Genes, Alleles and Causation

I will step back, as ever, to the question of causal credit. It strikes me as odd to say that genes cause when we don't have a good model of the transitions from genotype space to phenotype space. I have made this point colloquially in the past by asking whether we would would be comfortable assigning most (all) explanatory credit to selection when someone enters a McDonald's and leaves with a burger? Or if we would allow a political scientist to tell us that the explanation for the election of a president was selection by the electorate and leave it at that. Surely in those cases whatever caused the menu to be as streamlined/attenuated as it is deserve some explanatory credit.
Dawkins and others have used genes as a rhetorical device to scare off or impress philosophers, journalists and a public who assumed that they were making genetic claims and that this sanctioned their statements as part of the reliable and gleaming structure of modern science. A cursory examination shows that no actual genetics was part of the selfish gene talk. Highly repetitive short sequences that proliferate in the genome were annexed to "selfish genes" but were not really initially part of the story.
I never liked Dawkins' rough handling of Mary Midgley as he condescended from the elevated status of biologist/geneticist when he wasn't really engaging from those fields.
Does the reasonable statement that "nothing is a complete and sufficient cause of anything" sanction genes as agents. Agency may require a bit more cause than this. I think we do have constraints on our actors (or what philosophers of science call theoretical entities) that require at a minimum that we have a sense of how states in the neighborhood of the state of interest would unfold. If a lot of variation is all funneled to similar outcomes then the parameters used to define this notion of local neighborhood don't matter and don't merit being called out as having explanatory credit (i.e. being treated as nouns in our theories). On the other hand, if small changes result in wildly different end states, this too suggests that the explanation is not "in" the initial condition, but rather in the dynamics governing the system.
You can imagine a 2 x 2 grid; small changes in input -> small changes in output, small changes in input -> large changes in output, large changes in input -> small changes in output, large changes in input -> large changes in output. Now this is a continuum and likely there are critical problems that fall into each of these.
For well behaved or maybe it is conveniently behaved entities we are looking for cases where small changes in input map to small (and predictable) changes in output and only in those cases do we nominate those variables as worth elaboration and experiment. Small changes in input resulting in large changes in output whether chaotic dynamics with sensitive dependence on initial conditions or control systems that change modes we will likely focus on the dynamics and control as much as the entities. When large changes are, essentially ignored and funneled to a small change in output, again we would focus on the regulatory dynamics that buffer the system. Cases where large changes result in large results are problematic for other reasons - if we move too far in parameter space we really have lost our notion of local. So for practical purposes when we name entities and measure them we will need a certain level of stability and input output coherence to allow us to study.
My friend Andrew offered an extreme (counter?) example:

I am standing, after an exhausting hike through Colorado, somewhere on the Continental Divide. I take a pee. Now, it seems to me reasonable to regard the ultimate destination of the liquid as caused by where I point myself. Will it end up in the Atlantic or the Pacific? I can choose; and, in that sense, the cause of its ending up in one or the other is my state of mind. Of course, this is only the cause because of a whole lot of other factors, involving gravity, geology, hydrology ... But science would be impossible without assuming that other things are equal in that way. This is a large change caused by a small variation, in which we can largely ignore the dynamics governing the system, since they don't change.

I think he is on to something but we have here a story that nicely builds up from a long and well understood series of causes with water flowing ever downhill from local minimum to local minimum which puts one or the other coast ever closer. It may be that this provides some methodological guidance. Look as closely as you can for dynamics you can work with.
In this sense is hair or eye color is caused by alleles of genes (which is what people often mean by "caused by genes") or are the alleles just a convenient accounting tool that surfaces a control element with out be an explanation.
For evolutionary biologists (as distinct from geneticists) a key thing to explain is the clumping in phenotype space. There are variations in the individual and patterns in the ensemble. There seem to be types (as an observation about our world that deserves explanation, not as an a priori claim). These types seem to be more than taxonomic artifacts. Though they are frustratingly reinforced and made more rigid by the shoe-horning that taxonomy encourages. And edge cases and exceptions are frequently not highlighted. Uniform distributions are rare and and so we do not need to assert a null hypothesis that all of phenotype space should be evenly filled. But I take it as an open question whether the clumping in phenotype space is fully/best/comprehensively explained by chance, lineage/history, the high dimensionality of the space, and/or biomechanical constraint. Selectionists would add fitness to that list. And I would add homology - developmental regulatory homology - which, to me, explains more of the clumping than selection (but lives inside any clumping due to my chance, lineage... list). What this means, in practice, is that when you start with a mammal you get a mammal and so bats fly with all their phalanges and saber-toothedness arises again and again and no animals have wheels...