Mike's Standard Rant #4: Stat/Skill systems

[Mike Holmes]

This one's title may be misleading. I am not against Stat/Skill systems per se. I am against their general mishandling, and this rant takes a look at a common problem with game currencies. This is a another of those well known problems that somehow people keep falling into.

An associated problem that I often refer to the GURPS problem (as it is possibly the most talked about problem on the Newsgroups associated with that game) will help to illustrate the . In GURPS CharGen, characters often get made in a specific way because it is cost effectiove to do so. Specifically, if you have a certain number of skills that default to a particular statistic, it becomes cheaper to increase your statistic, thereby increasing your skills, than to raise the skills themselves. Thus players are encoraged to create characters with high stats and low skill proficiency. If you don't believe me ask the designers who see it as a design feature. They claim that this helps create better "Beginnig" characters. You can buy this, but it begs the question, "What about non-beginning characters?" It's just a dubious statement all around, and a justification for what is actually a fault in the system.

Just to be clear, there is a double incentive to raise your stats instead of skills, because, given that in GURPS every skill defaults to a stat, you are actually increasing your effectiveness in all those other skills simultaneously. And then there's standard stat checks which are made more effective through raising stats. Even if you don't have the prerequisite number of skills, there is often enough incentive to raise your stats and sacrifice skills.

Taken from another perspective, this essentially means that players who create characters that do not follow this min/maxing and instead focus on the character's skills, are penalized by having a less effective character. This specific description is for the GURPs problem, and other games' problems will differ slightly. But it's remarkable how many games have this problem in some form or another.

OK, have I established that a problem exists? (not really a subject for debate, here, go to the appropriate newsgroups and read the miles of posts about it).

Anyhow, what causes the problem? If Skills default to stats, then raising a Stat increases the character's effectiveness over a broad area that includes those skills. In theory it includes an infinite number of these subspecializations. Which means that if Skills, being some of these subspecialities cost a fixed amount, then the value of reaising a stat which raises them all must perforce be infinite to "balance" equally. Even if you just looked at only the skills in the GURPS book, raising your DEX by one is equivalent to raising fifty skills by a single point. Given that it actually raises your effectiveness on any other task that is unlisted as a skill, you can see my point. As soon as you nail down the costs of each (even on a sliding scale like GURPS has) you create an exchange rate. If a player has skills for which the total cost for a level in each is less than the cost of the Default stat, they have some small incentive to buy the skills. But as soon as the total is exceeded, and it costs less to buy a level of the Stat, there is no incentive at all to take Skills.

Well, what does this mean? It means that you cannot use a system where skills default to stats and still have fixed values for both, without creating an optimum min/maxing opportunity. Which means that players have an incentive to make all their characters in a similar fashion. In GURPS it means that characters will often have high DEX and/or IQ stats and low amounts of skill.

So, how do you solve the problem? Unfortunately there is no perfect answer that I am aware of.

One "solution" I see attempted is to fiddle around with the exchange rate. This never works. Either stats become overpriced and pointless, or the exchange rate just exists at a different point.

There are a few simple solutions that throw the baby out with the bathwater. One is to just do like a lot of newer systems, and have only one set of defining traits that can either describe what would be stats or skills in other systems. In other words don't use a Stat/Skill systm at all. Not really a solution, however, just avoids the problems and throws out the benefits of the Stat/Skill split (which some people will debate, but is not really important to this discussion).

Another simple method is to not allow purchase of Stats and Skills with the same pool of points. Have separate pools for each. This works, but limits player creativity in defining the character as much or more than if you had just stuck with the original problem. Note that in many systems there will be an exchange rate hidden in the rules so as not to be obvious. But these are usually ferreted out fairly quickly, and then the problem returns. For example, in Ars Magica (some edition) there is a trait that you can take that allows more skills to be taken, or something like that. This creates a rate of exchange that can be exploited.

A third method is to create a system where skills and stats do not add in a linear fashion. I've seen some systems that do this fairly well, but some just hide the problem, again, simply making it difficult to see. And they almost all make determination of your chances of success hard to calculate at a glance. There is room here for real soutions, however.

My favorite solution is a bit math heavy, and so it tends to repel people. It's also one that is actually very old having been invented (though not employed correctly by) Hero System. They encountered the same problem in their design. The method of correction they used was packages. You could buy a subset of what you got for buying a Stat at reduced cost over what it would cost to buy them as individual skills. Essentially, they had discovered that fixed prices for skills does not work. But their implementation, as with all package systems I've seen is clunky, and innaccurate. mathematically.

Which is odd considering that the answer was under their noses all along. For powers, they had figured out an ingenious formula for calculating how much an adjustment to that power, one that either limited it or gave it advantages, should cost. The same principle can be applied to skills. A skill is, for all practical purposes, a limited form of the stat to which it defaults. The more skills you have, the less limited your stat is, and the closer in value that area of the character is. Or stated another way, if a Stat is the representation of a whole category of effectiveness, and a skill is representative of some smaller segment, and further, the Stat's cost is fixed, then purchasing any number of skills should always be somewhat less expensive than purchasing the Stat.

As I said, this is mathematically complex, however (every point purchased has to be calculated separately). In fact, I only use the principle to do character generation for computer powered games. I wouldn't subject unaided table-top players to such.

Perhaps there are other solutions, but from what I've seen, they all fall into one of the categories above.

The subject of Stat/Skill splits is a complex one, and fraught with potential problems. And the solutions (at least those that I have seen) are imperfect. So if you are going to include such a system in your game, beware. Do not just throw something together without giving it some serious analysis! Just because you wouldn't abuse your system, or don't mind the system abusing you does not mean that this will be the case for others.

Mike

[Ron Edwards]

Hi there,

As I'm sure everyone has figured out by now, two threads are basically raising the same issues. The other one is Attributes + skills. I posted some relevant links into that one.

I guess we'll just conduct the threads simultaneously; just be sure to reference both before replying and to cite others' points carefully so readers can find them.

Inelegant, but we'll see how it goes.

Best,
Ron

[Mike Holmes]

Huh, sorry, I didn't even see the other thread.

But I think there is a subtle difference. Ralph is talking about whether or not Stat+Skill is a useful paradigm at all. I'm just saying that if you do go that way, watch out for the potential min/max pitfalls.

Mike

[Valamir]

Yeah, they really aren't similiar threads at all except by coincidence.

I will point out the same thing here that I mentioned in the other thread and that I think is confusing two seperate issues together.

This isn't really a skill vs stat rant, its a  rant about point based character creation systems that just happens to be useing stats and skills as an example of the numerous pitfalls there in.

[Mike Holmes]

True to an extent, Ralph. The thing is that many other forms of CharGen structure wil produce the same problems, or are actually the same solutions as some of the above. But true freeform generation avoids these pitfalls, yes (while having problems all it's own).

Mike

[Valamir]

Well that depends on whether you define the problem as the system having breakpoints, or a system where the breakpoints are in the players control to take advantage of.

For example the easiest (and most traditional) way of dealing with the issue is to make the Traits randomly rolled while the skills are purchased.  As long as you are not using random generation methods that are so generous as to pretty much be defacto "take the scores you like" the problem is largely dealt with.

For instance in d20 there is a break point in translating the 3-18 score to the die modifier.  This isn't really a problem using the straight forward random rolls because the player isn't in control of the breakpoint.  Its a minor problem when combined with the ability to raise attributes at certain levels because there will definitely be a motivation to raise attributes based on where that next break point is.  This is entirely a feature of wanting despirately to keep the legacy 3d6 attribute rolls, however, and not a function of random traits.  If they'd skipped the 3d6 roll and conversion and went directly to determining the modifier this problem would be non existant.


Turning it around a bit, the problem of point based character creation goes way beyond balanceing skills and stats.  Your example is the easiest way to prove your point (and is likely the most distasteful to you being you're mathematically inclined) but the issue is essentially the same as trying to balance the cost of a "Lightning Reflexes" advantage, with a "Poor eyesight" disadvantage.

The only difference is that the former example is objective and totally mathematically proveable, while the latter example is totally subjective and relative to how each is used in game.

[Paganini]

This one's title may be misleading. I am not
An associated problem that I often refer to the GURPS problem (as it is possibly the most talked about problem on the Newsgroups associated with that game) will help to illustrate the . In GURPS CharGen, characters often get made in a specific way because it is cost effectiove to do so. Specifically, if you have a certain number of skills that default to a particular statistic, it becomes cheaper to increase your statistic, thereby increasing your skills, than to raise the skills themselves. Thus players are encoraged to create characters with high stats and low skill proficiency. If you don't believe me ask the designers who see it as a design feature. They claim that this helps create better "Beginnig" characters. You can buy this, but it begs the question, "What about non-beginning characters?" It's just a dubious statement all around, and a justification for what is actually a fault in the system.

Yup. One way to fix this is to go the way WEG did with the D6 Legend games like Xena / Herc: once defined attributes do not change. Every character in the game has each attribute permanently defined on a scale of 2D to 5D at chargen. So, it can be helpful at the begining of the game to jack up a particular attribute in order to get higher levels in skills associated with that attribute, but in the long run this just makes it harder to get high levels with skills associated with lower attributes.

Another way is to still use a stat + skill system, but determine which stat is associated with which skill on the fly. That is, a sword skill might be associated with the coordination skill when fighting, but with the perception skill when examining a sword for flaws. This way there's no guarantee that spending points on a particular attribute is going to rais a particular skill. Alternatively, you could say that any point given to an attribute improves ALL skills by one point, only for certain uses. By distributing points among the different attributes the player is making a statement about his character by defining the attributes he thinks his character will use the most often.

Well, what does this mean? It means that you cannot use a system where skills default to stats and still have fixed values for both, without creating an optimum min/maxing opportunity. Which means that players have an incentive to make all their characters in a similar fashion. In GURPS it means that characters will often have high DEX and/or IQ stats and low amounts of skill.

This isn't quite true, though. It's true for GURPS, obviously, but it is possible to have a system that doesn't encourage minmaxing where skills still default to attributes. D6 Legend is one such as mentioned previously. (Although it has a different problem in that some skills / attributes are more *useful* than others in a game system sense.)

So, there you have two easy fixes that don't throw the baby out as far as I can see. Either one can work with point based chargen without *requireing* separate pools. Although, I think the D6 method works better with separate pools, because the nature of the game would encourage everyone to load all their starting points into attributes and pick up skills later. (That is, get the biggest boost you can when starting up, then learn skills later.)

[xiombarg]

Mike: What about the way Storyteller handles it? To increase your effectiveness, you need to increase your Stat AND your Skill, as both just contribute dice, and both of them max out at 5.

(This is assuming the Vampire 3rd ed rules for botches to having a 10-die pool DOESN'T increase your chance to botch.)

[Seth L. Blumberg]

Another possibility is to relate stats and skills such that the stat connected to a skill becomes a cap on the skill's maximum level, rather than contributing directly to the skill. In some versions of FUDGE, I believe stats are used to determine the point at which buying skills becomes more expensive, so that a high stat makes high skills cheaper but doesn't actually increase skill levels.

Unfortunately, this does not work in practice, as it means that high-stat characters tend to be better at the things for which they have not bought skills (because they can roll against their stats).

[Mike Holmes]

Ralph,

Randomizing one portion falls under the Splitting Pools as I enumerated it above. Yes, this works, but it eliminates some freedom.

X,
Well, that's only to be really good. There is still an exchange rate as long as your total effectiveness is less than 5 dice. The rate wil still come into play as a player will not likely be really good at everything. Anyhow, this is just limiting a character. Why not allow higher ratings (in act, they do, don't they for NPCs)? Because it makes the min/maxing easier. So to fix the problem we break it in another way. Not good. Hero system tried this with a soft limit called Normal Human Characteristic Maxima for non-superheroic games. Which just makes the rate harder to calculate, it doesn't eliminate it.

Mike

[Mike Holmes]

Another possibility is to relate stats and skills such that the stat connected to a skill becomes a cap on the skill's maximum level, rather than contributing directly to the skill. In some versions of FUDGE, I believe stats are used to determine the point at which buying skills becomes more expensive, so that a high stat makes high skills cheaper but doesn't actually increase skill levels.

Unfortunately, this does not work in practice, as it means that high-stat characters tend to be better at the things for which they have not bought skills (because they can roll against their stats).

Same problems hidden behind slightly harder math. If the governing stat does not produce effects in play, then you get the effective rate by simply doing the multiplication again (x increase in Stat gives me y increase in output times z Skills affected). That format really bugs me, as it adds complexity and breakpoints without teh actual benefit of the two-tier system. If such Stats do affect things in-play, and they limit Skills then that's just further incentive to raise your stats, which simply makes the calculation of the exchange rate a bit harder.

Mike

[Valamir]

Ralph,

Randomizing one portion falls under the Splitting Pools as I enumerated it above. Yes, this works, but it eliminates some freedom.

Quite so.  I offered it to illustrate again that IMO the problem is one caused by point based character generation, and is, if not non existant at least more difficult to find, in non point based systems.

This was in response to your comment about finding the same problem in other forms of Character Generations.

[Ron Edwards]

Hello,

I suggest that randomized character creation often results in multiple characters being made, either simultaneously or sequentially, until a "good" one results. If a Currency problem exists based on the interactions of (say) Attributes and Skills, then the same issues of canalizing playable character concepts remain in place, even though this system is not point-allocation-based.

Best,
ron

[Valamir]

Hello,

I suggest that randomized character creation often results in multiple characters being made, either simultaneously or sequentially, until a "good" one results. If a Currency problem exists based on the interactions of (say) Attributes and Skills, then the same issues of canalizing playable character concepts remain in place, even though this system is not point-allocation-based.

Best,
ron

Quite true, and hence why I used the caveat:

As long as you are not using random generation methods that are so generous as to pretty much be defacto "take the scores you like" the problem is largely dealt with.

[xiombarg]

Well, that's only to be really good. There is still an exchange rate as long as your total effectiveness is less than 5 dice. The rate wil still come into play as a player will not likely be really good at everything. Anyhow, this is just limiting a character. Why not allow higher ratings (in act, they do, don't they for NPCs)? Because it makes the min/maxing easier. So to fix the problem we break it in another way. Not good. Hero system tried this with a soft limit called Normal Human Characteristic Maxima for non-superheroic games. Which just makes the rate harder to calculate, it doesn't eliminate it.

Generally the only NPCs that have a stat or skill higher than 5 in Vampire are elder vampires, and even then they have a skill/attribute max set by generation.

Okay, so there's an exchange rate below 5. But overall, there isn't. Yes, it's limiting, but in my experience you have a reasonable chance of success at 6 dice.

However, I guess your general point still stands. I guess I just wanted to nitpick. ;-)

But I'm confused by one thing: What exactly do you want, here? Is what you want total player freedom without min/maxing encouraging a particular way of creating a character?