I swore I wouldn’t get into this anymore. The intelligent design crowd keeps pushing this crap, though, and I have kids who are at risk. Beyond that, I suppose intelligent design’s sort of dishonesty just galls me. Don’t say I didn’t warn you.
I am about to give an argument by analogy. I do so with trepidation, because analogies are always in danger of going off in the wrong direction and have no deductive validity in the first place. However, I think this form is the only fair tool for the subject since I’m going to use this argument by analogy to critique another argument by analogy: intelligent design (ID). Rather, I’m going to critique the positive arm of the intelligent design argument. This is the withered limb compared to the evolution-bashing arm, but it is necessary to the whole and the negative argument is a morass. Since scientific knowledge is never complete, critics have available to them an endless list of objections. The positive side of the ID case is more a philosophical than a scientific argument, so it can be settled on that level.
The positive ID argument is as follows: when humans start with a purpose (a problem to solve) and devise a tool to serve that purpose, the end product looks a certain way. Biological structures have a similar appearance, so biological systems must result from the same sort of process. An immediate problem arises at this point. Since proponents wish the analogy between man-made objects and biological structures to be precise the argument commits the fallacy of the undistributed middle (Designed structures look like X, Biological structures look like X, therefore Biological structures are Designed structures).
The proponents of ID subsequently introduce a number of modifiers to try to alleviate this situation. Complex specified information is not only used to portray designed and biological structures as essentially similar, it is also used to try to pare down the middle by saying it is statistically negligible. Appeals to abductive reasoning serve the same purpose, suggesting that design is so far and away the “best guess” that there’s no need to worry about the logic. There are deeper problems than the modifiers and the middle but those are the simplest things to consider, so let’s go there first. Let me tell you how I built my son a mountain bike.
The bike came in a big box in a bunch of pieces with a couple dozen pages of instructions. The instructions were complex and very specific. In fact, if you read the instructions through, you would know that a mountain bike and nothing but a mountain bike would result from following those instructions, even without actually building a bike to find out. But, I didn’t read the instructions. I know how bike parts work, so I just eyeballed the problem and figured it out. So, even though the Complex Specified Information (CSI) in the instructions has all the qualities that the advocates of biological CSI wish it to have, it doesn’t have the necessary relationship to the endpoint that those advocates want from it. There are multiple paths through the middle to the mountain bike. There is still a way out for ID, though. I must have picked up on the CSI contained in the parts themselves. It’s true. When I looked at the parts, they fit together in certain configurations and orders of assembly best, as I expected. My expectations were key, too. Over the years, bike designers had shaped those expectations, in effect teaching me to read the information they put in the parts so I could use my abductive reasoning to make a really good guess about how the parts should go together.
I use this “best guess” faculty all the time, because it is a great shortcut and pretty reliable in familiar circumstances. Notice I said ‘pretty reliable’. Also note that I said I didn’t use the instructions, but I didn’t say I threw them away. Sometimes the bike makers come up with an innovation and then my abductive reasoning is worthless and I have to rely on the instructions again to tell me what to do about the new part. I get in trouble if I try to rely on that ‘best guess’ shortcut in clinical medicine, too. The causal history of a manufactured bike is well-defined because people decided to make it that way. If the causal history of a patient’s symptoms is well-defined for me, it’s because I have decided so. Used outside of a situation with known prior constraints on the variables, the ‘good guess’ becomes confirmation bias. So, the problem is with the complex specified information. In biological structures, to achieve ‘specification’ and thus make the ‘best guess’ inference to design, the causal history has to be constrained after the fact. Anything less leaves open those pesky intermediate paths through the middle. ID imposes that constraint by assigning purpose to all biological structures it considers. Assigning purpose is easy for us humans and we like to do it because it lets us use shortcuts like guessing. Attributions of purpose (intent) are so appealing that we have trouble keeping them in the realm of human behavior where they belong (and not even there without some confirmatory process to check the attributions). Who hasn’t said their car “worked hard” to get up a steep hill? It’s just as easy to say that E. coli intends to swim to new food sources with its flagellum. In fact, it takes just that sort of presumption of intent to wrangle an object’s causal history into CSI, resulting in a bit of a Cartesian Circle (if the flagellum is made for swimming, then it has a complex history treading a narrow path to that endpoint, which shows that the flagellum was made for swimming). Still, we should be able to safely use this attribution of intent after the fact in limited circumstances, as long as we’re careful, right? For instance, it is surely accurate to say that the guys who designed my son’s bike did so from an original purpose. However, even that presumption of intent from the endpoint is not accurate, and the problem with the retrospective attribution of purpose/intent in such a case leads us back to the problem with intelligent design that predates any attempt to distribute or minimize the middle.
What do we know about design? We really just know what we do and what we do requires an agent (us), a purpose, and means. The problem is that the relationship between those three factors is not linear, nor is it even hierarchical. When considering bike design, for instance, we could go back to the origin of the means via the agent and examine the influence the means then had on the agent’s intent and subsequent development of further means. We could track back to the origin of the wheel in geometry, which is in turn based on observed properties of materials, which are in turn based on some basic laws of physics, all of which humans bothered to investigate and remember in the first place because, if you are a tool-maker, it’s easier to investigate and remember than to, um, ‘reinvent the wheel’. We could trace bicycle history back to the wheel and beyond, but let’s keep it brief and just consider the design of mountain bikes.
Mountain biking started when some California bike racers moved to the country. Their new environment confronted them with the problem of riding, and of course racing, on gravel roads. Their road bikes’ narrow tires were too unstable for that purpose, so they found some preexisting cruiser bikes with wide tires that would at least be rideable on the fire roads near their homes. The cruiser bikes were not perfect. They were heavy and hard to pedal, so the riders raced them downhill. Even that compromise lead to problems though. The brakes and bearings on the cruisers couldn’t survive that kind of abuse. The riders replaced the brakes and bearings with motorcycle and road bike components. The riders soon found that the revised cruisers, now possessing cassettes of gears with the road bike bearings, were capable of riding on rough trails as well as fire roads. Trail riding prompted further modifications to the bikes. These guys in rural California invented the mountain bike, but not all at once and not out of the blue. They worked through a progressive series of problems, each leading to the next, until they arrived at a relatively stable final design that did something very different from the structure they started with. The mountain bike evolved. Of course, this is microevolution; the mountain bike is just a tweaked cruiser bike. Neither the mountain bike nor the cruiser looks anything like an old penny-farthing with the giant wheel in front. The lineage is clear though, and bike development has proceeded by the same basic process from the wheel to the velocipede to the mountain bike. Moreover, the agents in this process acted as selective forces and were acted upon by selective forces – and not just physically. As designers altered the bikes, the bikes’ new capabilities altered their conception of where and why they might ride a bike and thus their purpose in the next set of modifications.
To fingerprint design as the ID scheme misrepresents it, we really must close that Cartesian Circle by presuming intent for any and all endpoints we wish to examine. Then the history of that point is seen separated from any branches or external contingencies. If the mountain bike comes from a mountain bike factory, surely the mountain bike factory holds the entire explanation for its structure. When defined after the fact like this, the history of a structure looks irreducibly complex; if you take away one part it is rendered meaningless because it is its own context. ID’s analogy between designed structures and biological structures not only fails to distribute the middle, it doesn’t even accurately depict design processes as we undertake them. People, the source of everything we know about design, don’t start cold from an undetermined purpose and design toward that purpose in an implacable, irreducibly complex chain of events. Replication may work a bit like that, but not design.
What this method really does is provide for a hierarchical relationship between presumed intent and biological structures, where the intent causes the structure. Such a relationship seems to allow for a supernatural cause. This is why ID’s advocates have gone through such contortions to make it work (or at least look like it might). Yet the intelligent design model fails even as a portal for the supernatural. It offers no solution to the interaction problem in dualism. This is a real problem because, as far as I can tell, one tenant of ID is that the design process in nature is ongoing. To drag a spiritual being into the material world and have it start doing things, one has to explain how it does so without being in some way beholden to the same laws, and thus part of the same causal history, as the rest of matter. If there is no good explanation, then the spiritual being from another realm is just a bizarre, unexpected new part of nature. Though this may seem an obscure technicality at first glance, here is an example of just how sticky the problem really is.
Descartes tried to defend the independence of the mind from brain processes. He offered the analogy of a virtuoso violinist asked to play on a broken instrument. The listener would have no clue as to the maestro’s true skill. Likewise the damaged, diseased, or intoxicated brain may just be a broken instrument unable to give voice to the intact mind which plays upon it. Unfortunately, this analogy raises the question: May the virtuoso be a virtuoso without a violin? Study of music theory or any other purely mental operation is insufficient. He must play a physical violin. Yet the skill he gains is a mental faculty which is subject to, and necessary for, his creativity, religious concepts, and emotion. The brain and its adjuncts affect the mind. There is no escape from this problem in a world of two supposedly separate substances in active contact. Deism or a strict idealism offer the only outs (and Deism may just push the problem back in time). Either of these scenarios keeps the supernatural supernatural, but thereby makes it irrelevant to any practical understanding of nature/matter. This is why it is best for religion that science adhere to methodological naturalism. This is why intelligent design is insidious as well as invalid, for all concerned. It robs religion of any hope of philosophical integrity, just as it misrepresents biology. Reason enough for everyone to drop this bullshit for good.
I think you are being too hard on ID. You cite the fallacy of the undistributed middle as if ID was doing that consistently, rather than suggesting links. I think I would prefer specific examples were ID proponents commit what accuse them of instead of mountain bikes. I’m not being cheeky, I just like real examples. That is how I learn. It doesn’t sound like you care to get into it, but thanks for the link.
It’s hard not to say “everything”, since I’m objecting to their premise. Their examination of the flagellum is the most complete example. A recent example from a Discovery Institute publication: Contrasting Astrobiology with Intelligent Design as Science, “…Intelligent design is not an argument from analogy. Analogies compare unlike things. A bacterial flagellum is not analogous to an outboard motor; it is an outboard motor. A molecular machine that employs energy to produce work is just as much a machine as an artificial machine…”. Restating an argument in stronger terms, here changing the ‘is like’ (is analogous to) to ‘is’, doesn’t change the structure of the argument. It is a debate tactic, though.
The second statement, besides equivocating the colloquial meaning of work with the term in physics, is an example of retrospective attribution. Assigning the work determines the structure; the molecular machine is made to do just what it currently does, just as the artificial machine is made to do just what it currently does. They must establish this for irreducible complexity and CSI to hold up. This specific relationship of current function to structure is not true, either for human designs or for biological structures. The only way to make it look true is to remove a structure from context. These guys are not stupid and I have a hard time believing they just made this kind of mistake over and over again. I can go looking for more examples if you like. They’ll come mostly from Meyer and Behe, though. I can’t bring myself to read anymore of Dembski’s stuff, he’s a worse writer than I am!