Stephens (2004) and others have identified the conventional Modern Synthesis view of evolution as a "theory of forces".   Evolution is defined as "shifting allele frequencies" in this view, and thus the "forces" of evolution push the frequency of an allele up or down over time.  That is, change in the frequency of an allele over time is the common currency of causation in evolution, just as change in location of an object over time is the common currency of causation in classical physics. Advocates of this view present the Hardy-Weinberg equilibrium as a "zero-force law" defining the state of the system (a reproductive population) at rest, when no forces are present.  

Ariew and Mohan (2009) have identified problems with the conception of selection and drift in this view.  

I realize that their criticisms may make the "theory of forces" view untenable already, but I would like to raise a problem that to me (as an evolutionary geneticist) is more profound. 

Advocates of the "theory of forces" point to the mutation-selection balance equation from population genetics as a case showing how separate forces combine in a way that is intelligible.  Mutation from allele B to deleterious alternative b pushes up the frequency of b, while natural selection pushes it back down.  Eventually a balance is reached.  In fact there is an exceedingly simple (approximate) expression for the equilibrium frequency, which is f(b) = u/s, where u is the mutation rate from B to b, and s is the selection coefficient against b.  

However, this equation applies to deleterious mutation.  For the case of a mutation event introducing a new neutral or beneficial (i.e., not deleterious) allele-- the kind of mutation underlying most changes in evolution-- the theory of forces breaks down.  When mutation introduces a new allele, acting in the role of an originating process, it does not shift frequencies generally, but specifically shifts the frequency of an allele from 0 to 1/N.  Selection and drift cannot accomplish this result: when mutation is acting in this role, it is acting alone, outside of the range of operation of other "forces".  In other words, the common currency of allele-frequency-shifting forces (mentioned above) is not a common currency.  In its role as the allele-originating process, mutation is a point process and not a mass-action "force"; it opens up dimensions along which selection and drift may act, instead of moving the system along a pre-established dimension.    

Although I accept the arguments of Mohan and Ariew about the weakness of the "force" conception of selection and drift, I am surprised that so much attention is given to relatively subtle arguments about selection and drift when the "theory of forces" view fails so utterly in a rather obvious way.  Why is this?  In one sense, the answer is simple: Stephens, Coyne and others seem to start out by assuming a variable population, so that new mutations are not necessary.  But this just begs a further "why" question.