At various times, my group has used rolling methods and point systems. Some of our solutions are presented below.
Points:
Most recently, we've been favoring the point system. However, it does tend to produce characters with very similar stats, and the "penalty" for high stats only reinforces this.
Our most recent campaign we used a modified "28+2" system. Basically, use the points to buy stats as normal. Then you get 2 more ability points to add to the stats directly. (+1 to two different, or +2 to one) The +2 essentially reduces the cost of getting a single high score, while not giving a "discount" on getting two high scores.
Rolling:
We've found that the standard 4d6 drop the lowest produces fairly average characters. It is rare to have scores outside the 10-15 range.
To alleviate this problem, I devised a system I call "Catch-22". (actually 2 sets of 11, but "2 x Catch-11" doesn't flow as well.)
The first step is to roll 11 dice. Group the highest 3 rolls as the 1st stat. The next 3 highest as the 2nd stat. The next 3 for the 3rd stat. The lowest two are ignored.
So if you rolled: 6,5,4,3,2,1,6,5,4,3,2
You'd end up with 17 (6,6,5), 14 (5,5,4), and 10 (4,3,3)
Roll the dice again and group them the same.
This method usually produces 2 scores in the 15-18 range, 2 in the 11-14 range and 2 in the 7-10 range.
The average should be around 13 per score, but the variance is much higher than 4d6 drop the low, so getting a set like 18, 18, 13, 10, 7, 4 is much more likely. (between sets of scores, I like similar sets of averages. within a set of scores, I like lots of variance)
A question often asked with this system is why 11 dice are rolled. The number was arrived at by trial and error, so basically: 10 was too few, and 12 was too many.
If tougher characters are desired, you can use 12-14 dice per set instead of 11. For "challenging" (or maybe "challenged" is a better term) try 10 or even 9 dice.
Relative Strength
Whatever rolling method chosen, one of the biggest problems is the differential between toughest and weakest characters. It is certainly no fun to play a fighter that is weaker and has fewer hit points than the party wizard.
To remedy this, we implemented a "stat draft" in one campaign.
To begin, the players (and dm) have to agree on the basic power level you plan on using, otherwise it will be hard to agree on whether a set is "acceptable".
First, we rolled 2 sets of stats for each character (12 sets total) using the Catch-22 method above. We compared all the sets and found 6 sets we all agreed were pretty good. (making sure they were all pretty equal) If needed at this point, roll more sets until you have enough sets that everyone is happy with. Then randomy distribute the sets.
The random distribution pretty much ensures that all the sets will be roughly equal. (it is unlikely that a group of players would agree to have a set in the pool that is clearly better than the rest because they aren't ensured of getting it. For similar reasons, they wouldn't want a clearly inferior set in the pool for fear they might get stuck with it) Again, play with whatever power level your group is comfortable with, and as long as all the characters are the same relative strenght, they should all be able to contribute to the the group equally. (nothing worse than feeling like your character is excess baggage)
This achieves multiple goals:
1) Everyone gets to roll lots of dice.
2) The sets of scores, while not identical, will be roughly equal
3) The random distribution means that the best roller won't always have the toughest character.
Overall Power Level
As for the power level we play at, we realized that the tougher the characters were at first level, the harder it actually was for them to survive. If the characters have an easy time dispatching enemies, the tendency will be for the DM to increase the EL's of future encounters. The problem with this is that these tougher creatures will be more likely to kill PC's off quicker (some times with one hit)
This is due to the fact that tougher characters at 1st level are mutants. They might be able to deal out damage like a "normal" 2nd or 3rd level character, but they still have 1st level hit points. We ran into a similar problem in 2nd edition when the Skills and Powers book came out. I made "Super Elf" and was able to kill an Ogre in a single round, provided I got to act first and hit with my attacks. Otherwise, the Ogre would likely kill me with a single shot. For this reason we usually have 30-32 point characters (28+2 fits in this range)
Umbrage
