2008/10/04

Darwinism is slow

Darwinism is the best framework we have currently to undertake the problem of evolution of organisms, but it may be incomplete, not enough powerful. When trying to resolve an extremal problem in physics, for example finding minimum energy states, using computational approaches like Monte Carlo simulations, one thing comes evident: high-dimensional spaces are slow to explore. This is known as the "curse of dimensionality" ( http://en.wikipedia.org/wiki/Curse_of_dimensionality ).

Darwinism is an extremal multi-dimensional method to evolve the species. Dimension in Darwinism means every different possibility available to a life being to evolve. And that number is really huge in the simplest case of a cell even a virus. Nature has the advantage of time. It takes millions of years doing evolution steps. On the other hand, today supercomputers are crunching numbers for months or years as much (1 or 2) to solve one problem, but computers may complete cycles way fast (10 GFlops implies 1 cycle per nanosecond). So, I would say that natural selection counts with a power similar to current supercomputers. If natural selection did it (to evolve species), supercomputers should be able to do it. The fact is that currently huge-dimensional extremal problems cannot be resolved successfully no matter how powerful the supercomputer you use.

It is correct to say that Natural Selection is greatly improved by sex, because sex will upgrade an organism's offspring with the advances achieved by other congeners. So evolution is not just a blind random trial and error mechanism but a sex-oriented one. On the other hand supercomputers also have a "secret weapon" in this "nature vs engineering race". In the Monte Carlo algorithm an "Importance Sampling" or "Score" may be applied to accelerate the convergence of the process. This way, supercomputers may also be oriented to the solution as sex is helping Natural Selection. In other words and roughly speaking: Monte Carlo simulations and Natural Selection remains equally powerful, but Nature can do it meanwhile Supercomputers cannot.


In the above video ( http://www.youtube.com/watch?v=oCXzcPNsqGA ) we find artificial organisms created in a computer by natural selection processes. The point is these organisms live in a simplified universe, so their evolution space dimensionality is very low. In Nature, the evolution space dimensionality for just one cell is huge. So, Natural Selection works, yeah, but too slow for itself.

Conclusion: maybe some day scientists will realize that the Darwinist Natural Selection is not enough fast to explain life evolution. And here is exactly the point where Creationism makes sense.

No comments: