Derived Stats: how far is too far?

[ACOS]

This is a question of First Principles (I'm having trouble figuring out how to effectively explain the concept, so bear with me):

How many derivatives are acceptable when generating stats for an RPG PC?

For example:
A primary stat would be like attributes.
Secondary stat (or, 1st derivative) would be like your awareness check in Vampire (i.e., Wits + Perception)
Tertiary stat "2nd derivative" .... (etc., etc., ...)
And then have everything that you actually use based off of 2nd+ derivative stats.

Does that even make sense?
(I'm at a loss for examples of 2nd or 3rd+ derivative stats).

The "purpose" of this would be changing the levers and how they work. The further derived the used stats are, when you change a primary stat, the effect is smaller in scale, but wider in scope.

I'm just spit-balling/brainstorming, and seeing what can be done in this direction (if anything).

Thank you.

[tussock]

They don't have to be smaller in scale or wider in scope. Traditionally derived stats are narrowly focused.

Damage can add triple Strength mod with a certain action, critical hits can triple damage. A weapon enchantment can double the number of critical hits. A fighting style can double the number of weapons. A spell can make a duplicate of you.

That's a 5th step (or 7th, depending where you start) damage function using 13 times your strength mod.

Strength -> modifier -> modifier multiple -> multiples of that -> further multiples, etc. You can limit or base those other multipliers on other stats if you like. Limit fighting style by Dex, spell access by Int, durations by Con, all sorts.



If all you're after is (Str+Con)/2 type functions, most games stick to 1 step, or rarely 2, in part because to get further you end up with ludicrous numbers of stats that are very hard to conceptually differentiate.

[GâtFromKI]

A function of a function of (x, y, z) is simply a function of (x, y, z). So, you can't even have 2nd derivative under your definition.

Anyway, is the goal of your game to play or to compute some shitty functions?

[CCarter]

Say for 4th Ed. there's
Primary: Constitution
Secondary: Hit Points
Tertiary: Bloodied Value

Here the tertiary layer there is just a rescaling though, it doesn't bring any extra attributes into the picture so may not be exactly what you mean.

Also as Gat says if you're only working with the twice-derived (third layer) numbers then the middle layer is irrelevant; you're just asking for a system where every mechanic is built off adding three things instead of two.

If the intent is to use both layers for calculations that's difficult since they're probably scaled differently- you'd need either different dice (e.g. roll under d10 for a Wits check, roll under d20 for Wits + Alertness check, roll under d30 for Wits + Alertness + Combat Reflexes subskill [initiative check]), to make the final 'derivation' an averaging operation, or to alter a suggested difficulty based on # stats being added (although still with RNG issues likely).

[Drolyt]

It depends on your goals, but in general it isn't a great idea to go beyond secondary stats. You want character generation to be simple.

A function of a function of (x, y, z) is simply a function of (x, y, z). So, you can't even have 2nd derivative under your definition.

That wouldn't necessarily be true if secondary stats could be bought up or down separately, but it sounds like ACOS just wants to split calculations into multiple steps for some reason, which is pointless and probably a bad thing.

[ACOS]

If all you're after is (Str+Con)/2 type functions,

yeah, that's the kind of thing I'm talking about.

I thought I had was something along the lines of:
* start with 12-15 primary attributes,
* mix-mash those in some fashion to get some # of secondary stats (1st derivative) - these would be the stats that you actually use as the basis of everything else,
* other aspects of your character would be derived from there.

Here's some (ass-pulled) examples:
Never mind. The more I think about this, the more tedious and complicated it gets.
I only had a minute kernel of an idea; but as I flesh it out in my head, it becomes infinitely cumbersome.
I just figured I'd bounce this off of you guys before I got too stuck in the weeds.
Thanks.

[PhoneLobster]

I was about to say its a complex issue, then I was about to say "what the ell 12-15 primary stats then you derive secondaries that you 'actually use'? ".

So I'll go back to the complex issue thing. There isn't a right answer without more information.

It matters what you do with the stats. And if what you want to do, apparently, is have large numbers of "secondary" derived stats of "X+Y" then you are probably best off with a "base attribute+specific skill" setup with lots of specific skills but only a small number of base attributes.

The important thing there is not to have any base attribute under-represented in number or importance of the specific skills it covers.

In a more broad sense never design a primary/secondary/whatever attribute derivation that leaves any layer of attributes unused or insufficiently used/important. If your primary attributes are never referred to you have a problem.

So for instance I would advise avoiding the basic D&D thing where your primary attributes are almost exclusively used to derive the attribute bonuses that you actually use. Make the bonuses the attributes and dump the unused base attribute itself.

[silva]

So for instance I would advise avoiding the basic D&D thing where your primary attributes are almost exclusively used to derive the attribute bonuses that you actually use. Make the bonuses the attributes and dump the unused base attribute itself.

This.

[hyzmarca]

If your primary stats don't do anything then you might as well skip them.

If your primary stats don't do anything particularly useful then you might as well skip them.

If your primary stats don't do anything particularly useful that anyone would actually want t do, then you might as well skip them.

The corollary is that there's such a thing as too much detail. If you have fifteen derived stats that do essentially the same ting slightly differently (lifting weight, pushing weight, benchpressing weight, clean and jerk weight), then you might as well just have the primary.

[ACOS]

It matters what you do with the stats. And if what you want to do, apparently, is have large numbers of "secondary" derived stats of "X+Y" then you are probably best off with a "base attribute+specific skill" setup with lots of specific skills but only a small number of base attributes.

This is what I've come to.

The important thing there is not to have any base attribute under-represented in number or importance of the specific skills it covers.

Like CON in D&D?

In a more broad sense never design a primary/secondary/whatever attribute derivation that leaves any layer of attributes unused or insufficiently used/important. If your primary attributes are never referred to you have a problem.

yeah.
It was more of a "simulationist" exercise for me. I wanted to break down a person's individual traits and attributes in to as many discreet base stats as possible; and then each type of action to be some function of how those individual traits/attributes interact with each other. For example, a "test of strength" - such as lifting a car (even for just a second) - could incorporate Focus in addition to Power (you need to focus in order to maximize your power output).
Even an extended endurance activity would require not only Power and Endurance, but also Focus and Mental Toughness. What if that endurance activity involved moving a heavy, oddly-shaped object over rough terrain?

Stuff like that.

But now that I've actually starting thinking this through, I've realized that the 5-page character sheet would be a giant pile of #s with flow charts and shit.
Yeah, bad idea.

The corollary is that there's such a thing as too much detail.

Yeah, I've just have to learn that abstraction is a good thing.

[Josh_Kablack]

3e gives an easy example

Primary: dexterity score
Derived: dexterity modifier
Secondary derived: Armor Class
Tertiary derived: Touch AC

[...You Lost Me]

I don't think it really matters how many steps you go, because if you get to too many derived stats you've probably got complexity issues otherwise.