Resistance

[Catwalk]

I think I'd prefer discussing it through thoroughly and then making a decision that we don't plan on changing later (still possible, of course). It's a lot of extra work trying to balance things for one resistance system, and if we have to balance it separately for both systems right away that doesn't sound appealing.

[Asfex]

How about gaussian model?
description: spell parameters: A:strength B:spellsafe. C:shift
say: confusion is casted by hero with -3 spellsafe item on spearmen (6 crosses):
A=(5+3)*1,4; B=-4 (native confusion spellsafe); C=0

then do A (8*1,4) times this:
start with (Unit resistance +B) (=2)
random add or ubstract 1 cross or do nothing,
check if zero is obtained.
if yes (zero at least once)- the spell success, if no -the spell fail.
end.

the step may be per 1/2 cross,
remarks: i about distinct the spellsafe from items and spellsafe from spells.
the fisrt is increase "probability"(linear in terms prob.sux/prob.fail), the second -"possibility", (non-linear).