I think that the main problem of Framstick problem is that in most cases the
gene pool is just too small. I realise that's nearly impossible to create
and stimulate as many genotypes as nature does, but the problem causes that
the project is still far from reality. The usual result is that the
Framstick evolution reaches its dead end and waiting for changes won't do
any good. I've recently revolved speed-oriented creature consiting of only
one stick and a few neurons, which walked aprox. 0,005! It was so
specialised it couldn't be changed in any way! In nature there's genetic
drift that randomly changes frequence of existing genes. Another thing is
that sometimes a strange genotype is introduced to the population and forces
it to evolve faster.
I tried to introduce a few genotypes in my speed-oriented population, but
it didn't do any good, since it was too rare and as a result new genotypes
got lost.
There should be something as central genotypes database, which I think FNR
is meant to be, that users retrive foreign genotypes to their popualtion
from. We should implement 'disperse proccessing' that will enable thousand
world wide users to join to common gene pool. Since SETI@home made such a
significant success, we should think over introducing such system in
Framsticks project.