Paizo posted the playtest for the new race-creation system for an upcoming Pathfinder hardback last week, to general buzz on the internet. The dominant note I saw was how easy it is to make a hideously powerful race that’s considered balanced by the system: for example, a race with +8 Intelligence, -2 Strength, -2 Charisma, and spell resistance equal to 11 + HD is priced the same as any basic character race. Part of the issue is not accounting for uber-specialization while another part is potentially undervaluing rules elements that can be exploited (Sean K. Reynolds’ point-buy feats list was mentioned as significantly undervaluing Natural Spell for druids, for example). But the biggest part of it, I feel, is just the danger inherent in trying to retrofit a system with the low granularity of D&D with high-granularity modifications.

Back in the early days of 3.0, I attempted to convert the system to point buy. I proceeded from a similar assumption as Paizo seems to with their race building guidelines: all classes are basically balanced against each other, so by accounting for their similar features one can also come up with valid pricing for their special features. The document did come out looking fairly authoritative, was easily understood in play, and could be used to recreate the existing classes pretty accurately. And then the cherry-picking began: wizards with barbarian HP, rangers with full Druid spell progression and shapeshifting, and so on.

Ultimately what I discovered was that, if D&D elements are balanced against each other at all, it’s because they are package deals. You can give classes like Druid a bunch of situational abilities like Resist Nature’s Lure and Trackless Step because they aren’t optional, the same way you can give Dwarves a +2 Appraise vs. stone or metal items. Players that want everything else the element provides will accept the situational abilities as what they are: lines on the character sheet that are only useful in rare circumstances. But as soon as you give a player the ability to trade 2-3 situational abilities for one consistently useful one, the dross falls by the wayside. It’d be suboptimal for a Druid to keep Resist Nature’s Lure if it could be traded in toward more attack bonus or HP just like it would be weird for a Dwarf to keep his miscellaneous bonuses to saves and appraisal when he could get a pretty beefy SR instead. That’s pretty much the benefit of a level/package-based game system: you can get players to take flavorful but less useful traits along with the things they really want, and nobody feels like they’ve made a bad choice.

The Pathfinder race system is, at least, intended to only be available to the GM, so it should have fewer optimization problems than my point-buy class system did. However, it does proceed from the faulty assumption that, because all the elements (races) are basically balanced against each other, there must by some mathematical solution to award value to each individual element. And that’s just not the case. A +2 Appraise can’t be accurately figured because it’s a flavor addition to Dwarves; it’s either worth nothing to you, or is a fun way to sometimes show off your Dwarfiness. But it’s certainly not worth the same as a free skill focus (no, really, the only reason in the world to take that instead of Skill Focus would be if you also wanted to take a Skill Focus in Appraise and become the best appraiser of metal and stonework ever).

In the long run, the best way to balance races is probably the one that has worked forever: you make one up, give it some abilities, and then start polling players to see if they’d play it. If it’s about equally attractive to your players as the other races, then it’s balanced, no matter what its actual array of abilities looks like.