Random drift/walk, backwards or forwards. As long as you don't come off the rails you'll end up somewhere.
There has been a little flurry on the internet about the question of the Zebra's stripes: What function do they serve and how did they arise? Evangelical atheist Larry Moran offers a possible alternative to the adaptive explanation:
There's a fifth possibility: maybe there's no reason at all and stripes are just an evolutionary accident.....The point is that the prominence of stripes on zebras may be due to a relatively minor mutation and may be nonadaptive. That's a view that should at least be considered even if you don't think it's correct.
This answer no doubt fits in very well with Larry's philosophical position as does his interest in "random drift" as one of the mechanisms of evolution; he might be right, after all he's the expert. But he seems to have little cognizance of the actual role of randomness in the greater scheme of things: For standard evolution to have the slightest chance of getting results the "random diffusion" of evolution must be working within a very constrained space indeed. Yes, evolution may go "backwards" and "forwards" at random, but for that to happen the current theory requires "rails" or "constraints", in terms of the fitness space*, to be a given. This is something that Larry just hasn't wrapped his head round. Am I being stupid in wondering why?
* If the fitness space is an "information" given then it follows that the problem of finding life has effectively been solved before the computation of evolution has started; This view is not a horse I'm backing. See http://quantumnonlinearity.blogspot.co.uk/2014/09/declarative-cognition-vs-procedural.html
The General Theory of Evolution: But it only works if you've got rails