The catalyst was this comment by 92g:
The English language is a teleological minefield. But the problem to which the above statement led me is much worse than a problem with language.
Engineering is an intellectual activity.
When we speak of a person's intentions, we tacitly acknowledge that such matters can get a bit complicated: a person's actions are not a simple matter of him doing what he wants, but reflect the greasing of the squeakiest wheel among an internal cacophony of various urges and counter-urges. Humans often demonstrate a phenomenal ability to make accurate guesses about which situations will cause another person's wheels to start squeaking, and which ones he is most likely to grease first. It is a skill that confers an advantage in the singles bar and the corporate boardroom alike; in some environments (such as the ancestral African savannah, the prison yard, or certain streets in most any large city), it is absolutely critical for survival.
Because it is so important, a considerable portion of our cognitive resources are dedicated to this task (one need only consider what percentage of the total bandwidth consumed by discussions about creation vs evolution are directed to the motives of a poster rather than to the content of his posts). I am convinced that high priority given to the performance of this task, and high reproductive reward for success was, directly or indirectly, a major driving force behind the evolution of the human brain.
Our brains leap instantly at the slightest hint of motive, and once they think they've got it, they don't like to let go. It is not surprising then that we always have an eye out for 'designs', and when we spot one, our minds immediately go to work formulating guesses about the designer and his intentions. If this practice is error-prone, the biologist may be the worst offender. Substituting "natural selection" for "designer", he points to the utter lack of purpose or direction in evolution, thinking that this is enough to free him of all hidden teleological assumptions. "Traits exist because they were selected for", he says, perhaps without appreciating the breadth of the leap he has just made. (Okay, maybe the evolutionary psychologist would be the worst offender.)
One problem is highlighted by the results of the recently completed human genome project: there simply are not enough functional genes to explicitly code for everything the biologist might like to refer to as a 'trait'. Apparently, the situation is even more complex than we thought, with a great deal taking place implicitly, or by side-effect; 'traits' (at least a lot of them) being emergent features of that complexity. Gould introduced his metaphor of the "spandrel" to explain the evolution of some features as the co-opting of side-effects of earlier features. Like many others, I have always found something vaguely unsatisfying about that explanation, but it serves at least to get us into the neighborhood of what I'm trying to express.
In his audaciously titled book, "Consciousness Explained", the philosopher Daniel Dennett proposed (for purposes tangential to this discussion) that among the properties which might be considered 'design features' of a 'designed object' (or, say, an organism) is its center of gravity. That's the sort of thing I'm talking about. If one were to consider that as a 'trait', where on the genome would one begin looking for the nucleotide sequences which produce it? How many other 'traits' might prove as elusive? Do we have mathematical models adequate to sufficiently isolate such 'traits' so as to even attempt to measure them in terms of 'absolute fitness', or 'relative fitness'?
What has been described as the "Central Dogma" of evolutionary biology is this mapping: DNA=> RNA=> Amino Acid=> Protein. There are nuances of that process enough to keep biochemists working for quite some time, but it does not seem as mysterious to me as it did when I was first introduced to it. Usually omitted, however, though tacitly presupposed, is one last mapping: Protein=> Trait. This I still find absolutely mind-boggling. It seems to me that reconstructing those mappings accurately enough to permit full appreciation of their evolutionary implications may not always be possible even in principle, a difficulty which often appears to be overlooked. Perhaps this is part of what inspired Gould to repeat the clichÃ© that evolutionary reasoning is just "cocktail-party speculation".
At the very least, a lot of empirical work remains to be done in this area, especially with regard to proposals made by the most ambitious of theorists. If I were looking for a chink in the armour of evolutionary theory (and loyalty to the highest standards of scientific inquiry seem to demand that I do so), I think I'd start there.