I know that some people like the randomization that comes with rolling your stats. There main philosophies boil down to:
1. Total Point Buy is as even as it can get. You get full customization options and you’re never left behind. This is the character-first method.
2. Rolling a character helps you build a character organically. You create a character from what you’re given, rather than from some starting concept. This is the stats first method. A subset of this method believes that you should generate scores completely randomly, where the averages make you a normal person, so that you can start from humble beginnings or overcome serious deficiencies.
I think there may be a best of both worlds. I think in Lost Worlds I will provide a method of generating random scores using approximately the same block of points. The main problem with pure randomization is that, if ability scores matter – and they will, you can be left with a character that is severely under powered or overpowered for the game. I know there are those out there that’ll say that shouldn’t matter, that we’re playing a cooperative game and if one player rolls incredibly – well that’s just better for the party. I think it does ruin the fun a little bit if we’re trying to play a game where we share the spotlight and it feels to everyone like half the party are sidekicks and not heroes.
I think part of this goes into play with the idea of “Difficulty Levels.” If you agree to play on Hard Mode, you have to generate a set of stats (using an approximated point buy system) and go by those stats where they land. I’ll talk a little bit more about the difficulty levels in Lost Worlds, but I think of them as a social agreement between the players and the GM that can be different from player to player.
Standard mode will be something along the lines of 25 points to build with. Hard Mode may be the same 25 points, but randomly generated scores using those 25 points, and randomly assigned to the 8 ability scores. Hopefully, this will make both philosophies happy, without setting one group too far ahead or behind in the power curve.
For the randomization method, unless I think of something a little bit slicker, I’ll probably generate a few arrays and the player taking random scores will roll to see which array they get, and then roll again to see how those scores are distributed.