Here are a few quick thoughts on ability scores.
I want to be able to allow folks to generate ability scores randomly and I want them to be able to not have to fall back on house rules like "you need 2 x 15 to have a viable character." If the ability scores are too important, then you either tend to get a Darwinian "survival of the fittest," where those with 8 CON are killed off by anemic goblins (or syphilis from a poor roll on the AD&D Harlot table), or you get the aforementioned house rules to ensure that everyone is on an equal footing. This is why, largely, I think 3.5 went to point buy. If a character is going to be around for awhile then you need to ensure that they will be relatively balanced.
I like random ability scores for a few reasons. First, they're fun to generate. You never know what will come out. Second, they tend to "funnel" choices very quickly. If you have a player who doesn't know what they want to play, you can have them roll scores and see what happens. "Oh, look -- a 16 in STR. That'd make a great fighter!" Third, they prevent uniformity. Not every fighter looks the same with random scores. Fourth, they're quick to generate. No hours poring over stats; just roll and go!
I see a few viable tools in the toolbox for evening this out.1) Ability scores increase the number of options, not the character's effectiveness or odds of success in carrying out those options.For example. AD&D gives a bonus to hit and damage for high STR. This means that characters with high STR are better in combat than those without. Period.
However, INT works differently. INT gives a different number of languages known, so high INT characters know more languages than low INT ones. However, high INT characters are not any better at speaking Elvish or Common than a low INT character.
CHA works both ways. On one hand, high CHA characters are strictly better because they have bonuses to morale and loyalty. On the other hand, they also increase the number of options (i.e., the number of employees).
I think its bad to let random chance determine whether you're an awesome fighter or a sucky fighter (AD&D: +1 to hit / +3 damage for 18/01 STR vs. +0 to hit / +0 damage for 15 STR). Then you get a need to let everyone get on the same playing field. It also introduces the Choice to Suck. Bob wants to play a brainy swordsman so he puts a high score in INT instead of STR, not realizing that he's screwed himself. Ooops.
I think its far better to allow ability scores to drive tactical options and/or operational effectiveness, not tactical odds of success. By that I mean that two characters with 5 STR and 15 STR should be equally good at hitting people with sticks; the guy with 15 STR should just have more choices in how he does it. By operational effectiveness, I mean that the score could contribute to the ability of a character to sustain themselves through multiple challenges/encounters, but not necessarily in one fight. So, a character with high CON might have more reserve points or healing surges, but not more HP than someone with low CON. They are more operationally effective but are both on the same tactical playing field.
This has a nice side effect of (A) letting higher level characters undertake longer operations (days), and (B) giving higher level characters more options but in a manageable way (that fighter can now carry an extra stone of equipment, so he has a new choice for gear; the magic-user can speak a new language, etc).
2) Allow ability scores to increase, but let the poor catch up to the wealthy.By this I mean letting characters with poor abilities catch up to those with strong abilities. For example, you could let folks roll 3d6 (or 1d20) every time they level up. If its higher than a given ability score (they need to pick before they roll) then they get +1 to that score. Obviously, trying to increase your 17 to an 18 is unlikely. But bumping an 11 to a 12 is quite possible.
In this way, you can allow random rolling for starting characters and know that by the mid-levels, things will have evened out some. It can lead to some boring characters that don't have any "dump" stats, though. I like dump stats, because you know most (good) fighters will be strong and have high CON; what differentiates them is whether they're stupid (INT dump), smelly (CHA dump), clumsy (DEX), or foolish (WIS). Or maybe all of the above.
This also has a nice side benefit of allowing the DM to occasionally drain ability scores without screwing over players too hard. If someone got a STR penalty because of a brush with a shadow, they will eventually recover to an appropriate score.
ADDENDUM: How about this... Roll 1d20. If its a prime ability for your class, roll 1d20+3. This ensures that in the long run, the prime abilities will end up one standard deviation above the norm. It also helps with the Dump Stat issue by making it harder to choose between bumping your 12 in a prime ability score and the 7 you have in something else. Intead of being a wide spread in the odds its only a few points now. I also like using 1d20 instead of D6s; if you use D6s, then players are FOOLISH to use their roll on anything other than a low to middling score.
DOUBLE ADDENDUM: You could allow (force?) folks to designate one ability score as a "weakness." Any roll to improve it would be 1d20-3 or even -6. Thus, keeping the dump stat alive and well. That dump stat will end up being one standard deviation below the norm.
With both addendums in play, I'd expect to see scores end up around here by the end game:
16, 16, 13, 10, 10, 8.
3) Soften the impact of the raw numbers.By using the -3 to +3 modifier system, you soften the impact of the 3-18. Since the ends of the spectrum are pretty rare, you basically limit standard results from -2 to +2. That's a much narrower spread than 6-15 is.
4) Build quantitative bonuses into something else that isn't random.For example, I considered letting folks use STR to add "to hit" to any weapon, DEX to be used for "light" weapons (rapiers, daggers, etc), and WIS or CHA to be used with "simple weapons" (maces, spears, etc). This would be in lieu of a proficiency system, the idea being that fighters can use any weapon well, swashbucklers will tend towards rapiers and such, and clerics can still smite the heathens with their maces.
The problem with this is that it violates rule #1 -- a fighter with 18 STR (+3) is strictly better than one with 13 STR (+1). Also, it leads to a mild numbers wank fest. By that I mean that you're basically giving everyone +1 or +2 to hit with a certain kind of weapon (and you're probably giving all the monsters +1 or +2 to AC), so its a wash. So why not just tell them that they can only use that kind of weapon, give them +0 to hit with it and -2 to hit with everything else, and adjust monster defenses accordingly? Its less math for everyone and gets the same point across.
The wank fest gets worse if you allow ability scores to increase, because now the player who doesn't pump their "to hit" score as much as you think they should falls further and further behind the power curve (4E has this problem). Don't give folks the choice to suck.
So, if you want fighters to be better at hitting stuff, just say "fighters get +1 to hit everything." If you want mid-level characters to be better at hitting stuff, then say, "when you hit level 5, everyone gets +1 to hit." Remember, we're talking about making randomly generated stats viable here, so its not a strategic choice between pumping STR for damage at the expense of CON for durability -- we're talking about getting lucky or not lucky.
Here's a sample Ability Score system that might address all those concerns:1. Roll! Roll 3d6 in order for each ability score.
2. Primary Scores. Designate two ability scores as Primary scores. There are two rules here: They must be at least a 9 and they must match up with a class (if using the Base 3 classes only, then at least one must match with a class). Son, one from CON/INT/WIS and one from STR/DEX/CHA.
DM's OPTION: Allow players to swap any two ability scores or arrange scores to taste. This ensures that everyone gets the primaries that they want.3. Flawed Scores. Designate one ability as a Flaw. You may drop your Flawed Ability by 3 points to add 1 point to both of your Primary abilities. You may not drop your flawed score below 5 in this way.
DM's OPTION: Players may designate another two ability scores as Flawed in order to raise a third ability score to be a Primary. In order to raise ability scores now, the player must drop TWO flawed scores by three each to give +1 to all three primaries.4. Increases. Every other time you gain a level, roll 3d6 for each ability score (4d6 for your Primary scores; 2d6 for your Weak score). If the number showing on the dice is greater than your current score, increase that Ability Score by one.
DM's OPTION: Every even level gained, roll for CON, INT, and WIS. Every odd level gained roll for STR, DEX, and CHA.5. Derived Statistics. Each ability score has a derived statistic associated with it. It is usually determined by 4 +/- MOD or perhaps 1 +/- MOD.
6. Chances of Success. You receive +2 or perhaps +3 on a D20 and +1 on a D6 when attempting a task linked to a Primary Score. You receive -2 or -3 on a D20 and -1 on a D6 when using your Flawed score. This applies to saving throws, to-hit rolls, etc.
Example "build:"Joe rolls 3d6 in order and gets the following:
STR 14, CON 9, DEX 7, INT 8, WIS 15, CHA 11
He wants to play a fighter, so he designates STR and CON as his prime scores. He debates about where to place his Weak score and ultimately decides on INT. He drops INT by 3 to a score of "5" and adds +1 to his STR.
When he gains his second level, this fighter will roll 4d6 for CON, 2d6 for INT, and 3d6 for WIS. Any score on the dice higher than his current scores will result in a +1 increase.
This fighter is good at physical challenges like breaking open stuck doors or wrestling. He's bad at math and foreign languages.
ADDENDUM - MATH
I just ran the numbers on the two systems (3d6/2d6/4d6 vs. 1d20/1d20-3/1d20+3) in Excel and these are the summary findings.
As one might expect, the methods using multiple D6s act very strongly to bring up scores below the average (10.5/7/13). Once this average is reached, they slow down growth significantly compared to the D20.
Average Stat: If you start with 10.5 (strictly average), these are the number of rolls required to get to 13 (a +1 modifier):
1d20: A hair under 6
3d6: A hair under 8
After 10 rolls, 1d20 will get you to a final score of 14.5. 3d6 will get you to 13.481.
Low Stat: If you start with a 5 (lowest score), these are the number of rolls required to get to 9 (-0):
1d20-3: Just under 8
2d6: Just under 9
After 10 rolls, 1d20-3 will get you to a final score of 10 and 2d6 will get you to 9.279. 2d6 is better than 1d20 until you reach the average (7) around roll 3 or 4, at which time the 1d20-3 becomes the better bet for advancement.
What does it mean?If the intent is to bring low scores up to at least the average, using multiple D6 is better. If the goal is to increase scores which are already at least average, then using 1d20+/-3 is better.
Also, assuming that a player has three "average" stats (10.5), it is more likely than not that one of them will be increased to 13 (gaining them a +1 bonus) after 2-3 sets of rolls. So, you need to let folks roll at least two or three times on the same set of three ability scores before the player will actually see significant improvement.
This also has slow and measured effects. You'd get more dramatic effects by rolling 3d6+3 (0r 4d6) for the middle scores and 2d6+3 (or 3d6) for the low scores. That'd be appropriate for a high powered campaign.
All in all, the methods are pretty equivalent, with the D20 based methods producing slightly higher scores and the multiple D6 methods ensuring that low scores are brought up quickly.
