glacial pace is that evolution builds systems through gradual accrual of
beneficial bits and pieces,rather than through systematic design or rapid
learning from the environment.
As a consequence of this piecemeal tinkering approach,what Wimsatt
(in press) terms “evolution as a backwoods mechanic,”designs that evo-
lution ultimately comes up with are not intended to be clean,or simple,
or easy to understand—they are just whatever worked in the particular
situations encountered.The implication for artificially evolved music-
composition systems is that,even once they do fulfill the preset fitness
criteria to a certain extent,they will likely be unfathomable in their
workings.However,this need not be a problem if one is interested in the
evolutionary dynamics or mere appearance of particular types of musical
behavior without understanding the internal mechanisms generating that
behavior.So far,most research on evolving musical systems has been
done by musicians interested in building interesting artificial composi-
tional tools,rather than answering scientific questions,and so they have
taken this pragmatic approach (see Todd and Werner 1999,for a detailed
review of evolutionary music composition systems).We can still look at
a few examples of this work,though,and get a feeling for the way in
which different types of fitness evaluators can lead to different musical
results.Appreciation for these more artistically inclined efforts can help
us make better choices of tools to use in exploring scientific ideas.Humans as Critics
Part of the reason that evolution in nature is often slow is that forces of
selection can be very noisy and temporarily ineffectual.Weak,sickly,or
just plain ugly individual organisms may still succeed in finding mates,
having offspring,and thus passing on their genes,whereas organisms with
a new advantageous trait may not manage to live long enough to find a
mate and influence the next generation.One way to speed up evolution
is thus to implement a more ruthless,strict,and observant selective pres-
sure on a population.This is the principle behind artificial selection,in
which humans play the major selective role,letting only those organisms,
be they pet animal breeds or garden flower varieties,that meet certain
phenotypic criteria produce offspring for the next generation.With
such careful supervision,large changes in traits can be achieved in a few
generations.Darwin,for instance,discussed how breeders have effected
more or less rapid accumulation of human-desired traits in pigeons,dogs,
and cabbages,noting that such artificially selected domestic races “often
have a somewhat monstrous character”(1859:16).This is due in part to
breeders’ ability to rescue interesting new “hopeful monsters”(even
those only slightly monstrous) from a childless fate and ensure that their
desired traits are kept in the gene pool of successive generations.368 Peter Todd