Logical fallacy of Behe’s “IC means ID” notion


By Mark Perakh


Posted on February 20, 2005



Michael Behe’s book Darwin’s Black Box (Behe 1996) is one of the most popular and extensively reviewed books propagandizing intelligent design (ID) “theory.” The concept of “irreducible complexity” (IC) introduced in that book has been touted by Behe and other ID advocates as a great discovery and used as one of the main tools in their efforts to “destroy Darwinism” (the goal openly announced by such ID’s “leading lights” as Phillip Johnson (Johnson 1991) and Jonathan Wells (Wells 2002)).

            The concept of IC has been critically discussed in minute detail by experts in biology and found to be unsubstantiated --  see, for example, publications by H. Allen Orr (1996-97), Russell F. Doolittle (1997), David Ussery (1999, 2004), Kenneth R. Miller (1999), Gert Korthof (1999), Matt Inlay (2002), Pete Dunkelberg (2003), and others. The prevalent attitude of professional biologists to Behe’s IC concept is perhaps most unequivocally expressed by Kenneth Miller’s words: “… the notion of irreducible complexity is nonsense.” (1999, p. 150).

            While the critical analysis of the IC concept by professional biologists seems to be sufficient to dismiss Behe’s alleged great discovery in biology, there is another aspect to IC which has received less attention, although it, to my mind, makes the entire set of ideas bridging IC to ID so weak that the “IC leads to ID” concept sounds implausible even without discussing the biological facets of IC. I submit that, regardless of its merits or drawbacks from the viewpoint of biology, the very idea of IC as a marker of ID is illogical.

            A concept identical in all but name to Behe’s irreducible complexity was around for a long time before Behe’s book hit the bookstores. It was applied to the problems of evolution of various anatomical structures, such as the mammalian eye (recall the endlessly asked and as many times answered question, “what good is half an eye?”), or the snakes’ apparatus of venom injection (Marcell 1976) etc. A practically identical concept (“interlocking complexity”) was discussed from the standpoint of genetic already nearly 80 years earlier (Muller 1918, 1939). Even the application of the IC concept to the molecular assemblies within a biological cell (which is Behe’s playing field) was put into circulation some ten years before Behe (Cairns-Smith 1986).     

            Unlike Behe, Cairns-Smith did not claim to have made a grandiose discovery and did not suggest that the “irreducible complexity” (the term he did not use) points to the impossibility of cell’s evolution.  Of course, the absence of references to predecessors in Behe’s book, although contrary to traditions of a scientific discourse, is a minor point (especially because Behe’s book is not a scientific monograph but more of a popular or semi-popular tale obviously designed for a wide lay audience).


            The critical discussion of Behe’s ideas has mainly concentrated on three aspects of IC, to wit:

(1)   The first aspect of IC subjected to discussion has not been about the biological side of IC; it has been, rather, about the very definition of IC.  Critics have pointed out the logical flaws in Behe’s original definition which made uncertain how exactly a system can be judged to be IC. To my knowledge, Behe himself has never acknowledged that his definition was in any way imperfect. However, Behe’s colleague William Dembski (viewed by the ID advocates as their leading logician) admitted that Behe’s idea of IC was “neither exactly correct nor wrong” (Dembski 2002, p. 280).  Dembski suggested various modifications of Behe’s definition (Dembski 2002, 2005) which, however, did not seem to really “salvage” that definition; the most recent variations of these modifications seem to have confused the matter even further (RBH 2005, Perakh 2005).


(2)   The second aspect of IC subjected to critique has been somehow connected to the first one but mainly dealt with the more specific problem of whether molecular systems offered by Behe as examples of IC are indeed IC. A number of biologists pointed out that systems such as bacterial cilia or blood-clotting cascade which, according to Behe, exemplify IC, are in fact reducible without losing their “basic function.” (See, for example, Miller 1999).


      It seems clear that aspects (1) and (2), while relevant to the problem of the validity of the IC concept, are nevertheless not crucial.  If a definition of a concept is logically flawed, this is a drawback from both the viewpoint of overall logical consistency of a theory, and of its practical application, but by itself it does not necessarily invalidate a theory. Even without a logically impeccable definition of its main concepts, a theory still can be useful and plausible. Likewise, even if the systems Behe offers as examples of IC can be shown to be reducible, this in itself, while devaluing the examples, does not necessarily devalue the theory since, perhaps, better examples can be found. The very existence of systems which need all of their parts to be functional hardly invokes doubts.


(3)   The third aspect of IC subjected to critique is indeed of crucial importance. Behe asserts that IC systems (exemplified by the protein assemblies in biological cells) cannot have evolved via a direct “Darwinian” path because such a path necessarily goes through a sequence of intermediates each performing the same “basic” function. Since any system comprising fewer parts than the IC system in question, is, by definition of IC, dysfunctional, it could not be an evolutionary precursor of an IC system, or so says Behe.  Regarding evolution of IC system via an indirect evolutionary path, Behe admits that such process is possible but, in his opinion, so highly improbable that it cannot be considered a feasible option.


      The last point has been strongly disputed by those Behe’s opponents who are professional biologists. They suggested detailed scenarios showing how, for example, a bacterial flagellum could have evolved from evolutionary precursors with a sufficiently high likelihood (Matzke 2003, Ussery 2004, Musgrave 2004).

      Being not a biologist, I don’t feel qualified to discuss in detail the scenarios in question, but the overwhelming consensus of the vast majority of professional biologists seems to favor the views of Behe’s opponents.  Except for vague protestation from Behe wherein he and his supporters demand from their opponents  highly detailed proofs of the factual occurrence of indirect evolutionary paths leading to IC systems, Behe seems to me unable to offer substantive arguments against the consensus of the majority of biologists.

      Still, however realistic and likely the scenarios for a “Darwinian” evolution of flagella, of the immunity system, etc., may be, the ID advocates will most probably demand ever more details of the suggested evolutionary process, never conceding that a satisfactory demonstration of a possible evolutionary path has been offered. Moreover, even if an ultimately detailed and obviously realistic scenario can be suggested for the evolution of, say, the flagella or of the immunity system, the adherents of ID most probably will simply switch to another example of an IC system which supposedly could not have evolved, and demand again detailed proof of its evolutionary history. This portends an open-end game. 


      In this essay I will analyze the IC concept from a viewpoint different from the three aspects of the problem discussed above. I intend to show that even if the IC concept is valid, and even if many biological systems are indeed IC, this in itself does not logically lead to design inference. In fact, my contention is that IC in itself can more reasonably be construed as an argument against design inference.


            In an essay titled “Irreducible Contradiction” posted to the internet in 1999 (see reference) I suggested critical comments to Behe’s Darwin’s Black Box. This essay was translated and printed in Russia (Perakh 2001a) and in Israel (Perakh 2001b).  After its appearance in Kontinent the essay was reproduced on several Russian websites and invoked a discussion which sporadically continues even now (February 2005). By the end of 2003 my book Unintelligent Design was published wherein chapter 2 was essentially a slightly modified version of the same essay (Perakh 2004).  Recently that chapter was translated into Polish and appeared in the Polish journal Filozoficzne Aspekty Genezy (Philosophical Aspects of Origin) – see .

            In the nearly six years since the appearance on the internet of the essay in question, Michael Behe, whose ideas about irreducible complexity were critiqued in my essay, has never uttered a word acknowledging the existence of my critical remarks.  Nor has William Dembski, who has actively promoted Behe’s irreducible complexity  concept, ever mentioned my critical comments. Neither did anybody else from the intelligent design camp.

            Recently Dembski posted an article titled “Irreducible Complexity Revisited,” (Dembski 2005) which has initiated some discussion (RBH 2005). In particular, I placed a post both on the Panda’s Thumb (PT) blog and on Talk Reason website, where Dembski’s most recent interpretation of irreducible complexity was briefly discussed (Perakh 2005). Why, then, am I writing this essay?

            The reason I am writing once again about IC is that in some recent threads on PT Behe’s IC concept has emerged as a target of debates once again. Moreover, an article by Behe appeared recently in New York Times, wherein Behe repeats again the same much critiqued notions without any signs of having changed his position to any extent or of having accounted for a single point suggested by his critics. I think therefore it is worthwhile to revisit certain points which seem to be in need of clarification, regarding the IC concept and its alleged logical segueing into ID.

            Since I am not a biochemist, I do not try discussing Behe’s IC from the viewpoint of biological details – this seems to be taken care of by the aforementioned experts. There are, though, in my opinion, certain logical faults in Behe’s arguments which can be discussed without delving in the intricacies of the biochemical side of his discourse. Since my discussion of these logical fallacies (to which I pointed in my essay and in my book) has not only been not answered by Behe or by his colleagues, but also largely left out of discussion by IC opponents who (justifiably) concentrated instead on the biological aspects of the matter, it looks like another shot may be not excessive.

            I’ll not discuss here Dembski’s recent modifications of the IC concept (addressed in Perakh 2005 and RBH 2005).  Instead, I will refer to Behe’s original definition of IC, which, albeit suffering from a number of deficiencies (as admitted by Dembski 2002), does essentially reflect his principal idea whereas Dembski’s subsequent modifications, rather than “salvaging” Behe’s original definition (as Dembski asserted to be the case) in fact have only added unnecessary confusion. (For example, Dembski added to Behe’s definition the concept of an “irreducible core” of the system. Whatever Dembski chooses as the “irreducible core” always can simply be construed as the system in question itself; indeed, Behe has never defined where exactly the boundaries of a system are; so Dembski’s modification adds little essential to Behe’s concept).

            The essence of Behe’s original IC concept is as follows:

            A system is IC if:

            (a) It consists of several parts;

            (b) The parts are “well matched.” (Behe offered no definition of the notion of being “well-matched.”)

            (c) It performs a certain “basic” function (for example, clots blood);

            (d) It ceases to function if even a single part is missing.


            Having discussed several examples of protein “machines” in biological cells, which, according to Behe, are IC (although this assertion has been disputed by the biologists listed above who critiqued Behe’s idea), Behe then asserts that the existence of IC systems in a biological cell points to them being designed rather than having emerged as a result of evolution. I will discuss this conclusion and I intend to show that it, in my view, contradicts logic.

            Note that Behe’s concept of IC comprises two components: one is complexity and the other is irreducibility.

            Indeed, Behe expends a lot of effort to demonstrate how staggeringly complex the protein systems in a cell are. Although the concept of IC as defined by Behe is not limited just to very complex systems (indeed, Berlinski [1998] and Dembski [2005] apply it to systems as simple as a chair or a three-legged stool), Behe emphasizes the enormous complexity of the protein “machines” in a cell.  It is evident that for Behe the complexity in question is part of his idea, pointing to design as the alternative to evolution.

            Hence, according to Behe, biological systems must have been designed because they (A) are very complex; and (B) cannot function unless all of their parts are present.

            Regarding (A) – complexity – note that Behe has not provided a definition of complexity. Several such definitions have been suggested, though, by Dembski.

            As has been pointed out before (Perakh 2004), Dembski’s various definitions of complexity are often incompatible with each other.  There is among them, though, a definition repeated by Dembski many times, which is in tune with Behe’s point (A). According to that definition complexity is equivalent to small probability (Dembski 1998). For example, Dembski asserts in his book that “probability measures are disguised complexity measures” (page 114). Variations of this assertion are scattered over Dembski’s books. Thus the more complex a system, the less probable its spontaneous emergence as a result of chance, or so says Dembski.

            As I have argued (2004), this definition of complexity is contrary to facts. At this time, though, I am discussing not the validity of Dembski’s definition – my task is to discuss Behe’s concept of IC in its logical connection to ID, and for this end it is sufficient to keep in mind that Behe’s IC concept contains as a part the concept of complexity, the latter interpreted as a disguised probability, as rendered by Dembski. The more complex a system, the less probable its spontaneous emergence by chance – hence it is more probable that it was designed – this is the essence of point (A) in Behe’s concept.

            To summarize, Behe’s concept of IC includes the notion of a system’s complexity as a marker of design.


            Point (B) – irreducibility – in Behe’s concept of IC asserts that an IC system loses its function if even a single part is missing. 

            According to Behe, protein “machines” in a cell meet both requirements for being IC – they are very complex and they are irreducible. 

            As it looks to me, even though there are many simple systems which are irreducible in the sense of losing their function if a part is missing, the assertion about the complexity of protein assemblies in a cell seems to be correct: the protein “machines” as described by Behe are indeed very complex.

            As to the second assertion (about irreducibility), it is less certain. Perhaps some of these protein “machines” are indeed irreducibly complex as per Behe’s definition, while many others, including those chosen by Behe as his examples,  are in fact not (as shown by Miller, Doolittle, Dunkelberg and others - see references). However, in this essay I am not interested in the question of whether or not the two above points asserted by Behe are true. My goal now is to discuss: what if these assertions are true? Do they lead logically to the design inference?

            According to Behe (as well as to Dembski and to other ID advocates), both complexity and irreducibility (if they are indeed present) point to design.

            I submit that such a conclusion is illogical. Here is why.


            Start with complexity.

            As I have argued before (Perakh 2004), contrary to Dembski’s persistent assertions, complexity is certainly not just disguised improbability. Examples to the contrary abound. Imagine a pile of stones. Each stone has some irregular shape that resulted from a series of chance events. Among these irregularly shaped stones we find a perfectly rectangular brick or, say, a perfectly spherical stone. These two pieces both have very simple shapes which can be described by very short (i.e. simple) programs.  On the other hand each of the irregularly shaped stones can be described only by a much more complex program containing many numbers.  However, the probabilities of a rectangular brick or a spherical stone being a result of chance are very low: these pieces of a highly regular shape are reasonably (with a high probability) assumed to be products of design. For irregularly shaped stones the opposite is true – the probability of their origin in chance is much larger than in design. Here the relationship between probability and complexity is opposite that prescribed by Dembski’s definition (but compatible with the definition of Kolmogorov complexity – see, for example, Chaitin 2003).

            In this example simplicity rather than complexity is a marker of design. The brick (or spherical stone) which is a simple item is most probably a product of design while stones of more complex irregular shapes are reasonably attributed to chance. I submit that the described example shows not only that Dembski’s definition of complexity fails for certain situations but also that, generally, a more reasonable statement is that simplicity points to design while complexity as such points to chance (more about this in Perakh 2004).

            If this is so, then the first part of Behe’s IC concept – complexity - is more reasonably construed as a partial indication of “blind” evolution rather than of design.


            Now turn to the second part of Behe’s IC – irreducibility. Recall that Behe’s idea is that losing a single part of a protein “machine” makes it non-operational. Therefore, says Behe, such a “machine” could not have evolved via a “Darwinian” evolutionary process.  For such a process to work, a system must have a functional precursor which, modified by a random mutation, is, say, subjected to natural selection.  Since any protein assembly differing from the IC system even by a single part is, by definition, dysfunctional, it could not be a target of natural selection. Therefore, says Behe, an IC system could only have been “created” as a whole. 


            The simple fact is, though, that if an IC system has been designed, we deal with a bad design.  If the loss of a single part destroys the system’s function, such a system is unreliable and therefore, if it is designed, the design is very poor. When engineers design machines, bridges, skyscrapers, TV sets, or artificial kidneys, they always try to envision what can go wrong with their design and how to ensure that small defects do not result in a failure of their product; to this end they build in certain redundancies so that in case some part of the construction fails, its function will not be completely lost but rather taken over by certain self-compensatory features.

            IC systems, by definition, are highly vulnerable to accidental damage (which can never be excluded).

            IC systems, if they are designed, are very poorly designed.

            It must be stressed that in this case we go beyond the problem of suboptimal design. When we deal just with suboptimal design as such, the ID advocates suggest various arguments supposedly justifying the reasons for design being not optimal. For example, one such argument is that we simply don’t know anything about the designer’s reasons to behave as he does; perhaps whatever from human limited standpoint looks like suboptimal design has good reasons beyond our comprehension to be as it is, etc. 

Such an argument usually (albeit not always explicitly) presumes that suboptimality is a “side effect” rather than a deliberately chosen goal of the designer.

            Whether such arguments are convincing depends on the mindset of the particular persons.  For this discourse, however, such an argument is not really relevant. Indeed, Behe’s concept contains as a crucial part the assumption that the irreducibility of biological system is a marker of design.  Such an assumption is obviously beyond the problem of suboptimal design. It is not about a designer who has failed to provide an optimal solution or compromised in his design for some unknown reason. It is no longer  about some “side effect” which the designer has simply failed to correct or has kept for unknown extraneous to the design reasons although it is not a necessary part of the design.  

            Behe’s concept assumes that the very feature which makes the design bad means the system has been designed. In other words, Behe’s concept means that suboptimality is  viewed not as just an unfortunate oversight by the designer; nor is it viewed as something that, albeit seemingly detrimental for the designed entity, has some reasons known only to the designer but unfathomable to us. No, in Behe’s concept the very suboptimality is suggested as a marker of design: an IC system by definition is easily destroyed by damaging just a single part, so a system’s being IC means that its vulnerability is its ineliminable feature.  Behe’s idea implies that the system is IC (and hence suboptimal) because such was the goal of the designer.  “The system is suboptimal, therefore it is a product of design” – that is what Behe’s concept entails.

            ID advocates are welcome to accuse me of offering a caricature of their idea, but it cannot be helped when a concept’s essence sounds like a caricature or a parody; the idea that “IC means ID” can most succinctly be rendered by a maxim: stupid, therefore designed.

            If this is a satisfying logic, I don’t know what a lack of logic is.


            Remember also that Behe’s design inference is based not on some positive evidence but rather on a negative assertion:  IC systems could not have evolved via a “Darwinian” path. Since such a path is impossible, concludes Behe, the only remaining option is design.

            This is an eliminative argument of  the “either-or” type.  I will not discuss here whether or not there indeed are only two mutually-exclusive options. My point is different: if Behe infers design only because the direct evolutionary path, in his view, is impossible and an indirect evolutionary path is improbable, then, to be consistent, he should use the same probabilistic criteria for judging whether or not it is reasonable to assume that the feature which makes design bad is a marker of design.  How probable is it that the putative designer deliberately designs his products to be IC if this means the product will be unreliable?


            Dembski asserts that ID does not imply a smart designer. Designer can even be stupid, says Dembski (Dembski 2001).  However, from many other utterances of ID advocates, including Dembski, it is clear that all such statements are just a smoke screen and in fact they believe that their “designer” is the God of the Bible. (See, for example, Dembski 1999, part 3, or Johnson 2000.) This designer is supposed to be omnipotent and omni-benevolent. If Dembski and Co. are serious in assuming that the designer is some entity different from the God of the Bible, please say this in plain words: “because of design’s faults, the designer most likely is not the God of the Bible, but is some entity that is either of limited intelligence or is rather malevolent.”

            They will not say so. In fact, ID advocates want to have their cake and to eat it too. On the one hand they concede that the putative designer may even be stupid – this they say when trying to explain suboptimality of design.  On the other hand they speak about Christian values, cultural war, the Logos of John’s gospel and the imminent triumph of ID over “materialistic” science. (Dembski 1999, part 3; Johnson 2000). It is not by accident that the leading young earth creationist, Henry Morris, who is much more consistent and straightforward in his frank biblical literalism, referred to Dembski’s contortions regarding the nature of the designer as “nonsense.” (Morris 2005).

            How probable is that the very features that make design bad are markers of design (as follows from Behe’s discourse)? It is hardly less improbable than the evolution of protein assemblies via indirect “Darwinian” paths.

            If Behe infers design just because evolution of protein assemblies via indirect “Darwinian” paths looks improbable to him, design inference also has to be excluded because of the improbability of the putative designer’s deliberately incorporating in the protein assemblies the very features (like IC) which make the design bad.

            In fact, neither design inference from IC nor its denial because of IC being a feature of bad design are real decisive arguments, both being expressed in non-quantified probabilistic terms.  However, the above discourse is, to my mind, sufficient to reject the design inference based on the IC concept, as logically untenable.





Behe, Michael J. 1996. Darwin’s Black Box: The Biochemical Challenge to Evolution. NY: Touchstone.


Berlinski, David. 1998. “Gödel’s Question.” In W. Dembski, ed. Mere Creation. Downers Grove, Ill.: InterVarsity Press.


Cairns-Smith, A.Graham. 1986.  Seven Clues To the Origin of Life: A Scientific Detective Story. Cambridge: Cambridge U. Press.


Chaitin, Gregory J. 2003. “Randomness and Mathematical Proof.” In Niels Henrik Gregersen, ed. From Complexity to Life.  Oxford: Oxford U. Press.


Dembski, William A. 1998. The Design Inference.  Cambridge: Cambridge U. Press


Dembski, 1999. Intelligent Design. Downers Grove, Ill: InterVarsity Press


Dembski, 2001. “What Intelligent Design Is Not.” In W. Dembski and J. Kushiner, eds. Signs of Intelligence: Understanding Intelligent Design. Grand Rapids, MI: Brazos Press.


Dembski, 2002. No Free Lunch: Why Specified Complexity Cannot Be Purchased without Intelligence. Lanham, Maryland: Rowman & Littlefield


Dembski, 2005. “Irreducible Complexity Revisited.”

Accessed on February 1, 2005.


Doolittle, Russell F. 1997. “A Delicate Balance.” Boston Review, 23, no 1:28-29.


Dunkelberg, Pete. 2003. “Irreducible Complexity Demystified.” In Talk Reason. . Accessed on February 17, 2005.


Inlay, Matt. 2002.“Evolving Immunity.” In Talk Reason. . Accessed on February 17, 2005


Johnson, Phillip E. 1991. Darwin on Trial.  Downers Grove, Ill.: InterVarsity Press.


Johnson, 2000. The Wedge of Truth. Downers Grove, Ill: InterVarsity Press.


Korthof, Gert. Review of Michael Behe’s Darwin’s Black Box. In Was Darwin Wrong?  accessed on August 1, 2003


Marcell, Harry. 1976. “Evolution – Theory or Faith?”. In Aryeh Carmell and Cyril Domb, eds. Challenge: Torah’s Views on Science and Its Problems. NY: Feldheim


Matzke, Nicholas. 2003. “Evolution in (Brownian) Space: a Model for the Origin of the Bacterial Flagellum.”  In Talk Reason. .


Miller, Kenneth R. 1999. Finding Darwin’s God. NY: Cliff Streets Books.


Morris, Henry. 2005. “The Design Revelation.” In Institute of Creation Research. , accessed on February 17, 2005


Muller, Hermann J. (1918) “Genetic variability, twin hybrids and constant hybrids, in
a case of balanced lethal factors.” Genetics 3: 422-499


Muller. 1939. “Reversibility in evolution considered from the standpoint of genetics.” Biological Reviews of the Cambridge Philosophical Society. 14: 261-280.

(The references to Muller's papers were supplied by Douglas Theobald and Reed Cartwright whose contribution is thus acknowledged).


Musgrave, Ian. 2004. “Evolution of the Bacterial Flagellum.” In Matt Young and Taner Edis, eds. Why Intelligent Design Fails: A Scientific Critique of the New Creationism. New Brunswick, NJ: Rutgers U. Press.


Orr, Allen H. 1996-97. “Darwin vs. Intelligent Design (Again): The Latest Attack on Evolution Is Cleverly Argued, Biologically Informed – and Wrong.” Boston Review, 21, no 6.


Perakh,  Mark. 1999. “Irreducible Contradiction.”  Talk Reason. . Accessed on February 17, 2005.


Perakh  2001a. “Razumniy Zamysel ili Slepaya Sluchainost?” (Intelligent Design or Blind Chance?). Kontinent, No. 107, pp. 338-362. (In Russian).


Perakh 2001b. “Razumniy Zamysel ili Slepaya Sluchainosrt?”  Vremya Iskat, No 5, pp.30-50.  


Perakh. 2004. Unintelligent Design.  Amherst, NY: Prometheus Books. (Chapter 2).


Perakh. 2005. “IC’s Irreducible Inconsistency Revisited.” In Talk Reason, , accessed on February 17, 2005.


RBH. 2005. Posts at;f=13;t=001884;p=0). 


Ussery, David W. 1999. “A Biochemist’s Response to ‘The Biochemical Challenge to Evolution’”.  Bios, 70: 40-45.


Ussery 2004. “Darwin’s Transparent Box: The Biochemical Evidence for Evolution.” In Matt Young and Taner Edis, eds, Why Intelligent Design Fails: A Scientific Critique of the New Creationism. New Brunswick, NJ: Rutgers U. Press


Wells, Jonathan. 2002.  “Darwinism: Why I Went for a Second PhD.” In The Words of the Wells Family. .  Accessed on February 17, 2005