5 RESULTS AND DISCUSSION
5.2 Intelligent Design and Biology (II, IV)
5.2.1 Finding Design in Nature
Implications of Dembski’s theory for biology and the methodology of science are considered in Articles II and IV. Dembski himself emphasises that “the focus of the in-telligent design movement is in biology. That’s where the action is”.259 The importance of finding traces of design in the natural word is also highlighted in the definition of intelligent design offered by the Discovery Institute, where the theory of natural selection and “certain features of living things” are explicitly mentioned:
Intelligent design (ID) is a scientific theory that employs the methods commonly used by other historical sciences to conclude that certain features of the universe and of liv-ing thliv-ings are best explained by an intelligent cause, not an undirected process such as natural selection.260
In Dembski’s view, some biological structures are too complex and too rich in infor-mation to have emerged through the “blind” processes of nature. Dembski relies on the idea of “irreducible complexity” introduced by another prominent promoter of intelligent design, Michael Behe. An irreducibly complex system, according to Behe’s definition, is “a single system composed of several well-matched, interacting parts that contribute to the basic function, wherein the removal of any one of the parts causes the system to effectively cease functioning.”261 Dembski explains that
because the irreducible core of an irreducibly complex system can’t be simplified without destroying the basic function, there can be no evolutionary precursors with simpler cores that perform the same function. It follows that the only way for a direct Darwinian pathway to evolve an irreducibly complex system is to evolve it all at once and thus by some vastly improbable or fortuitous event.262
In other words, Dembski holds that irreducibly complex systems are necessarily formed in one go because the probability that any putative precursors of an irreduci-bly complex system could have evolved into the system through evolutionary means is extremely small.263
258 For Dembski critique of the Bayesian approach, see Dembski 2004a, 232–248.
259 Dembski 1999, 14.
260 Discovery Institute Staff 2009.
261 Behe 1996, 39.
262 Dembski 2004d.
263 Dembski 2004d.
Dembski believes that a supernatural intelligent designer—or, God—must be the source behind these “irreducibly complex” structures which (allegedly) are unexplain-able by natural causes. Dembski asserts that God controls the development of some biological objects by entering information at convenient places in their development process, perhaps through controlling the outcome of seemingly random quantum events which are then amplified on a wider scale. The idea of quantum amplification, however, seems to be rather speculative and is not supported by the current theories of quantum physics.264
Irrespectively of the method by which God influences nature, Dembski is certain that God definitely does that. According to Dembski, it can be calculated that God has entered complex specified information into the development process of some biological organisms. Since “irreducibly complex” biological structures are formed abruptly without predecessors, they can be treated as “discrete combinatorial ob-jects”. Dembski maintains that the probability of the formation of such an object can be calculated with the formula pdco = porig × plocal × pconfig, where porig is the probability of originating the building blocks for the object, plocal the probability of locating the building blocks in one place once they are given, and pconfig the probability of config-uring the building blocks once they are given and in one place. Now, if pdco < 10-150 (the universal probability bound265) and if the object is specified the object is—according to Dembski’s theory—designed.
5.2.2 Applying the Theory of Intelligent Design to Biology
The only example of a biological object the development of which Dembski explicitly deals with is the flagellum of the Escherichia coli bacterium. Surely, one plausible coun-terexample is sufficient to disprove the claim that all biological organisms have been produced by natural causes. In other words, if it can be confirmed that the bacterial flagellum exhibits complex specified information (and if it is assumed that the criteri-on of specified complexity is a reliable method of detecting design in the first place), it must be accepted that supernatural causes have played part in the development of some biological structures (the flagellum, in particular).
Focusing on this one example does not mean that Dembski would not think that there are no other biological objects that are designed.266 So when Dembski suggests that the promoters of intelligent design do not need to be “committed to every biologi-cal structure being designed”267 but merely to “find some clear instances of design and
264 Quantum amplification might work to a restricted extent but is unable to account for all God’s actions.
For a thorough discussion on quantum amplification and its problems in the context of divine action, see Saunders 2000; Saunders 2002; Koperski 2000 (a more general analysis is Laughlin & Pines 2000).
265 Dembski’s estimation regarding the universal probability bound is rather conservative when compared to some others presented in the literature. For example, Henry Morris and Gary Parker (1987, 270–271) have estimated that the maximum number of events ever to take place in the universe would be 10170. As a special case pertaining to the formation of biological structures, Fred Hoyle and Chandra Wickramasinghe (1999) have argued that the probability of already-available amino acids spontaneously combining in the right way to yield living organisms would be 10–500000.
266 See, e.g., Dembski & Wells 2007 for a comprehensive presentation of alleged examples of design in biol-ogy. More generally, Dembski (2001b, 222) believes that “God created nature as well as any laws by which nature operates. Not only has God has created the world, but God upholds the world moment by moment.”
267 Dembski 2004a, 63.
nail them down,”268 this does not appear to fully reflect his personal views. Instead, it might be a strategic choice to only highlight “some clear instances of design.” As stated above, in order to make a case against naturalism, it is not necessary to show that design exists everywhere in the world (although Dembski might believe that it does), but merely to construct one credible example of design.
Dembski calculates that both the localisation probability (plocal) and the configura-tion probability (pconfig) of the bacterial flagellum fall below the universal probability bound. Therefore, the bacterial flagellum would clearly constitute complex informa-tion. (For simplicity, Dembski estimates that the origination probability (porig) of basic building blocks of life such as amino acids would have to be close to 1 or else life would not be possible in the first place.) Moreover, Dembski takes every functional system in biology to be specified, so he deduces that the flagellum exhibits complex specified information and is designed.
However, as argued in Article II (and explicated further below), Dembski’s calcu-lations are problematic.
1) First, his understanding of how biological structures are formed appears to be incorrect. Namely, Dembski calculates the probability of the formation of the bacterial flagellum on the condition that it was formed through a random assem-bly of proteins. The theory of evolution, instead, suggests that proteins are not drawn together randomly but evolve in interaction with other molecules from simple forms and only gradually form more and more complex structures.269 It has been shown that in the case of the bacterial flagellum, there are also plausible scenarios for the structure having evolved gradually without a non-functional intermediate, with selective benefits at each step.270 In other words, it is quite possible that the flagellum is not irreducibly complex and cannot be treated as a “discrete combinatorial object”. It is probable that ordinary evolutionary mechanisms also apply in the case of the flagellum.
2) Second, Dembski grounds his claim that the flagellum is specified upon the assertion that “humans have developed motor-driven propellers well before they figured out that the flagellum was such a machine”. However, it is not clear that a random “motor-driven propeller” constitutes a pattern corresponding to the flagellum. On the other hand, as Dembski points out, it might not be nec-essary to have a precise one-to-one correspondence between the object under examination and a pattern specifying it. All that is required is a “near-enough”
resemblance. More precisely, the pattern describing the object must be one of a very limited number of patterns belonging to a sub-group of “target patterns”
among a wide variety of all possible patterns. However, it is difficult to objec-tively define how near counts as “near enough”, for example, in the case of the flagellum.271 Furthermore, Dembski’s more general assertion that “no biologist
268 Dembski 2004b, 210.
269 Alberts et al. 2015, 109–172.
270 Pallen & Matzke 2006; Wong et al. 2007; Chaban, Coleman & Beeby 2018.
271 This supports the idea that defining specifications is subjective (see Ch. 5.1.2). For example, it is debat-able whether Dembski’s example of a “bidirectional rotary motor-driven propeller” specifies the bacterial flagellum, in other words, whether the concepts “bidirectional”, “rotary”, “motor-driven”, and “propeller”
are sufficient to define the flagellum (Dembski 2005, 18).
I know questions whether the functional systems that arise in biology are speci-fied” (in addition to being an argument from incredulity) is somewhat mislead-ing since the term “specified” is not commonly used in evolutionary biology.272 Based on this, it can be inferred that Dembski was not successful in constructing a convincing case for detecting design in biology.
5.2.3 Biological Information and Evolutionary Algorithms
Dembski has also approached the issue of producing biological information from the perspective of evolutionary algorithms, aiming to show that the processes of evolution can only produce complex specified information if the process has been fine-tuned to produce complex specified information by a designer in advance.273 Evolutionary algorithms are search algorithms based on models of biological evolution, operating on a population of potential “solutions”. A typical example of an evolutionary algo-rithm is a search based on the processes of recombination, mutation and selection.
After an initial population is created,
[t]he environment delivers quality information (fitness value) for new search points, and the selection process favors those individuals of higher quality to reproduce more of-ten than worse individuals. The recombination mechanism allows for mixing of parental information while passing it to their descendants, and mutation introduces innovation into the population.274
Dembski (with Marks and Ewert) scrutinizes several widely-used evolutionary algo-rithms, reaching the conclusion that none of the models is able to properly describe the process of biological evolution. All analysed evolutionary models are shown to have a teleological goal (“imposed by an omnipotent programmer”)275 and a particu-lar known fitness function associated with the goal, whereas the process of evolu-tion is assumed to be undirected and lacking any knowledge of the fitness funcevolu-tion.
Therefore, although it might be possible to produce complex specified information through evolutionary algorithms, it is only possible if the ability to produce such information is built-in in the process itself.276 Indeed, Marks’s statement that “there exists no model successfully describing undirected Darwinian evolution” potentially holds true, although rebuttals have been made.277 This, however, only means what is says, in other words, that there would currently exist no evolutionary algorithm properly describing undirected evolution, not that the actual evolutionary processes would not be able to generate complex specified information.
Yet, Dembski goes further and maintains that there could never even in theory be an evolutionary algorithm capable of producing complex specified information. He
272 Some biologists, although not endorsing intelligent design, find it reasonable to talk about biological organisms as being specified (Felsenstein 2013; Dawkins 2015, Ch. 1).
273 Dembski 2002a, 179–237.
274 Bäck 1996, 7.
275 Marks, Dembski & Ewert 2017, 187.
276 Marks, Dembski & Ewert 2017, 105–250.
277 Marks 2017. For rebuttals, see Felsenstein/Kuhner lab 2019.
appeals to the no-free-lunch theorem which shows that “evolutionary algorithms, apart from careful fine-tuning by a programmer, are no better than blind search and thus no better than pure chance”.278 In consequence, because blind search is a hope-lessly slow process (this is common mathematical knowledge) but still as efficient as any evolutionary algorithm, Dembski claims that there cannot be an evolutionary algorithm which could meet the performance claims of the theory of evolution or produce “anything of even moderate complexity”.279
It is important to note that, as Dembski also does, only “when averaged across fitness functions, evolutionary algorithms cannot outperform blind search”.280 When applied to a particular fitness function, algorithms differ in their efficiency. Contrary to what Dembski seems to assume, the case in which all possible fitness functions need to be considered is purely hypothetical and does not model real-life situations.281 Moreover, the no-free-lunch theorem is only valid if the fitness function either remains fixed throughout the execution of the programme or varies in a manner independent of the search algorithm. In the case of evolution, the fitness function changes over time in response to changes in the population under the influence of the evolutionary algo-rithm.282 In other words, the fitness function is neither fixed nor independent from the algorithm. Therefore, the no-free-lunch theorem is not relevant for a proper modelling of the processes of evolution. To sum up, the evolutionary algorithm approach adds little to Dembski’s argument on the development of biological information.
5.2.4 Intelligent Design and Methodological Naturalism
As noted above, Dembski believes that there are biological objects the development of which cannot be explained by natural causes. Consequently, he argues that since methodological naturalism only accepts natural causes, it should be abandoned as a basic principle of scientific inquiry. Dembski’s argument against methodological naturalism can be formulated as follows:
1) If design inferences concerning some natural phenomena are warranted, then an intelligent agent has brought about such phenomena.
2) If an intelligent agent has brought about some natural phenomena, then natural causes will fail to explain such phenomena.
3) If natural causes fail to explain all natural phenomena, then methodological naturalism will in some cases lead to errant explanations.
4) Any methodology that leads to errant explanations should be abandoned.
5) Design inferences concerning some natural phenomena are warranted.
6) Therefore, methodological naturalism should be abandoned.283
278 Dembski 2002, 212.
279 Marks, Dembski & Ewert 2017, 59.
280 Dembski 2002, 212. More generally (to quote the words of the developers of the theorems, “any two optimization algorithms are equivalent when their performance is averaged across all possible problems”
(Wolpert and Macready 2005, 721). See also Wolpert 1996a; Wolpert 1996b; Wolpert & Macready 1997.
281 Droste, Jansen & Wegener 1999.
282 Wein 2002; Perakh 2003b.
283 I am here extending the formulation presented by Murray (2012, 603).
Furthermore, Dembski reasons that methodological naturalism should be replaced by intelligent design which does not a priori disallow the possibility of supernatural causes as an explanation of natural phenomena.
In Articles II and IV, it is argued that Dembski regards the consequences of his theory for the methodology of science to be more significant than they actually are. First, in order to replace methodological naturalism and to become a new scientific methodol-ogy, intelligent design would need to provide an alternative research programme. At the moment, such a programme does not exist, the main reason being that no cases of biological design have been confirmed. A more principled problem is that it is difficult to construct a design-based research programme capable of making any kinds of pre-dictions, because the intentions of designers cannot be predicted.284 Second, even
a) if Dembski’s theory of intelligent design would provide a reliable method to detect design, and
b) if the method could be applied to natural phenomena, and c) if some of these phenomena exhibited specified complexity,
it would still not be justified to abandon methodological naturalism. It would only mean that in some (rare) cases naturalistic explanations should be complemented with further explanations accounting for design. The success of science has shown that, on most occasions, methodological naturalism would still work and be a justified methodology.
In my view, Dembski’s argument against methodological naturalism should be refined as follows:
1) If design inferences concerning some natural phenomena are warranted, then an intelligent agent has brought about such phenomena.
2) If an intelligent agent has brought about some natural phenomena, then natural causes will fail to fully explain such phenomena.
3) If natural causes fail to fully explain all natural phenomena, then methodolog-ical naturalism will in some cases lead to partial explanations.
4) Any methodology that leads to partial explanations should be complemented with further explanations.
5) Design inferences concerning some natural phenomena are warranted.
6) Therefore, methodological naturalism should be complemented with a theory that includes design.285
In conclusion, Dembski’s claim that methodological naturalism should be completely abandoned if traces of design were found in the biological world comes across as excessive.286 However, if methodological naturalism is taken in the strict sense that it either has to apply to all imaginable cases of scientific inquiry or else it is not a useful concept at all (which seems to be a false dilemma), Dembski’s view becomes under-standable and closer to what I have presented above.
284 It is, of course, debatable whether predictability is a necessary requirement for good science.
285 Loikkanen 2015b.
286 Another way to justify the compatibility of methodological naturalism and intelligent design would be to argue that since the theory of intelligent design does not comment on the identity of the designer, the designer could be a natural being. This, however, would require a different kind of definition of the term
“supernatural” than offered in this study (see Chapter 5.3.2).