I think bell curves can have a place in gaming. For things like rolling up stats or determaning damage a curve makes sense. But certainly not for task or conflict resolution.
R
I agree. If I were rolling stats, I wouldn't be comfortable with merely rolling a d20, as it would produce as many STR-1 characters as STR-10 and STR-20 characters. In that case 3d6 would make more sense.
And I'm not saying that bell curves are inherently bad. I'm saying that (a) their purported "advantages" seem overrated to me; and (b) their limitations are usually misunderstood by even veteran game designers. Bell curve mechanics are extremely sensitive to modifiers and few game designers are mindful of this, in my experience. The result is usually a crappy game. The system either remains broken (and therefore no fun to play) or it gets weighed down with additional clumsy mechanics designed to correct the problem. An example of the former would be Classic Traveller, with the system-ruining Book 4 weapons and character generation systems. An example of the latter is GURPS' clumsy parry/defense roll.
I also note that bell curve systems *seem* to have more room than they effectively have. Take the GURPS 3d6 system -- it has 16 "slots", from 3 to 18. Ignore automatic success/failure.
If the base success number (average person, average task) is set at the midpoint (10 or less in this example), then it takes a net modifier of 8+ to "crash the system" (i.e., make success or failure automatic). That appears to be a pretty large range, and a designer might well have lots of modifiers. As GURPS does.
But in reality, a much smaller net modifier will make success or failure *effectively* automatic. A detrimental modifier of 3 will result in failure 85% of the time and a positive modifier of 3 will result in success 85% of the time.
In other words, there is seldom much real world difference betwee (say) a +3 and a +6 in GURPS. Particularly when one considers that a 3-4 is an automatic success and a 17-18 is an automatic failure.
It's pretty easy in GURPS to get a net modifier of 3+.
Compare this with a d20 mechanic. Assume you replace the 3d6 roll in GURPS with a d20 roll (something I've long advocated). Now, it will take a net modifier of 7 to make success or failure effectively automatic (85% chance). The d20 allows more than twice the range of modifiers.
If you redefine effectively automatic to mean a 90%+ chance, nothing really changes. The 3d6 mechanic is broken with net modifiers of 4+; the d20 mechanic is broken with net modifiers of 8+.