Attribute Scores: Chicken or Egg Problem

Tadatsune

Registered Magic-User
Validated User
#1
So, I like attribute scores. I know it's trendy to discard them in favor of more direct methods of defining a character (like traits, direct bonuses, etc), but I feel like - if done well - they can be helpful aids for conceptualizing a character. This is especially true if you want a system that places an emphasis on random character generation - something I have been increasingly interested in, of late. Unlike many players, I don't tend to go into RPGs with a strong idea of what my character is going to be like, but rather prefer that that character develops more as an emergent property of play. Random char-gen also has other benefits, such as allowing for quick turn-over in the case of permanent character death or promoting exposure to role types you would not normally choose for yourself. I recognize that not everybody prefers this style of play, and there are certainly aspects of traditionally random char-gen that I find unappealing - such as variable power level between starting PCs. I'd like to make a system that produces (relatively) mechanically balanced characters, either via random generation or - for those with a reestablished character concept in mind - by manually selecting options from tables. Characters would start with identical base stats, which would then be adjusted depending on the selected class/heritage/background/traits & etc., such all characters would have the same sum-total attribute scores.

Anyway, the issue I'm currently having is which order to determine interdependent features - namely, should attribute-related traits determine attribute scores, or the other way around? My initial concept was that you would start with a base number, say a "Might" attribute of 5, and then randomly selected traits would increase or decrease this value, as well as providing ancillary benefits. So, "Mighty Thews" might give you +1 Might as well as a bonus to damage output or something similar. Not only is this friendly to "Random Char-gen" engine I am envisioning, but it allows you to express different, potentially conflicting aspects of a generalized attribute: Take "Intellect" for instance, an attribute that encompasses everything from wits, to wisdom, to knowledge, to acumen, to insight & so on. With this system, I could create a character that has a "Booksmart" trait for +1 Intellect, but also a "Naive" trait for -1 Intellect. These bonuses/penalties cancel out in regards to the general attribute score - something that would be used for general Intellect "attribute tests" and the like, but situational bonuses/penalties would kick in under the appropriate circumstances (say, overcoming a challenge requiring academic knowledge, vs. overcoming a challenge requiring street-smarts). In addition to the mechanical aspects, these traits would also supply role-playing queues for players to follow.

The other potential way of doing this is generating the attribute scores first, and then selecting traits or bonuses based on the total number. So, a character with a Might of 7 would have not only larger numeric bonuses for any skill tests based on might, but would have a wider selection of "might" linked traits/bonuses to choose from as compared to a character with a Might of 5 or 3. This unfortunately decreases the systems ability to mechanically represent "opposing" aspects of a particular attribute (such as booksmarts vs. streetsmarts, as detailed above) but does prevent potentially counter-intuitive skill/attribute score combinations, such as a character with "Mighy Thews" that none the less has a low Might score due to possessing multiple opposing traits.

Anyway, these are just some preliminary thoughts. I'd be interested in hearing what others thought about the advantages or disadvantages of proposed systems.
 

pstjmack

Registered User
Validated User
#2
Sounds reminiscent of Pendragon and its plus/minus approach to personal traits. That said, one of the main reasons I adore GUMSHOE is that it puts all skills and personal characteristics onto the same mechanical basis, as Abilities. Heroquest does the same to an extent. And the simplicity of such systems makes me wonder why anyone bothers with the complexities of interdependent attributes, traits, etc. One set of character stats, and that's all.

Take a look at Call of Cthulhu 7th edition, for instance. Now all personal characteristics are rated 1-100, to bring them into line with the percentage skills that the rest of the system runs on. To me that begs the question: why have those separate characteristics at all? Most of the time in-game, you're making skill rolls, except when you hit a grey area and go for characteristic tests. To me, again, it feels very artificial to be suddenly rolling an abstract STR or INT test, when a skill like Athletics or Research can just as well define what you're doing. The characteristics give me this impression of hanging there in the background of the system, complicating it but not actually doing very much beyond giving players hooks to hang character identity on, which they could get just as well by writing down their hair colour or gait.

If we posit something like Pendragon's D20 system, I'd say balance it out by making certain ability/attribute ratings crimp others. For instance, every point of Occult you have above 10 reduces your Sanity/Stability by 1. Very high ratings in Academic categories reduce your maximum Health or Athletics ratings (too much time at your desk means too little exercise, etc.)

Essentially, in my very lean-and-uncrunchy view, numbers in a game system are there to do stuff, not to hover as background definitions. The more they can be put to work doing stuff, on a simple uniform basis, the faster and more furious the fun.
 
Last edited:

Tadatsune

Registered Magic-User
Validated User
#3
Thanks for the response. Honestly, though I'm not really interested in attribute score alternatives - I understand the - often compelling - arguments for ditching the stats and going for direct bonuses and abilities, but ultimately I'm not convinced. For one thing, I do actually think having a numeric score for classic attributes (STR, DEX etc) has value for helping define a character - especially if you are going into the defining process without a strong preconceived concept. Then there is the fact that ditching stats would require me to create a much more comprehensive ability list than I would care to or feel comfortable with. I also really like the idea of derived ability scores, where you use combinations of base attribute scores to generate ability scores (so your base ranged weapon capability might be dependent on your Perception and Agility scores, for instance). Basically, I can randomly generate as discrete set of numbers, and then use those numbers to generate scores for a wider array of logically connected abilities. Furthermore, I don't need to define every possibly category of skill, because I have the base scores (or new combinations there-of) to fall back on whenever a novel situation calls for it.

Having an "Athletics" rating is neat, for instance, but it doesn't offer you much granularity. Take something like sprinting vs. distance running - I might use a combination of a Might and Agility to determine a base rating for challenges involving the former, and a Might and Fortitude rating for the latter, on top of which I might add some sort of situational trait or background related bonus - like a "Marathon Runner" trait.

Edit: I will agree that a system utilizing attribute scores should make sure those scores have an actually mechanical function, rather than just using them to determine a finite set of direct bonuses - at that point you might as well ditch the base states and go with the bonuses directly.
 
#4
I'm not against indirect attribute scores, usually players find them a good way to figure out the character at creation step - and actually that's the goal of it, if the value doesn't actively play any role just after. It also allow to play more with accessory scores, using attributes as blueprints.

I'll go more toward a set of natural attributes and a set of untied "virtues". Avoiding linking them in such a direct logic - high INT == more INT traits, sound like granting a double advantage.
Just choose/roll a number of virtues, being attributes your physical skillset (including "intellect" as a phisical trait), and virtues your aptitudes.
So a "Stubborn" character can apply that modifier to both running a marathon, focusing on a boring lecture or being assertive on a debate, summing up with different attributes. It could turn into a malus when more flexibility is needed. It also allow for more mechanic indepths, like stressing by using virtues, or having virtues influence differently the dice mechanic.
Or giving default virtues by race/culture/class/zodiac for quick characterization.
 

kenco

Registered User
Validated User
#5
...attribute scores... can be helpful... for conceptualizing a character... especially... random character generation... also has other benefits, such as allowing for quick turn-over in the case of ... death or ... exposure to role types you would not normally choose for yourself.
OK!
...I'd like to make a system that produces (relatively) ... balanced characters, either via random generation or - for those with a reestablished character concept in mind - by manually selecting options from tables. Characters would start with identical base stats, which would then be adjusted depending on the selected class/heritage/background/traits & etc., such all characters would have the same sum-total attribute scores.
So far this all of this sounds like your 'initial concept' to me - i.e. the first of the two options you describe.

I can't tell for sure whether you are:
a) committed to the idea of a mixture of random generation and manually selecting options; or
b) want to provide both options - random generation and manual selection and/or a mixture - in a single system; or
c) open to using either or both features, provided the system produces 'relatively balanced characters'

I think* the more manual selection you offer, the less of the interesting benefits of random generation you are likely to get. There is an intrinsic tension between random generation and (mechanical) balance. But balance is a complex, multi-dimensional concept in RPGs, and hard to pin down.

*But my conclusion below seems to contradict this thought... so o_O

You might come up with a system that supports either option (or some arbitrary combination) equally well. As one example, a fixed series of choices you make by choosing options from one or more lists. This is easily adapted for random character generation by forcing the player to roll for each choice, rather than choosing one. However, if you are concerned about balance, a major drawback is the intractability of 'balancing' all possible combinations that a player might choose. Generations of Hero System designers have spent decades try to balance combinations with an elaborate point system, but the rule book comes near to saying that 'balance' is always going to involve case-by-case judgement.
...should attribute-related traits determine attribute scores, or the other way around?
Does it have to be one or the other? Surely your system could be a hybrid.
My initial concept... you ...start with... a "Might" attribute of 5, and then randomly selected traits would increase or decrease this value, as well as providing ancillary benefits... friendly to "Random Char-gen" engine ... allows you to express ... conflicting aspects of a generalised attribute... These bonuses/penalties cancel out in regards to the general attribute score ... used for general Intellect "attribute tests" and the like, but situational bonuses/penalties would kick in under the appropriate circumstances ... these traits would also supply role-playing queues for players to follow.
Sounds OK. But what does the resulting 'generalised attribute' represent? When is it used? There is a danger that the generalised attribute will end up being a meaningless number that simply fills a mechanical gap for a default where none of the traits applies. Which defeats your desire to use the attribute score as a meaningful indicator of something.

One of the interesting things about the old '3d6 in order' procedure is that you have to interpret what DEX 8 means (or doesn't mean). But your idea reverses that sequence defining what your character is like in more detail (Mighty Thews with auxiliary bonuses etc) and then telling you how many net pros/ cons you have in the general field. You go from defined specifics to a general score rather than general score to defined specifics. The former is a convergent/ rules driven approach. The latter is a divergent/ interpretive approach.

You've highlighted below the possibility of contradictory traits (Mighty Thews plus Lightweight) appearing in the same randomly generated character. If you're worried about the contradictory sets you could design the random selector so that it never generates pairs you regard as 'impossible'.

The obvious advantage is that it plays very strongly to your 'random character' ethic. Who cares how contradictory the trait combination sounds? It's your job as a player to make sense of that somehow and play your character.
... [an]other potential way ... is generating the attribute scores first, and then selecting traits or bonuses based on the total number. ... a character with a Might of 7 would have...larger ... bonuses for...tests based on might, [and] would have a wider selection of "might" linked traits/bonuses to choose from as compared to a character with a Might of 5.... This .... decreases ... ability to... represent "opposing" aspects ... but does prevent ... skill/attribute score combinations, such as a character with "Mighy Thews" that none the less has a low Might score due to... multiple opposing traits.
The obvious issue here is that you're introducing a lot more player choice in the character design. This seems to depart from your random char-gen ethos.

But I think it is actually closer to the original 3d6 in order process in the sense of allowing for interpretation. You start with a general score and then 'interpret' it by choosing pro/con options that somehow 'add up to' the randomly generated total. What you are doing is adding a finite list of rules for mechanical elaborations around the random general attribute.

The balance issue is likely to be in the foreground of your design considerations here, because some combinations of traits might easily be much more 'valuable' than others 'purchased' with the same underlying general attribute score.
Anyway, these are just some preliminary thoughts. I'd be interested in hearing what others thought about the advantages or disadvantages of proposed systems.
So both of these look to me like ways to elaborate and specify the mechanics. That will necessarily come at the expense of added complexity and slowness: so if rapid replacement characters is a high priority, you will need to keep them simple. The first approach might be easier to balance, but the second is arguably closer to the creative/ interpretive process involved with random generation.

After having though about this, I am surprised to find that I lean toward the second approach. However I personally don't enjoy choosing from long lists of traits/ powers.

Interesting ideas!

K
 

Tadatsune

Registered Magic-User
Validated User
#6
I can't tell for sure whether you are:
a) committed to the idea of a mixture of random generation and manually selecting options; or
b) want to provide both options - random generation and manual selection and/or a mixture - in a single system; or
c) open to using either or both features, provided the system produces 'relatively balanced characters'
So, the idea is that you would have a char-gen system that would be largely automated - key aspects of the character would be generated by die rolls on random tables, and the player would be left to fill in the details. However, at any point in the process the player could opt to skip the random rolls and simply make a selection from the table. Thus, if you want to create a specific character concept, you would just pick the options you wanted, whereas if you wanted to let the game surprise you, you'd let the dice do the work.

Obviously, a constructed character would have greater potential for min/maxing or similar behavior - something I'd like to avoid as much as possible - but the hope would be that the options presented would be sufficiently balanced such that there wouldn't be that much difference in mechanical "power" or utility between any two starting characters, whether randomly generated or otherwise.

I don't think its entirely possible, or even desirable, to randomize the entire process, mind you. The char-gen system is supposed to provide a basic framework for a new character - one that you might not have come up with on your own - not spit out full-formed characters without any player input. That said, I would attempt to automate as many aspects of character generation as is feasible.
 

Tadatsune

Registered Magic-User
Validated User
#7
Sounds OK. But what does the resulting 'generalized attribute' represent? When is it used? There is a danger that the generalised attribute will end up being a meaningless number that simply fills a mechanical gap for a default where none of the traits applies. Which defeats your desire to use the attribute score as a meaningful indicator of something.
I'm not sure why having a "default" score for generalized tests is considered a bad thing. I'd much rather have a small list of generalized attributes that can cover a wide array of situations than have a lengthy list of ultra-specific scores. Same goes for skills and other abilities. I'm perfectly OK with a degree of abstraction here, and don't have any problem with delegating authority to the Referee to make fexible, on-the-spot calls on how to resolve skill tests.

To better illustrate, let me give you the set of attributes I've settled on: (nothing revolutionary here, its more or less the standard set for a fantasy RPG)
Might (M) - not just physical strength, but the projection of physical power
Agility (A) - grace, coordination, manual dexterity, reflex, balance, speed, initiative & etc.
Fortitude (F) - physical resilience, endurance, vitality, constitution, stamina & etc.
Intellect (I) - wits, wisdom, knowledge, insight, acumen, metal acuity & etc.
Perception (P) - sensory perception, awareness/alertness, & etc.
Will (W) - mental resilience, self-discipline, force of personality, influence, courage & etc.
I am also thinking of including a Luck/Fortune stat as well.

The way you would approach a "skill test" would be first to select the two most appropriate attributes for the test, as determined by the referee, and then adding them together to get a base rating. I haven't determined the actual mechanic I want to use, here, but basically the rating would interact with a die roll (3d6) and any applicable bonuses, such as those provided by a background or talent, and the result compared to a difficulty rating to determine success/degrees thereof. To give an example, lets imagine a character wants to climb a precarious cliff face: the referee determines that Might (M) and Agility (A) are the appropriate attributes for the test, so M+A provides us with the base rating. A character with a talent for climbing would then receive a bonus. Possession of climbing tools might provide an additional bonus, or decrease the level of difficulty (again, haven't fully fleshed this out yet - I'd like to keep the number of bonuses/penalties to a minimum to reduce number bloat and time spent on calculations). In contrast, for a pure "feat of strength" like lifting something heavy or bending iron bars and the like, you'd simply double the relevant attribute to arrive at the base rating (ie, Might + Might). The rulebook would naturally provide a list of standard "ability" combinations for common situations, but ultimately the call of which attributes to test would be in the hands of the referee.
 
Last edited:

kenco

Registered User
Validated User
#8
Sorry, I wasn't very clear.

There is nothing wrong with generalised attributes or generalised tests IMHO. And there is nothing wrong with having a "default" score. My concern was that the default score might not seem meaningful in the same way if it is the result of 'adding up' randomly generated specifics that might vary from character to character.

What I meant was that there seems to be a psychological difference between 1) starting with the idea of Might (big, strong, tough... however you conceptualise it) and adding traits that adjust the score in particular specialised circumstances; and 2) starting with a list traits that apply in diverse specific circumstances and using them to generate a generalised score that represents... whatever is left over/ averaged out/ somehow common to those traits... and using that score in a bunch of situations that might not seem have much to do with the specific traits that generated it. In the first case, Might means something pretty clear; in the second case it seems less clear.

I grant that the two approaches might be mechanically/ arithmetically identical; but going from generic attribute to specific traits that vary from character to character is a different mental process than going from a list of specific traits (plus the unlisted traits a given character doesn't happen to have) to a general attribute that is shared by all characters.

I don't think there's anything intrinsically wrong with the second approach, only that it seems further removed from the idea of random stats plus interpretation (which was your starting point for the post) than the first.

Hope that helps.

Cheers

K
 
Last edited:

Alban

Registered User
Validated User
#9
As for myself, I like having broad abilities (whether "characteristics" or broad "skills") and more specialized ones (skill" or "specializations"), but I prefer the former to be defined in such a way that they don't overlap.

Note that if you systematically add 2 characteristics together, you create an implicit list of 30 broad aptitudes, one for each couple of characteristics (even though some will rarely be used in play).
 

cloa513

Registered User
Validated User
#10
I'm not sure why having a "default" score for generalized tests is considered a bad thing. I'd much rather have a small list of generalized attributes that can cover a wide array of situations than have a lengthy list of ultra-specific scores. Same goes for skills and other abilities. I'm perfectly OK with a degree of abstraction here, and don't have any problem with delegating authority to the Referee to make fexible, on-the-spot calls on how to resolve skill tests.

To better illustrate, let me give you the set of attributes I've settled on: (nothing revolutionary here, its more or less the standard set for a fantasy RPG)
Might (M) - not just physical strength, but the projection of physical power
Agility (A) - grace, coordination, manual dexterity, reflex, balance, speed, initiative & etc.
Fortitude (F) - physical resilience, endurance, vitality, constitution, stamina & etc.
Intellect (I) - wits, wisdom, knowledge, insight, acumen, metal acuity & etc.
Perception (P) - sensory perception, awareness/alertness, & etc.
Will (W) - mental resilience, self-discipline, force of personality, influence, courage & etc.
I am also thinking of including a Luck/Fortune stat as well.

The way you would approach a "skill test" would be first to select the two most appropriate attributes for the test, as determined by the referee, and then adding them together to get a base rating. I haven't determined the actual mechanic I want to use, here, but basically the rating would interact with a die roll (3d6) and any applicable bonuses, such as those provided by a background or talent, and the result compared to a difficulty rating to determine success/degrees thereof. To give an example, lets imagine a character wants to climb a precarious cliff face: the referee determines that Might (M) and Agility (A) are the appropriate attributes for the test, so M+A provides us with the base rating. A character with a talent for climbing would then receive a bonus. Possession of climbing tools might provide an additional bonus, or decrease the level of difficulty (again, haven't fully fleshed this out yet - I'd like to keep the number of bonuses/penalties to a minimum to reduce number bloat and time spent on calculations). In contrast, for a pure "feat of strength" like lifting something heavy or bending iron bars and the like, you'd simply double the relevant attribute to arrive at the base rating (ie, Might + Might). The rulebook would naturally provide a list of standard "ability" combinations for common situations, but ultimately the call of which attributes to test would be in the hands of the referee.
Might needs a little more work. I'd say might is physical and psychological effect of said physicality.
 
Top Bottom