found. (Good thing he's not in charge of drug discovery.) He doesn't look at possible ancestral variants that have been out competed by newer mutations, though that is a key issue.

      Durston is a Christian apologist, not a working scientist. His co-author Abel is a retired veterinarian! No, these guys are not in the same league as Szostak and Hazen.

      "Very short functional sequences are trivially easy to obtain." Well, that's kind of the point here, Censored. Meyer's whole Bignum argument falls down when you start admitting that "functionality" doesn't start at 150aa in length.

      BTW, I saw that this review attracted the attention of the DI, with predictable results. I discuss it on my blog, Invisible Hand. (

  10. vun Kannon,  your comments suggest you are more swayed by authority and personal bigotry than by science. Let us keep the comments focused on the science please.


    First, your classification of Durston’s equations as ‘made up’ suggests that you may benefit from familiarizing yourself with the concept of Shannon information and the more rigorous way of calculating it that uses the summation of the probabilities of the variables. Start with Shannon, C. Part I and II: A mathematical theory of communication. The Bell System Technical Journal XXVII, 379-423 (1948). Both Durston and Hazen use a form of Shannon information filtered for functionality, hardly ‘made up’. Hazen’s equation is a simplified form, as I mentioned earlier, and as you need to grasp. You cannot knock one without knocking the other. Both have their shortcomings, but both are useful so far as they go.


    You are wrong about Durston’s method needing to examine all possible sequences for functionality. It is Hazen’s method that fails here. Using Durston’s (really Shannon’s summation of probabilities approach), if you do not have all the functional sequences (which we do not have), you can calculate the probability distributions of each aa at each site and base your calculation on that. If your input sample is large enough, the probabilities of each aa at each site will level off in a series of horizontal asymptotes. As I mentioned earlier, I have been playing around with Durston’s software, adding what I think are some improvements of my own. If you do this, and plot the probabilities vs. sample size, you will see them well on their way to leveling out by the time a sample size of 500 is used. It is an elegant way around the problem of not knowing M(Ex). It does not give a precise answer, but it does give us a very good estimate, which gets us further than Hazen’s equation does. (Actually, Durston’s method significantly over estimates M(Ex), it is much smaller than what their method suggests, because they assume site independence.)


    Finally, it appears as though you are under the impression that you can build novel stable, repeatable 3D protein structures by assembling a series of, say, 5aa sequences that are functional ‘recognition motifs’. That is so far from reality that ‘it isn’t right, it isn’t even wrong’ (to paraphrase Wolfgang Pauli).

    1. Censored, your comments suggest you don't know how to read the scientific literature or understand the use of impact ratings and citation indexes. To state that Dunston and Abel (repeatedly) channel their articles through a 'we publish anything' journal is merely warning the reader of what most already know. It is like the old Rivista di Biologia.

      Hazen is explicit that all possible configurations need to be tested. That might be inconvenient, but it is correct. Counting the examples of function that you happen to know about is not correct. That was the force of my comment about drug discovery, earlier. Many drugs have the same function as a target, but vastly different composition.

      Your final point is just wrong. If you'd like to know what I think, ask, don't assume. A recognition motif is a specific kind of sequence that binds to RNA, that is its function. I haven't said anything about all proteins being built out of recognition motifs.

      Of course there are highly functional protein motifs such as the alpha helix that are going to score very low on the "functional information" scale, because you can substitute just about any AA into any position. In these cases, function isn't correlated with AA specificity, yet any chain of 4 or more AAs can be "functional" in building a helix.

      As you'll see if you read Hazen, we're talking about more than proteins. We're also talking about RNA. RNA stem loop structures (hairpins) are novel, stable, 3D structures with function. And short, in the range of 20 nucleotides long. Yes, there are short functional biochemicals that can be built by random processes. Get used to it.

  11. Censored wrote: ‘ Hazen is explicit that all possible configurations need to be tested. That might be inconvenient, but it is correct.”
    And it is for precisely that reason that Hazen’s method cannot be used for protein families, which either you don’t seem to grasp, or you refuse to admit. You seem to be unaware that there are an enormous number of possible functional sequences for each protein family. If you were to understand Hazen’s equation, you would begin to understand just how large the search space N is. If you used Durston’s equation, you would start to get an idea of just how large M(Ex) might potentially be. You need to do the science, and not excuse yourself from doing science merely because some religious person published it first.
    I know what a recognition motif is, but we were talking about proteins. Go back and look at the examples I have provided instead of trying to change the topic. Your belief that, since we can generate short functional sequences by a random process, therefore we can generate long ones is naive, to say the least. Science needs to be testable, verifiable and falsifiable. You need to test your belief to see just how woefully inadequate it is. Write yourself a genetic algorithm to search for a functional sequence in some protein family. I am sure your selection function would be most interesting.
    We know that it is trivially easy for intelligence to produce the level of information required to code for biological proteins, say, 400 bits, so it is not the case that we do not know what can do it. Until you can provide a testable method to show how nature can generate, say, 400 bits of functional information, you really only have one empirically verifiable candidate on the table …… intelligence.
    Do the science …. and it seems you have a ways to go before you even understand what the problem is. I would suggest beginning with a graduate course in protein structure. Maybe do some homology modeling where the fit isn’t the greatest. Also, I would also suggest that if you have a problem with a paper, that you show what is wrong with the content of the paper, rather than whine about one of the authors being religious. You have just enough knowledge to give you a false sense of confidence, but too little to actually discuss these things in any depth at all. For that reason, I need to move on to more productive things and leave the last word to you (which I’m sure will be rich indeed).

Leave a Reply