Ok, here's a specific proposal: right now, the way an attacker
is chosen is that all of the potential attackers are considered
to be equally likely. An equivalent way of doing this is to say
that given two sides A and B, the probablity of selecting an attacker
from side A is sizeof(A)/(sizeof(A)+sizeof(B)), while the probablity
of selecting an attacker from side B is sizeof(B)/(sizeof(A)+sizeof(B)).
Once a *side* is chosen, the actual attacker is then selected randomly
from that side, and the target is selected randomly from the other side.
This has the same result as the simpler method in use now. The advantage
of this formulation is that you can play around with each part separately.
I'd proposed that you choose which side gets to attack based on the
square root of the size: sqrt(sizeof(A))/(sqrt(sizeof(A))+sqrt(sizeof(B))).
If the two sides are of equal size, then this makes no difference. If
the sides are unequal, then the smaller side gets a dis-proportionally
large number of attacks. (If using the "speed" modification, then
the "sizeof" a stack is the sum of the speeds, rather than just a count
of units.)
Big mobs can still overwhelm (as it should be), but smaller high value
stacks become more viable.
I also like the suggestion that nobles only control some limited
number of men/creatures.
-- Bron Campbell Nelson bron@sgi.com or possibly uunet!sgi.com!bron These statements are my own, not those of Silicon Graphics.