[Right Wing Professor]

But doesn't that describe something that is stable and driven towards stasis and not change?

Well, locally stable. But introduce a new degree of freedom, and you won't be stable along that axis. And existing degrees of freedom won't be stable if you change something.

[Right Wing Professor]

One other thing I wanted to say; if you've done a multivariate optimization, you realize how complicated the surface of a graph of comparatively simple independent functions becomes once you put in more than a few dimensions. Say you do a non-linear least squares optimization to a data set with ten or so independent variables (this isn't a big deal if you have a well-conditioned problem). What you find is that the error surface is frequently enormously complicated, and your fit swings all over the place, exploring all sorts of valleys and local minima, beofre you find the globally optimum solution, which may take a very long time. I see evolution as being like this; as species become more and more fit, they can go way out to the extremes of some characteristics, and then head off in an enirely different direction. They can bifurcate as a result of small local differences in the environment, and end up in totally different places on the graph.

[AndrewC]

But introduce a new degree of freedom, and you won't be stable along that axis. And existing degrees of freedom won't be stable if you change something.

But then why are there still monkeys?