Well, it does make some sense. Let me explain:
* A stack with 1 dragon attacks with an average AF of 500.
* A stack with 1 peasant leader and 1 dragon attacks with an average
AF of 250.
+++ This is a bad thing for the 1 peasant leader + 1 dragon stack.
* If the opposing stack has N units in it, then the stack with 1
dragon in it attacks with frequency 1/N+1 and is attacked with
frequency N/N+1.
* The stack with the peasant leader is attacks with frequency
2/N+2 and is attacked with frequency N/N+2.
+++ This is a good thing for the peasant + dragon stack (it
gets a bigger share of the total attacks).
* Both the dragon stack and the dragon + peasant leader stacks
defend the same, since all attacks in both cases are directed
against the dragon.
+++ This is neutral.
The question is, does there exist a case where the loss in average
attack factor outweighs the advantage in attack frequency? My best
guess at this point is no. Certainly for the cases I've simulated
there doesn't seem to be a disadvantage (if I run enough trials :-).
I'm still thinking about it, but there doesn't seem to be a simple
case where this is really a problem.
-- Scott T.