Cryptohedonology

Teleological arguments for the existence of god are my favorite theological debates. Not because I agree with them in how they are typically applied. And not because they are good logical arguments as usually outlined. Rather, amongst all the approaches to prove or explain god’s existence, teleological arguments 1) have a hint of evidence-based thinking in them and, in spite of being pretty weak, can still serve as inspiration for real scientific questions of teleology; 2) spark my imagination and, 3) unlike ontological arguments, still generate considerable contemporary controversy, so you can in principle make a difference in people’s belief structure by examining and posing alternatives. In contrast, for example, ontological arguments for god (arguments from being) tend to be pretty dry and always violently beg the question (assume the thing they are trying to prove). As far as i can tell, no one takes ontological arguments for god seriously (not even hard core believers) because they are unchartable in their illlogic by any modern standard — yet while pretending to be rigorous. On the other extreme, faith-based arguments are by far the most popular for theists, but such arguments are so difficult to confront and leverage against because they are really just expressing a subjective emotional state. They aren’t even pretending to be based in reason.

A teleological argument is just a fancy way of saying “argument from design or prupose.”  Such arguments are frequently abused by creationists in an effort to support their case for creation and criticize evolution. This thought provoking argument gains a bad reputation because of the company it keeps, not because the idea is without merit. Although I will be accused of generating a straw man, here is the usual teleological argument for god as I see it, as frequently used by creationists, in cartoon form:
1) By inspection e.g. by personal observations) elements of nature (in particular life, but not limited to that) appear engineered or created;
2) The truth of the observation is self-evident, therefore there must be a creator because self engineering not possible;
3) It is also self-evident that this creator must be god;
4) That god is invariably the same god of the bible [or insert appropriate holy book] who is responsible for creating the universe, dictating moral code, etc.;

Classic gedanken experiments posed by creationists include the famous watch on a beach, painting in the forest, and tornado in a junkyard scenarios and their variations. The first two situations are addressing point (1) above. The moral of the story as told is if you saw a watch on a beach or a painting in a forest, you would instantly know it was created. The claim is humans have intrinsically good observational skills when it comes to assessing if something was created or not. The tornado situation is indirectly trying to supply evidence for (2) by setting an implicit background for the signal: undirected design (“self engineering) or total randomness is insufficient to create anything more complex than another pile of junk.

The argument and scenarios above are deeply flawed and riddled with both thinking, factual, and argumentative fallacies. Nevertheless, I believe teleology, though greatly abused by modern (typically Christian) creationists, it is the best intellectual evidence-based foothold theists actually may have into demonstrating a creator’s existence (something resembling god), subject to qualification. The irony for them is that it may be a careful-what-you-wish-for situation. For example, their god/creator might be some kind of alien being (but natural, not supernatural). If super alien scientists seeding pre-biotic earth with proto-DNA replicators (which eventually evolve into humans) seems unlikely to you, I agree. But such a scenario, however strange, is still within the realm of the natural world, so is infinitely more likely than some sort of supertantural process. In short, by examining the fallacies of the standard argument, you can actually reformulate the issue into something approaching scientific terms. Unlike ontological or faith-based arguments, almost all parameters of the teleological argument can be reformulated into a coherent, naturalistic framework, at least in principle.

Let’s examine issues in point (1) in the argument while using the watchmaker/painter framework to highlight problems with the reasoning.

Fallacy or reasoning errors:
The argument mistakenly assumes that because we are pretty good at identifying human-made things, humans are good at identifying all things created for some purpose by any intelligence. Both the watch and painting thought experiments highlight the issue. If you saw a watch on a beach or a painting in the forest, you would certainly identify the watch and painting as having been made by humans. You have experience with the concepts of watchmaking and painting and have seen watches and paintings before in context. But notice in both scenarios, to contrast the created object, advocates place the items in an environment intended to be a neutral or uncreated space. For example, the watch problem is never posed as “you find a watch in a watch store.” Everyone would agree right away the watch was created under such conditions. No, the watch is placed on a beach. The sand, ocean, sky, wind, sun, etc. are meant to be representative of an uncreated background. Similar with the painting. The problem is never framed as “you find a painting in a gallery in New York.” The implication is that the trees, ground, sky, rocks, etc. in the forest form a contrasting neutral background. The irony, of course, is that advocates of the argument by design would also claim that god created all the elements of the background too, yet our inherent design skills apparently are intended to fail us in these analogies. This can quickly become an exercise in the absurd because if everything was created by an intelligence, including things you would normally dismiss as clearly not-created using your best human experience, the entire “by inspection” approach just implodes on itself.

But in fairness, let’s agree there are different degrees of design clarity. An ancient arrowhead might look like a rock to an untrained eye, but a supercomputer would be an obviously human designed object to most modern people. Perhaps to an ancient person, the rocky arrowhead is obviously designed while the supercomputer would be simply unidentifiable, perhaps something from the gods. However, they would try to identify it with things they knew. Perhaps the supercomputer would resemble a very abstract adobe oven with strange clay pots and flashing lights attached. But this is basically my point. One’s degree of familiarity with the object, even in the abstract, and ability to visualize, even in principle, the creation process is a key element to our design tagging capability. Like the ancient man being unable to properly process the supercomputer, if you saw an alien artifact that was well outside your experience, even in the abstract, then you might easily mistake it for something undesigned. A flying saucer would be tagged as designed because flying saucers ideas themselves are human made, based on quasi-real machine and aircraft (and dishware) modifications. But an alien ship that resembled a blob of protoplasm (think Zerg production facilities in Starcraft or various bizarre otherworldly transdimensional alien technology from the Cthulhu mythos) or a sea cucumber, might easily be mistaken for, well, a blob of protoplasm or a sea cucumber, not an alien technology. A man is walking along a beach and sees a Rolex watch and a personal teleportation device designed by alien — which just happens to resemble a dead sea urchin. Yet the man reaches for the Rolex marveling by its intricate workmanship…

To reframe the problem in naturalistic terms, let’s isolate the problem: can something be identified as having been created by simply observing it? In other words, is there a creation measure: a single parameter that you can ascribe to an object that tags it as having been created by an active intelligence? As a physicist, my inclination is to frame the problem in thermodynamic terms. Is this creation measure a state variable (i.e. an inherent property of a system) or is it a path dependent variable (you must know the initial state, final state, and the process connecting them to assign a value). Temperature of an ideal gas in thermal equilibrium is a state variable. At some instant, you can take a thermometer and obtain a single value for the temperature. But a value like “work done on the gas” is path dependent, meaning it depends on the exact process by which work is transferred.

Creationists over the years have tried to come up with creation measures using information theory. Many opponents are cynical about this because it is usually perceived as a distortion of science to push a religious agenda. This is indeed frequently the case. However, I actually admire the effort to obtain this evidence and do these calculations because this is exactly the sort of thing you must do to build a scientific case. But using information theory to assess design probability is wrought with problems, both experimental and theoretical. Information content is something that can be assessed in principle. It is roughly a measure of the configurational entropy, which is a close cousin of Shannon’s negentropy in computing. One problem is, in order to calculate such a number for the entropy and start doing calculations, you need to know all the configurational constraints. In thermodynamics for simple systems, these constraints come in many varieties, but there are three very common ones taught in an undergraduate physics class. The microcanonical ensemble is a system where every energy configuration is equally accessible. The canonical ensemble is one where all configurations are accessible, but the average energy of the system is fixed. Then the grand canonical ensemble also has all configurations accessible, but both the average energy and average particle number is fixed. Although the math machinery and calculation power is similar for each case, the sorts of results and values you extract based on the assumptions are not.

Another deeper problem is that, although it is perhaps a good first guess, higher information content (i.e. lower configurational entropy, the thing being used as a creation measure) is not a very good measure of design. To highlight this (and also demonstrate how important constraints are), lets examine the phase transition of liquid water to ice (at standard pressure). Ice has a lower configurational entropy than liquid water. The configurations of ice have lower configurational entropy and higher information content because they are highly constrained (some would call this more ordered, but we don’t want to evoke qualitative statements of order, although that frequently correlates with lower entropy, but not for the reasons usually taught). The system arose not because each crystal was actively created by an intelligence but because of the 1) the microscopic geometry of a water molecule and 2) the weak attractive force between these molecules. The relationship between 1 and 2 is not entirely independent — both are byproducts of fixed properties of the atoms and how they couple. This information generating process occurs predictably and systematically for many gazillions of molecules of water on a regular basis when the temperature changes in a certain way — all without the help of an active designer. And it comes at a cost: some other part of the environment must lose some informational content to compensate. Yet I don’t think anyone is advocating that individual ice crystals are “created” in the usual sense of the word. However, taking it to the next level, one might go so far as to say that to create ice, the designer configured the atomic structure and couplings to be exactly constrained to generate ice at a certain temperature and pressure. That’s fine to hypothesize. But this sort of creation isn’t a property that can be probed with configuration entropy. If so, you would conclude that liquid water or vapor was not created, while ice was — yet this built-in design feature as I described it would still be present in all cases.

Another simpler example is the Lottery. Draw 7 numbers and compare it to 7 numbers you picked. This is a microcanoical case because each configuration of the system (assuming it is a fair game) is equally probable. If one didn’t apply the right model, one might conclude that the number combination {1, 2, 3, 4, 5, 6, 7} was less probable than the random combination {12, 32, 41, 2, 4, 18, 8}. But it isn’t true. Each configuration has the same probability and thus the same amount of information. It is only when you start comparing collections of combinations will the probabilities shift. For example, if you compared all consecutively ordered sets of 7 everything else, then the former would clearly have a much lower probability. This is what the junkyard example is trying to illustrate. Treated as a microcanonical ensemble, a tornado will randomize the configurations of a junkyard, sampling all configurations equally. In spite of that, you would never expect to see a car emerge from some such situation because the probability is so low. However, you can imagine a fixed game where the system *appears* microcanoical, but isn’t. What if there were conditional probabilities built into our Lottery like if you draw an even number, the next number will be odd or the next number drawn will never be more than two away from the previous…or…given these constraints, or combinations of them, then configurations like {1, 2, 3, 4, 5, 6, 7} could be vastly more probable than an arbitrary one like {12, 32, 41, 2, 4, 18, 8}. Again, without knowing the constraints of the system, you can’t calculate the configurational probabilities and thus can’t calculate the information content. Interestingly, the chemistry of life acts more like the fixed lottery example. Certain seemingly unlikely chemical combinations are vastly more likely if other “mildly unlikely” combinations occur first.

In summary, there are other technical problems with using such creation measure such as configurational entropy or information content. The short answer is that 1) it is difficult to calculate and measure this number for arbitrarily complex systems because you need to know a lot about the system to begin with and 2) it doesn’t generally tell you anything about design or creation, the very thing you set out to do.

So is there a creation measure that acts as a state variable? That is, can I just look at something, perform a single measurement, and assess if it was created by an intelligence? I doubt it. My hunch is that any creation measure will be inherently path dependent. When a human assess design, they are weighing the whole path of creation based on experience. This may be faster for some familiar things than others, and may not be an entirely conscious thing, but at the very least we as humans are making assumptions about that path. Today, with computers, this can throw off a humanity’s ability to measure created things. For example, it is easy to read computer generated poetry these days and not know it was computer generated. It has the look and feel of human generated poetry — and even may tell a story that feels human — but it was created at random by an algorithm, not a human.

It will not be long before computer generated poetry, prose, paintings, etc. will easily pass Turing-like tests. So, to distinguish between human and machine-made creative works, we will *certainly* need to know more than simply the final state of the artistic system (we should be asking that anyway). As a side note, the path to creation has always been an important part of art. However, people can also simply observe the final state (the art piece itself) and get pleasure and personal meanings from that as well. This will still be true in the case of machine-made art, but perhaps to fully understand a piece one needs to fold in the path of the artist — as usual.

I would like to address elements (3), (4), and (5) in the teleological argument as I outlined it above. Imagine that some evidence existed that life one earth was created in some fashion. As outlined above, I’m not entirely sure what that evidence would look like, but it probably will not be a simple measure, rather a network of knowledge that builds up describing a process and a path. Now, the essential question is, if we were created, must that creator be god? Specifically, does that creator have to be the same creator of the universe or one of the gods outlined in the world’s holy books (e.g. the bible, etc.)? Clearly, in my opinion, no all accounts. The kind of evidence you can expect would be contingent upon what kind of creator it was. As many science fiction works have dabbled with, the creator could be aliens, for example. As Michael Shermer once pointed out during a debate, this issue also forces you to examine the issue of what level we were created. At the individual level? Species level? Pre-biotic level? For example, to be consistent with evolution, we have roll back things all the way back to before life existed. Perhaps they seeded the planet with pre-biotic muck or tinkered with existing muck. It is clear that, in this scenario, these aliens would not have to be the creators of the universe. I’m not saying I believe this super-alien hypothesis. I’m just saying that there is at least the possibility of finding evidence for such a scenario, so it is naturalistic. However, I don’t think aliens are what creationists are thinking of when they make their case for argument by design in its generic form. Ironically, if we were created, it turns out this is one of the only reasonable natural, non-spontaneous things it can be! Careful what you wish for, intelligent design people! If the creator is a divine god then it isn’t at all clear what evidence would possibly support that. However, “by inspection” and “god told me personally” or “I read it in the bible” are all insufficient. If these are your only arguments, you should retreat back to the subjective comfort of faith-based arguments and not try and pretend the process to be science.

Years ago, I mentioned the above argument to a colleague. In his opinion, if it could be shown that we were created, no matter my whom, then this was enough for him to believe in god. His reasoning is basically following the creation chain all the way back: if we were created, who created the creators? And so on. The problem with this argument is manifold, and has been shredded by many others before me. Why terminate at god? If you are following a creation chain you are arbitrarily terminating it at something you personally choose: god. Why not terminate it sooner? Or later? Or not at all? For example, if we were created then those that created us could have simply formed out of a pre-biotic ooze spontaneously, just as we now believe for ourselves at this time. Another example extrapolates to the future. Imagine a future where there are intelligent robots (wasn’t there a running joke on the radio to that effect in Grand Theft Auto: Vice City?). We humans are extinct and all obvious evidence for us is gone, but we were responsible for creating prototypes that learned to self-replicate microscopically, leading to the robust robot society. These robots now ponder their hypothetical creator, who is beyond them. From their perspective, they evolved via evolution from a microscopic state to where they are. But now imagine the robots eventually find evidence that we (their human creators) did exist. Perhaps a city is uncovered and vast amounts of information is revealed about the origin of robots. Could the robots, now knowing they were created, conclude that we humans were also created? No, clearly not. We could have been generated through spontaneous means then evolved by ourselves. We just happened to figure out how to make self-replicating robots who then evolved into hyper-intelligent machines. The basic message here is that just because you can show via evidence one link in a creation chain was actively created by the previous, you cannot conclude the previous chain link to that was created as well. And you certainly can’t conclude the chain goes all the way back to a single source (indeed it violates your own assumption of a creation chain!). Each link must supply its own evidence of being actively created. However, conversely, if you can show just one link could spontaneously generate the next, then there is no reason to insert a creator for any link in the absence of evidence.

Who knows? My guess is that it is a very dark horse, but perhaps it is worth seeking evidence for a naturalistic creator, be it (they) super alien(s) or whatever. Just don’t do it in my backyard, with my tax dollars, or on on my (created by Timex, thank you) watch — unless, that is, you find a personal teleportation device shaped like a sea urchin, then we’ll talk about aliens. But be prepared to prove your point beyond a reaonsable doubt.