Friday, May 13, 2011

An Interesting Metric

Working On Right Now
I am testing "packages" of TAP attributes--groupings of stuff like Fast Company or Science Agents to see what they empirically test at vs. what the apparent "additive" cost comes out to.

What I Discovered
I've discovered a few things in this process, nascent as it is.

  1. Buying in Bulk Pays: That's the nice way to put it. The other way to put it is this: the individual costs of the TAP attributes such as a combination of, say, -4 Damage Mod, +1 to Dodge, AGI and Dodge both work vs. Range, and +5 Initiative test at, let's say, 50% of your AP (so 16 AP for a 32 AP character). However if you "buy them off the list" they add up to around 18 AP or so. I'd seen this before--but testing it more rigorously it seems that there is some "overlap" or "rounding" or something going on. It is also possible that for these large groupings the remaining APs are so low that something else is at work. What could that be?
  2. Sometimes An Extra Attack Ain't 'Worth Anything.' This was the shocker. An extra attack (a full-force strike for 1 REA, allowing the character 3 strikes in a Round instead of the customary two) is one of the more expensive things you could buy. When I ran the numbers for a group of TAP attributes (Fast L1 + Cyber Dodge and Fast L2 + Cyber Dodge) I discovered something: the Fast L2 had the extra strike and it just came out with an identical win-percentage to the L1. I was like that isn't possible. Examination showed nothing broken. The Fast L2 character was, indeed, striking more often. Both of these were around the 32 AP range so while most of the points were spent on the Fast (defensive) attributes they weren't, like, down to 0 APs for an attack--but: then I realized: the Fast L1 character was winning as often as the Fast L2 character but they were taking, on average, 4 rounds longer to finish the fight. What that meant was that our metric (Percent of Victory--PoV--was simply missing another key element: how long did the fights take).
I believe--on examination of the fights--that the issue was this: because the damage for the AP level was so low the two characters generally won whenever they could hurt their opponent (they won almost never against the Full Armor guys--like 0 to 17%). However, for the most part, the fights weren't in doubt percentage wise (they won, each, about 54% of the time)--however, the guy with the extra attack simply won faster.

Now what do we do about that?

The first answer is "not much"--that was a hypothetical test and it's not an actual package we're vending. But that same scenario could come up again and the real answer is: we'd manually adjust the costs to take into account the longer battles (probably by making the Level 1 version cost 10-25% less than it currently tested at). The exact number we'd choose would be based on thinking about what extra long rounds really "cost" the character (more attacks from other parties primarily).



  1. Just a note: don't write off time-to-finish in combat. In terms of the actual game scenario, sometimes those things matter (rescue the sacrifice, stop the bomb, so on).

  2. We're not--it's just not one of the things I was /generally/ tracking. It never occurred to me that it would be the deciding factor in an otherwise balanced Percent-of-Victory evaluation.


  3. Yeah, I can see that'd be a bit odd; I'd have expected the going-to-win-but-just-take-longer scenario to only come up in cases where you could see, on the face of it, that the character was going to win (usually because the opposition was incapable of hurting them barring Hail Mary's). I'd have thought those would be outlayers.