As she was hiding in the woods during 2nd world war, Rita Levi Montalcini, '86 Nobel for Medicine, discovered some important things about optimization and evolution.

In fact, she noticed that bee's brain is morphologically speaking perfect, and for this reason it didn't evolve.

Otherwise human's brain, due to its imperfect shape, continued developing till the present homo sapiens sapiens form.

"Evolution is nothing but the gradual insertion of more and more freedom into matter. In that sense, we humans, custodians of the most advanced form of Mind, are simply the most free material entities in the universe."

S.Kwinter_"Leap in the void"

According to darwinism, random modifications produce some relevant variation in a x-number of stochastic reiterations; then natural selection process favours those changes that tend towards certain optimization.

Considering the "normal" range of variability inside species, these relevant modifications could be seen as "mistakes" in comparison with the "rule". in this case we could speak about error finding, both towards optimization or desoptimization, in any case, what G.Deleuze called anomaly.

That's exactly what is considered progress in non-linear development models.

S.Kwinter, in a an article about a "general theory of models", explains how according to catastrophy theories, as in thermodynamic systems, the posibility of escaping from the bounds of the system depends on the distance from the stationary equilibrium state of the system.

“Now it is a basic tenet of the laws of thermodynamics that in order for something to happen within a system, there must be first a general distribution of differences within the system.”

Landscapes of Change

So probability of producing a relevant change basically depends on the excitation status of the system and on the number of reiterations or tentatives.