Traditionally in Dungeons & Dragons, there is a relatively sharp divide between inherent character qualities and the main capabilities that interact with important game systems. If we were talking more abstractly about what makes up a person, one might see nature and nurture in this split, but that is not quite it, because there are many aspects of a character that would most properly be considered nurture (such as AD&D secondary skills) which have little impact on game capabilities outside of the occasional edge case.
As you have probably already guessed by now if you have any experience with D&D, this division is realized by ability scores and character class. The former may have some influence on marginal effectiveness of the later (such as a dexterity bonus to missile attacks), but it would not be unreasonable to say that ability scores represent a characters raw potential whereas class represents something more about a character’s particular life experiences and training.
A major benefits of this approach is that two characters of the same archetype can be differentiated without needing to resort to more complicated trait or skill systems. For example, traditional D&D rules support both charismatic, leader fighters and brutal, strong fighters in a way that gives mechanical weight to the distinction without undue complexity. This is possible because most important game capabilities are either located in class (attack bonus, weapon proficiency, spell casting, thief skills) or static across all potential character build options (such as the 1 in 6 search chance).
In this older approach, ability scores are relatively static, baring supernatural augmentation. If a first edition fighter begins with 10 strength it is totally conceivable that the same fighter will still have 10 strength at 20th level, assuming the character manages to survive that long. Despite only being of average strength, this imaginary fighter will still be competent, because increasing combat effectiveness is tied mostly to class level.
Another approach is to treat stats as more direct measures of game effectiveness rather than seeing them as a way to describe the totality of a fictional person. Third Edition takes a few steps in this direction with its regular (every fourth level) stat increase, though this is small enough that it can still be understood as minor fictional personal growth, staying within the “nature” conception of ability scores. Other games take this further, such as Green Ronin’s Dragon Age (which I have previously discussed in more detail), where each level an ability can be incremented and, for example, stealth is just a dexterity test, meaning that characters of any class can improve into that area.
The recently released Fifth Edition D&D version 0.1 basic rules are closer to this second style. Very few bonuses are directly traceable to class. For example, there are no class-based attack bonuses. Instead, there is a general level-based “proficiency” bonus which applies to different things. Proficiency in something can be awarded by background, race, class, and presumably feats (though the details of feats will not be revealed until the publication of the Player’s Handbook). The proficiency bonus ranges from +2 to +6, and otherwise the ability score bonuses apply to all tasks. So, a fighter could gain proficiency in thieves’ tools (thus being able to apply the proficiency bonus). A wizard could get better at shooting bows by increasing dexterity during one of the regular chances to increase ability scores. These opportunities happen approximately every four levels, though it varies by class, but they grant a +2 or two +1s, making the improvement more impactful than how it is done in 3E.
At first one might say that this is not all that different than the older approach. It’s just a few bonuses, right? If you actually look at the numbers though, the way things work out is that most characters, if the player cares even a little about mechanical effectiveness, will end up with 20s (which is the max) in the ability scores most important to the class. At this point, if you can assume that all melee fighters will end up with 20 strength and all wizards will end up with 20 intelligence, the ability score system is no longer attempting to describe the “nature” side of the equation.
This is not a bad design decision, but I think it might be frustrating for players expecting something with a feel like 3d6 down the line. That set of random numbers, in addition to providing game bonuses, is also something like a personality profile. That use of ability scores has been marginalized, which may be part of the reason that other personality mechanics (ideals, bonds, flaws, inspiration, etc) were added.
Dark Souls uses a similar scheme. Class in that game determines only initial stat values, starting equipment, and starting spells. It is possible to level a “warrior” character into, essentially, a sorcerer by improving attunement and intelligence scores. The system is elegant and easy to understand. It is probably slightly more flexible than is appropriate for a tabletop game, which will almost certainly involve multiple PCs cooperating. Dark Souls, however, being at heart a one player game, needs to support accessing different strategies through the same fictional avatar. Though there are some cooperative multi-player features, most of the game is not about a party of adventurers solving problems by working together. That said, the way the stats are improved, and what they affect within the game world, are almost directly appropriate to the tabletop context, and are interesting to compare the “increase ability scores” models that can be seen in Dragon Age and Fifth Edition D&D. For reference, the Dark Souls stats are vitality, attunement, endurance, strength, dexterity, resistance, intelligence, and faith.
In The Final Castle, I have also settled on an approach where stats are more about game capabilities than complete representation. Like Dragon Age, the scores are much smaller, what would be probably recognized as the “modifier” in D&D terms, and each time a level is increased one is incremented. The basic array is described here (strength, dexterity, constitution, magic, perception, charisma), though the mechanics have changed somewhat. Starting values range from 0 to 3, with no possibility of negative modifiers, as with the original Gravity Sinister system, and the 3-18 number is no longer recorded, though it is used to determine initial values. I am also considering narrowing the scope even more for some of the abilities, to make it even clearer what is going on, such as replacing perception with aim. The current improving bonus from perception sits uncomfortably with the scale of the d6 skill rolls (once perception reaches +4, the basic skills such as search and listen become optimal, with only 16% chance of failure).
For my pirates game, I did away with the 3-18 too. You can have a +0 or +1 to start and they are modified stepwise from +1 to -1, or if you go below -1 in CON, dead.
This design decision was made to countermand the variance in stat obsession between players at my table; two of them are happy with normal stats (whatever that means) and one is obsessed with 13+ in everything.
Saying “you’re in the normal range, but you have Exceptional (Stat X),” should tamp that down a little.
As a side note, I would like to pont out that Gravity Sinister did away with negative modifiers, as well, but it probably doesn’t matter much now.
Years of playing RPGs like RuneQuest III have soured me on ability scores and they’re modifiers. This is why I adore OD&D. Mostly just experience bonuses and penalties, which you can easily ignore. The 3-18 range works perfectly as a descriptive role-playing tool and maybe not so much for the meta aspects of post-Greyhawk supplement D&D.
If you go the bonuses only route, you could try pilfering Ars Magica’s ability score generation method with a little Traveller5 mixed in for fun. Make an opposed roll with the GM or another player and subtract his roll from yours. For B/X use d4’s (mod range -3 to +3, avg. 0), post-3rd edition use d6’s (mod range of -5 to +5, avg. 0). If negative mods bug you, try rolling two dice of the same size and subtract the low roll from the high one (range 0 to +5, avg. +2). Do that three times and split each one amongst two ability scores.
I feel similarly about OD&D ability scores. I love how they get out of the way.
The approach of later D&Ds to ability scores, with the highly impactful modifiers, feels intrusive in that it steps too far into the area of class competency. You find that players often don’t want to play a wizard with less than 17 intelligence (or whatever), and then what’s the point of the variation? Just roll the stat into the class and be done with it. And with many of these systems, the underlying principle of ability scores being the “nature” part of the character means that the player is mostly stuck with those scores, baring powerful magic.
Which is to say, I guess, that if the stat is an important part of class competency, it should be able to be improved like class level. That sounds a little more dogmatic than I really intend, but I think there is some basic principle there that I am just unable to express more eloquently right now.
In one seemingly ancient D&D campaign of mine, I allowed player’s to roll a d20 over one ability score after achieving a new level and if they rolled above it, it increased by one. Obviously, I stole that from Basic Roleplaying. We always stuck with 3d6 in order, so it worked okay. Nowadays, we often start play without even knowing the characters ability scores and just roll for them when they’re are needed in play. Confession, that idea was snipped from HeroQuest, but it’s damn fun and seems to fit in with the wild frontiers of OD&D.