Circularity of using CSI to conclude Design?

At Uncommon Descent, William Dembski’s and Robert Marks’s coauthor Winston Ewert has made a post conceding that using Complex Specified Information to conclude that evolution of an adaptation is improbable is in fact circular. This was argued at UD by “Keith S.” (our own “keiths”) in recent weeks. It was long asserted by various people here, and was argued in posts here by Elizabeth Liddle in her “Belling the Cat” and “EleP(T|H)ant in the room” series of posts (here, here, and here). I had posted at Panda’s Thumb on the same issue.

Here is a bit of what Ewert posted at UD:

CSI and Specified complexity do not help in any way to establish that the evolution of the bacterial flagellum is improbable. Rather, the only way to establish that the bacterial flagellum exhibits CSI is to first show that it was improbable. Any attempt to use CSI to establish the improbability of evolution is deeply fallacious.

I have put up this post so that keiths and others can discuss what Ewert conceded. I urge people to read his post carefully. There are still aspects of it that I am not sure I understand. What for example is the practical distinction between showing that evolution is very improbable and showing that it is impossible? Ewert seems to think that CSI has a role to play there.

Having this concession from Ewert may surprise Denyse O’Leary (“News” at UD) and UD’s head honcho Barry Arrington. Both of them have declared that a big problem for evolution is the observation of CSI. Here is Barry in 2011 (here):

All it would take is even one instance of CSI or IC being observed to arise through chance or mechanical necessity or a combination of the two. Such an observation would blow the ID project out of the water.

Ewert is conceding that one does not first find CSI and then conclude from this that evolution is improbable. Barry and Denyse O’Leary said the opposite — that having observed CSI, one could conclude that evolution was improbable.

The discussion of Ewert’s post at UD is interesting, but maybe we can have some useful discussion here too.

210 thoughts on “Circularity of using CSI to conclude Design?”

  1. keithskeiths


    If we take before and after snapshots then if *INSERT CURRENT ITEM HERE* can be measured by IDists then I’d expect them to be falling over themselves there to do it. It proves their case!

    Exactly. They’d be demonstrating it every chance they got, instead of sulking and avoiding our questions. They’d be saying “I’m glad you asked” and “Let me show you.”

    Instead we get these peevish responses from HeKS, complaining about our questions and referring us to old comments that don’t answer them. HeKS clearly has no confidence in alt-CSI or in his abiilty to use it for making design inferences.

    And for good reason. His proposed method for quantifying it — adding up the number of base pairs, multiplying by 2, and then throwing in some extra bits for methylation and whatnot — will lead him straight into trouble, because he won’t be able to demonstrate that 500 bits (or whatever threshold he uses) is beyond the reach of evolution.

    And amusingly, any attempt to do so would require him to revert to Dembski-style P(T|H) calculations — which no one can do anyway for the evolution of real biological structures in real-life fitness landscapes.

    So barring a miracle, alt-CSI will go on the junk heap with the other CSI variants.

    One thing I’ll give HeKS credit for: the name “alt-CSI” is less of a monstrosity than KF’s “FSCO/I” or gpuccio’s “dFSCI”.

  2. Joe FelsensteinJoe Felsenstein Post author

    If most of the other commenters here are correct, the alt-CSI argument amounts to saying that (1) this looks like Design to my intuition, (2) I can’t imagine how evolutionary processes such as natural selection can bring this about, (3) it’s up to evolutionary biologists to provide a detailed, step-by-step explanation of it, and (4) ahai! they have not provided the detailed explanation for this case! Therefore Design.

    If that’s what it amounts to, it isn’t really a cousin of William Dembski’s argument, nor of Michael Behe’s argument.

    I will give HeKS credit on one point, though. HeKS has argued that most of the regulars at Uncommon Descent are actually making that alt-CSI argument rather than the more technically demanding Dembski arguments. I think HeKS is quite right about that.

  3. PatrickPatrick


    Hi Patrick,

    Sorry for the extreme delay but this is the first chance I’ve had to write a response to you.

    I’m glad to see you back and willing to continue the discussion. You’ve written a very detailed response. For the sake of clarity I’m going to try to break it into separate topics because I don’t want to inadvertantly quote mine you by leaving out context. The first topic I’ll address is Dembski’s CSI calculation and its implications.

    Thank you, I think I understand your argument now. Essentially you are summarizing Dembski’s math in English, focusing on the structure of his conditional statement. If my understanding is accurate, you are saying that Dembski’s equation for CSI means:

    “If known evolutionary mechanisms are not sufficient to explain some biological artifact (that is, P(T|H) is small) then it is reasonable to conclude that artifice (design) is an explanation.”

    Is this an accurate restatement of your claim?

    Thank you for trying to accurately understand what I’m saying. I recently took issue with one of your TSZ compatriots for seeming to intentionally do the exact opposite.

    In any case, you have very nearly understood what I’m saying, so let me reuse what you wrote and simply change/add a few words:

    “If known evolutionary or other relevant purely naturalistic mechanisms are not sufficient to explain some biological artifact (that is, P(T|H) is small) then it is reasonable to tentatively infer that artifice (design) is the best causal explanation based on the current state of our scientific knowledge.”

    There are only a few changes there, but they are not insignificant ones, as in some cases they speak to the points you subsequently raised.

    I don’t think our two formulations are inconsistent, but I’m happy to use your version that you are willing to support.

    You are making a fundamental error in your reasoning, one that I have seen other ID proponents make on several occasions. Specifically, you are leaping to the conclusion, however tentative, of “design” without any evidence whatsoever. In the situation where “known evolutionary or other relevant purely naturalistic mechanisms are not sufficient to explain some biological artifact” the only rational statement we can make is “We currently don’t know.” If you want to conclude “design”, you need evidence for design, not just lack of evidence for another explanation.

    “Design” (which, let’s be honest, means creation ex nihilo by the Christian god, to most ID proponents) is not a reasonable default assumption. Without positive evidence for the existence of a designer, and of when, where, and how that putative designer implemented his, her, its, or their design, defaulting to “design” is simply a god of the gaps argument: “Science can’t (currently) explain this, therefore god.”

    Further, rarely is there a complete lack of hypotheses about the provenance of a particular artifact. Even if specifics aren’t known or one hypothesis isn’t significantly better supported than another, investigators have leads to pursue. Historically this has always led to better understanding of natural phenomena. Never has it led to identification of a designer of biological systems. That’s not to say it couldn’t, but it’s not the way to bet.

  4. PatrickPatrick


    The next topic I’d like to address is ID as religious apologetics.

    I understand that you think “ID is nothing more than religious apologetics” but I simply don’t agree. I see it as proceeding from a different (and I think better) philosophy of science, which says that scientific knowledge should not forever be held captive to a priori philosophical commitments. If design is, at present, a uniquely adequate causal explanation for certain features of life, as some think it is, why should people be prevented (possibly on pain of losing their careers) from making an inference to design as the currently best explanation? If people are unconvinced by naturalistic just-so stories, why should they be forever bound to wait until the ‘correct’ naturalistic explanation is found? What if it turns out that there is no correct naturalistic explanation and design really is the right one? A refusal to ‘allow’ it as an acceptable causal explanation on philosophical and methodological grounds would permanently prevent us from recognizing design as a scientific truth and would instead consign us to eternal intellectual darkness, demanding that we refrain from making a design inference on the basis that maybe, just maybe, there might be some unknown naturalistic process out there that can satisfactorily explain (rather than simply explain away) effects that appear to be the product of design … and with no expiration date on how long we are required to wait.

    (As an aside, despite the claims of the faux documentary Expelled, people have not been fired for supporting ID. Some ID proponents have had their contracts not renewed or have been denied tenure because they failed to meet the requirements for their positions. If you have evidence otherwise, I’d be interested in seeing it.)

    You are mistaken about the nature of scientific investigation. There are no “a priori philosophical commitments” in the scientific method except, arguably, methodological naturalism. That doesn’t mean that science can’t investigate the supernatural (which again, being honest, is what ID purports), it simply means there must be objective, empirical evidence to support any claim and that hypotheses must have testable entailments.

    ID has neither evidence nor testable entailments. It is not a scientific endeavor in any sense of the word. It is certainly far from “a uniquely adequate causal explanation for certain features of life” since it is not adequate to explain anything at all. In practice, ID is nothing more than religious apologetics, recognizable by anyone who has read Paley. As further evidence of this, I’m copying a paragraph from later in your comment:

    Finally, the claim that there is no evidence that any such entity exists is also a little odd. . . . There is plenty of evidence and argument for the existence of some highly intelligent and powerful being, ranging from the realms of chemistry to cosmology. That those who don’t want to accept the conclusion of such a being’s existence try to undercut or explain away the evidence by reference to unsupported assumptions, unwarranted extrapolations, or untestable flights of fancy does nothing to change the facts on the ground.

    Clearly you are referring to the standard arguments that attempt to demonstrate the existence of a deity. I’m going to be unapologetically blunt: There is no evidence for your god or any other gods. People have been creating gods for tens of thousands of years but none have ever been substantiated by evidence and arguments that are convincing to any but those who already believe.

    The vast majority of ID proponents support it because it provides a veneer of pseudo-scientific respectability over what is, at its core, simple religious faith. That is why it is clear to any external observer that ID is apologetics, not science.

    There is nothing wrong with religious apologetics, per se, in their place. That place is not, however, in public school science classes.

  5. PatrickPatrick


    The third topic I’d like to address is who, what, when, where, and how (why is optional).

    It seems that the real problem anti-ID people often have is with the technique part. How exactly would the designer have done it? How did he or she get the right parts in the right places and in what order?

    That’s not the problem I see with ID. One of many problems with ID is that there is no evidence for any entity capable of doing what is claimed. There is no agreement on what this supposed entity did or when it did it. There is no evidence or theory for the capabilities of this unevidenced designer. There is no scientific theory of ID that has testable entailments.

    Well, apart from the case of Young Earth Creationists, you’re probably not going to get all that many disputes over when the designer actually did stuff. In most cases an ID proponent will think the designer did something relatively shortly before it showed up in the historical/fossil record, so there won’t be all that much dispute with otherwise accepted timelines.

    In my experience, ID proponents are remarkably reluctant to make explicit claims of when the designer intervened. I would be very interested in the results of a survey at UD on this topic. I suspect you won’t find the consensus you expect.

    Based on what you’ve written, though, am I correct in concluding that you believe that the designer is at work continuously, guiding evolution to his, her, its, or their ends?

    Also, what this entity would have done doesn’t seem like much of a mystery at the macro scale, so I have to wonder if you really are referring to the technique/methodology issue that you said wasn’t your problem with ID.

    So, what do you believe the designer has done, specifically? It is currently a mystery to me at least.

    When you have described that, please explain what evidence you have for your claims and how they might be tested.

    Also, to say that there’s no theory for the capabilities of the designer is a little odd. There are no scientific theories for the full complement of the designers ultimate abilities, but basic logic would entail that the designer possesses the attributes and abilities necessary to produce the effects in question.

    Saying that the designer could do whatever the designer did is tautological. As with the “when” question, ID proponents in my experience shy away from describing the capabilities of the designer. This is, of course, because they consider the designer to be their omnipotent god.

    The problem with an omnipotent being is that it can be used to explain anything we observe, which means it explains nothing.

    We typically observe the attributes and abilities that need to be present in humans for them to produce comparable but significantly less sophisticated effects. One can reasonably infer that the designer would need to have similar attributes and abilities but on a significantly more advanced scale.

    Are you limiting your designer to human capabilities? If not, are you suggesting any limitations on the capabilities of your designer? What is your evidence for the existence of the being or beings with the capabilities you posit?

    Who, what, when, where, and how are essential questions that must be answered for ID to be considered a scientific hypothesis. The that no ID proponent has ever addressed those questions is yet another reason why ID is properly seen as nothing more than religious apologetics.

  6. HeKS

    Hi Patrick,

    I have to say, while I disagree with your responses and assessment of some of my prior comments, I very much appreciate that you seem to be trying to actually understand me and take my comments in their intended context. In some cases I think your replies tend to ignore comments made elsewhere in the post you’re responding to that would address some of what you’re saying, but I find your general approach honest enough that I don’t think these are cases of you intentionally ignoring my comments so that you can misrepresent them. I have some replies to direct at the other people, which I’ll hopefully have a chance to do tonight, and then I’ll address your comments if I have time. As I’ve said in some other posts here, I’m on the verge of starting the next phase of a programming project I’ve been working on over the past year or so and it may start as early as tomorrow, but I will eventually get to your comments, whether it’s in the next few days or a couple of months down the road like it was this past time.

    Take care,

  7. keithskeiths


    If most of the other commenters here are correct, the alt-CSI argument amounts to saying that (1) this looks like Design to my intuition, (2) I can’t imagine how evolutionary processes such as natural selection can bring this about, (3) it’s up to evolutionary biologists to provide a detailed, step-by-step explanation of it, and (4) ahai! they have not provided the detailed explanation for this case! Therefore Design.

    HeKS’s position has changed quite a bit during this thread, but as it currently stands, he appears to see alt-CSI as having two components: a complexity component and an information component.

    The complexity component is the “I know it when I see it” part, for which there is no calculation and no comparison against a threshold:

    In Alt-CSI, the C is primarily a descriptor of an observable attribute of some effect, system, block of code, etc., rather than a specific measure. As such, there’s is no need for an additional numerical threshold of complexity over and above the required information threshold.

    The information component is the “take the number of base pairs, multiply by two, add some bits for methylation and other epigenetic factors, and compare to 500 bits” component:

    Now, when it comes to calculating the information of Alt-CSI (since, you know, we’re talking about Complex Specified Information), I already gave an example of how you would go about doing that for some molecular machine that should be none too surprising. As for the informational threshold required of Alt-CSI to infer design, I’ve never heard it as being any different than the Dembski-CSI threshold of 500 bits. In both cases you’re talking about information that is specified to fulfill a useful function.


    I would say that when it comes to something like molecular machines, we can at least begin by looking to the genetic code that specifies their parts lists. There would be approximately 2 bits of genetic information per base pair of nucleotides (1 member of the pair specifies the other but there are 4 options to choose from). We may be able to add another bit per base pair signaling methylated or non-methylated where appropriate (and there are other types of epigenetic information, but we’ll ignore them for this purpose). On this basis one could determine the number of base pairs that impact the folding of the individual protein parts and their ability to do their job. This would give a low-ball measure of information in bits, since there’s also the matter of assembly instructions for molecular machines, which would likely include further epigenetic information (and perhaps further genetic information) as well.

    Hopefully he will explain to us why he’s using Dembski’s 500-bit probabilistic threshold when his alt-CSI information number scales linearly with the number of base pairs. It doesn’t make much sense to me.

  8. Joe FelsensteinJoe Felsenstein Post author

    Unless there is also some argument as to why natural evolutionary processes, including natural selection, cannot put into the genome 500 bits of whatever information is in the SI part of CSI, there is not much use computing it.

    As far as I know, substitution of alleles can produce SI, where SI is defined the way Dembski did in his 2001 book. I discussed a simple example of this in my 2007 article critiquing Dembski’s arguments. I even made a post here at TSZ in 2012 showing an example of a simple scheme of natural selection which puts SI into the genome.

    I hope HeKS will explain here what is wrong with those examples.

  9. keithskeiths


    Unless there is also some argument as to why natural evolutionary processes, including natural selection, cannot put into the genome 500 bits of whatever information is in the SI part of CSI, there is not much use computing it.

    It’s the perennial dilemma for those pushing CSI and its variants. Either

    1) you base it on P(T|H), in which case

    a) no one can calculate it for actual biological systems, and
    b) using it to infer design would be circular, even if you could calculate it; or

    2) you base it on a simplistic calculation like HeKS’s, in which case it says nothing about evolvability.

    It must be disheartening to try so hard to defend an irrational belief, only to fail utterly.

  10. PatrickPatrick


    I look forward to your response when you have the time. I appreciate that it can be difficult being the only ID proponent currently commenting here who is polite and participating in good faith.

    It would cut through a great deal of the back-and-forth if you could clearly and concisely state a scientific hypothesis of ID, a summary of the evidence that supports that hypothesis (not simply arguments against modern evolutionary theory), and some testable entailments that could potentially disprove the hypothesis. If ID really is scientific and really is detectable, these are the minimum required to establish that.

    In the meantime, I’d like to make one more response to your long comment, this time on the topic of the capabilities of known evolutionary mechanisms.

    On the other hand, we also know that we have never observed any natural process or mechanism actually producing any complex (in the sense of many well-matched parts), functionally specified system at all, much less one that is even at the level of sophistication that humans are currently able to produce.

    We don’t know that. In fact, we know the opposite. Known evolutionary mechanisms are capable of producing new functionality. If you provide a precise definition of “functionally specified information” I suspect I can provide examples of those mechanisms creating it.

    First, I referred to a complex, functionally specified system, not just functionally specified information. However, even when referring just to the latter, the term “complex” cannot be left off, though I will again use it in the sense of ‘many parts’. When asking whether evolutionary mechanisms are capable of adding functionally specified information to the genome, we would have to answer yes, in a sense. For example, if some existing functionality has been knocked out with a simple point mutation, it’s possible for a chance mutation to revert the code back to its original form and thereby restore the prior functionality. In doing this it has, in some very limited sense, added functional information. Such events are fortuitous, but not all that surprising or impressive. It’s kind of like having a pre-built machine with electricity supplied to it and all that is needed for system functionality to arise is for some final connection to be made between two points on a circuit board. If you start flipping pennies in the direction of that board until one happens to bounce into the right spot and complete the connection, system functionality will ensue, but the fortuitous fall of the penny conferring functionality is not what needs to be accounted for. It’s the fact that everything else was in the right place, all the necessary parts being present and each part being correctly related to the others, such that only one final step was needed for functionality to arise.

    Of course, there are other means by which functional information might be added to the genome, such as in the case of gene duplication, but in terms of information creation this is tantamount to accidentally hitting CTRL+V twice when you’re trying to paste something. What needs to be explained is not that more content is added to the document in the second paste action, but how the functionally-specified information that was originally copied arose in the first place; not to mention the hardware and software that allows for that information to be copied and pasted at all. The duplication does not constitute the de novo creation of complex, functionally-specified genetic information.

    There are two issues here. First, gene duplication can result in changes to the behavior of an organism, for example by producing more of a certain protein. More importantly, gene duplications provide an opportunity for mutations to take place without impairing the existing functionality of an organism. This has long been known to be a common and important source of raw material for selection to operate on. In fact, the nylon-eating bacteria found in Japan appear to have developed their unique functionality by exactly this mechanism. If your calculations of the probability of a particular biological artifact arising by “natural” means don’t take this mechanism into account, you’re going to have a lot of false positives for “design”.

    Second, you are moving the goalposts when you switch to discussing “how the functionally-specified information that was originally copied arose in the first place….” If a gene duplication followed by mutation can create CSI or FSI or CFSI or whatever your metric is, then we know that your metric is not a reliable indicator of design.

    I see that other participants here are discussing your “alt-CSI”, so I won’t pile on. I’ll simply say that unless and until you provide an objective way to measure whatever it is you’re measuring, you can’t say with any certainty that any mechanism can or cannot create it.

Leave a Reply