In a comment about how races should impact a character mechanically, Brian suggested that races should have minimal impact, saying:
I think that this would help to diversify race/class combos away from what may be “optimal”.
On one hand, I agree with this sentiment. I don’t want players to feel like they need to select a specific race in order to play the class they have in mind. Do we really want all the thieves to be elves and halfings? On the other hand, sometimes this stretches my imagination. Should halflings really have 18 strength? Why shouldn’t elves be more agile than dwarves? So, if race has nothing to do with vitals, it seems as though we’re trading some consistency for options and balance. While on the face of it, this trade might be worth it, I worry that these are the kinds of trades 4e made that individually don’t seem like much, but when taken together start making the system difficult to immerse in.
I’m not sure what the answer is on this. My first thought is to place what I think are truly “realistic” vital changes for different races. This could be extreme, for example, a Halfling might have a strength capped at 14 (or -4 to STR). If I do this, there would need to be a way for a Halfling fighter to compete and play a role approximately as well as a human fighter or a half-orc fighter.
Another option is to have races that are not all that different from a “vitals” perspective. Different variation of humans may have only slight, if any, differences in vitals and would be mostly a flavor choice in initial setup. How much different in terms of stats would a Lannister have than a Stark?
One possible compromise is to adjust scores at point by using a minimum bump. For example, if the point buy from a score of 1 to 2 is a point, and the point buy between 2 and 3 is 2 points, maybe a particular race might start with a score of 3 in their prime attribute, and never go underneath it. If the point buy costs go up exponentially (or by Fibonacci or some other escalating pattern), then the point gain from a minimum jump isn’t as drastic as a bonus on top of an already maxed stat (not by a long shot if you’re using some sort of exponential growth in points), and the savings could be spent on any of the other stats.
Maybe that’s even too minimal to be worth doing.