I thought I was done with my historical research missions, alas there was one more I needed to complete. Luckily this was a short one! :-)
In the various incarnations of D&D, rangers have always had a special advantage against a certain category of creatures. The details, however, have changed at least three times.
The original ranger class by Joe Fisher (Strategic Review Vol. 1 No. 2) as well as the AD&D 1st edition ranger received their advantage against "giant class" enemies, defined (for whatever reason) as anything remotely "goblinoid" between kobolds and (actual) giants. Mechanically rangers get +1 damage per level against these enemies, so a level 6 ranger who hits a troll gets +6 to damage.
The AD&D 2nd edition ranger drops the "giant class" notion and instead let's the player decide (and the referee approve) against what monster the ranger is particularly effective. Since a specific monster has to be chosen (at least that's what the wording implies) the ability is seriously nerfed. Also the mechanic changes: Rangers now get a constant +4 on their to-hit roll instead of a damage bonus that increases by level. So on the one hand the improvement aspect is also nerfed, but on the other hand a to-hit bonus is probably a better mechanical choice overall. What's cute though is that along with their +4 to-hit bonus rangers also get a -4 reaction penalty against the creature they chose.
The "D&D" 3rd edition ranger changes things up again. First we now get a list of "monster categories" (such as "goblinoid" or "undead" or "dragon" and so on) from which the player picks their "favored enemy" as it were. This is probably the best approach yet, provided those categories can be chosen sanely in such a way that they actually matter in the campaign. Of course the mechanic changes again, but also for the better: Rangers now get a +2 on their to-hit rolls against favored enemies. That may seem like a step back at first, but we also get level-based improvement! Roughly every 5 levels the character can pick a new favored enemy, and attacks against them will again be at +2 to-hit. However, the existing favored enemies gain an additional +2 at the same time. So now a level 15 ranger can be +2 against undead, +4 against dragons, and +6 against goblinoids. That's not too shabby, although it does open up the door to a lot of fiddling during character creation and advancement. Edit: Forgot to say that the reaction penalty from 2nd edition is gone again, which presumably means that 3rd edition rangers aren't infamous killing machines of their favored enemies anymore? :-/
I shall once again ignore whatever "D&D" 4th edition did, feel free to enlighten me if you think I am missing out.
So looking at these different approaches to the "favored enemy" ability, I think it's safe to say that 3rd edition got it mostly right in terms balance: It's somewhat broadly useful, there's some potential for players customizing their characters, and yet it's not overpowering. Good job!
However, for my own games, I'd rather go with a slightly simpler procedure. Yes, some rangers may develop "new grudges" over time and so letting them choose new enemies may be cool. But why would they get better against already favored enemies as well in such a situation? To get the "new grudge" they must have been up against many of those "new favored enemy" creatures and presumably they didn't keep fighting their "previously favored enemies" all that much during that time. So I'd give the player the choice to either pick up a new favored enemy at +2 or to improve the to-hit bonus for an existing favored enemy by +2, but not both at the same time. I'd probably also go with a smaller increment, say every 3 levels instead of every 5 levels. Sound good?