[atltais]

This is mostly used in components and the like on Star Conquest, but it's still somewhat relevant here.

After setting up a baseline, we generally create algorithms that determine the 'level' of something automatically. After that, reasonable statistics are assigned. After playtesting, if balance issues (or other, similar issues) arise, the statistics are reevaluated and the algorithm is (potentially) refined.

Now, on other (more relevent) things, we generally go down to a more generic metric that can be used between a variety of things. (for example, weapons, damaging spells, healing spells, regeneration, damage over time, and so on) These are also algorithmically balanced, and we have a number of fine tuning measures. (and, if need be, explicit overrides in all cases)

Most of this is relatively transparent to players, and I think it works fairly elegantly. In short, we solve our problems with mathematics.