fun with Bitozoa2

The alife application available at:

http://www.bpp.com.pl/bitozoa2/bitozoa2.html

might seem pretty primitive, just a bunch of dots
chasing each other around a toroidally connected
square playfield.

This turns out to be anything but the case, and so
I'm writing this posting to entice other people to
explore it in greater depth.

If you don't meddle with a run once it's
(re-)started, Bitozoa2 gives you the opportunity
to run a simulation I just ran, by duplicating my
starting random number generator seed and my
initial settings.

I cannot, unfortunately, give you the ability to
replicate what I'm seeing and describing here,
because somewhere early in the simulation, I
meddled. The carnivores were about to go extinct,
so I dumped 500 more random ones in to keep things
interesting, an act you cannot replicate exactly.

Interesting indeed the racing dots have become,
since I have a computer available which normally
has nothing useful to do but browsing, so that I
can leave the simulation running for weeks at a
time with little penalty.

I messed around with four of the startup parameter
settings:

Changed the default starting 5 neurons to 20 to
support more complexity at the cost of lots more
initial unfitness.

Changed the default flagella setting to around its
usual evolved setting, 16 degrees apart or so,
because I wasn't very interested in watching this
happen again.

Changed the default mutation rate from 0.1 to 0.03
(since I had time to kill and didn't need to eat
dessert with the first course of the "meal") so
that favorable mutations wouldn't so easily be
lost through a combination of genetic drift and
unfortunate re-mutation before they had a chance
to spread into the population.

Disabled asexual reproduction, because it has lots
of effects I don't like, like whole populations
that never have to find another member with whom
to mate as part of their evolved behavior.  This
is probably _why_ the carnivores were about to go
extinct; carnivores have a hard time catching the
hints to chase herbivores to feed, and other
carnivores to breed, until really dense
populations of herbivores exist to make the first
part available more or less by accident by running
over unintelligent herbivores.

Things I've seen since:

A period when the young carnivores were _afraid_
of the herbivores, not all that difficult to
understand when you realize that mature herbivores
can trample immature carnivores.  Thus the "cub"
carnivores need to learn to avoid the big "bulls"
while snacking on the little "calves" until they
are mature "lions" themselves capable of taking
down a "bull". single-mouthedly. This isn't an
easy behavior to evolve because it is quite
complex, and I seem to have tuned in when the cubs
had learned to be very tentative in approaching
any herbivore.  This was detected because the
usual sexual reproduction act involves two adults
colliding and being replaced by a perfect circle
of few to over a hundred offspring going in random
directions. The perfect circle was suddenly
becoming imperfect because it "lagged away" from
any concentration of herbivores encountered.

A change of the population balance from slightly
more carnivores than herbivores to perhaps four
times as many herbivores as carnivores, as the
herbivores evolved the skill of running away from
the carnivores, and the carnivores weren't yet
very good at pursuit.

Not in this run, but in others, the carnivores
evolving "base sticking" behavior, where a
carnivore would start either circling around or
lurking in a shrub bit (inedible to it) in hopes
a naive herbivore would pay more attention to
getting dinner than to the chance of being dinner.

Herd and pack behavior, where the mature "bull"
herbivores move in groups (without touching lest
the replication behavior be triggered), apparently
in hopes of trampling a "lion", which they could
do apparently by running over it one by one, even
with potential loss of life: altruism evolving, if
indeed that is what I am seeing, is worth a
headline.

Similarly, and to me this is the most hoped-for
spotting of all, immature carnivores, instead of
just radiating from the point of birth as before,
are starting to clump together from sub-arcs of
the reproductive explosion, and travel as
packs, which very likely improves their hunting
efficiency by blocking off more routes of escape
for the herbivores.

Another headline grabber was the carnivore visual
angle evolving to be narrower, while the herbivore
visual angle evolved to be wider, exactly what you
would expect from studying eye placement in living
beasts.

Bitozoa2 -- worth more than a glance.

FYI, I don't live in Poland or have any relation
to the developer except satisfied user and
occasional bug reporter.  [Now about that bottom
statistics graph that is set too high so it
overlaps the top graph's range indicating
floats...]

By the way, it is not at all obvious that you have
to PAUSE the simulation to use the statistics graphs
and some of the individual entity exploration tools.

xanthian.