The methodology used to do the TAP cost stuff was, really, fairly experimental. I had to hope our system was tight enough that I could take a 32pt (or whatever) character, add 1-12 (or whatever) points of Armor and that then the %-Percent-of-Victory (POV) improvement that the characters showed against the herd would match up with the POV increase that something like -8 DM showed.
It turns out, it's looking "pretty close."
I was concerned that when I added, let's say, 8 AP of armor to a 16 AP character I 'd get a 24 AP character who had only 1/3rd of his points in attack. That's /legitimate/ but it's not what we think of as a "common build." I'll recap here that we consider 1/4th AP in attack "Low Damage" (LD) and 1/8th or less AP in attack to be Very Low Damage (VLD) but above 1/4th is "Normal" so these guys were still in the "Normal" range (wherein they get no special breaks on things like extra REA)--just the lower end of it.
But I wasn't sure if that was skewing things or not.
I also knew that there were "floor" costs for these things: no matter what the formula said you couldn't really get -8 DM vs. everything for 1 AP if you were only 8 AP as a character. These abilities have minimum costs I haven't tried to calculate yet.
So my approach was to "triangulate." To do this I'm taking the three herds and modifying their points spent in traditional defenses (Armor, DP, Force Field) and adding in TAP stuff like +2 AGI vs. Ranged, -4 or -8 DM, and so on. I deduct from the points in traditional defenses the estimated cost of the TAP defense and see if the POV changes.
If our estimations are right, it shouldn't. If our tests say that for a 16 AP character -8 DM is worth 5 AP then if the character normally has 8 Armor (for 8 AP) then he should be balanced with 3 Armor and -8 DM. Same difference.
Turns out, while I've only done the 32 and 64 herds thus far, that's turning out to be "about right." If anything some of those values on the chart below are sometimes 1 AP higher (but I've not found anything that's /too/ out of line ... yet).
So it looks like the methodology at the high end works!