One thing that's always a difficulty in game design is the relative weight of attributes and skills.

Talent or experience, which is more important?

For example, Classic Traveller had attributes ranging 2-12, and skills of 0 to usually not more than 3 or 4. So it was about 75/25 weighting in favour of attributes. Your raw talent mattered more than your learned skills.

Whereas old Rolemaster had attributes give bonuses of -25 to +25. You'd naturally choose skills which matched your attributes or vice versa, so you'd find in practice the attribute bonuses were 0 to 25. Each "rank" in a skill gave you +5, so that at level 1 you'd have +10 or so in a skill, but by level 5 you might have +50. So in the beginning, attributes were about 50/50 with skills, but as the character became more experienced, skills had a stronger weight than skills.

The old World of Darkness rules had rolls almost all being attribute + skill, and both ranged from 0 to 5, so that it was 50/50.

The relative weight of attributes and skills is important because it flows on to other things. If attributes are as or more important than skills, then the system will usually say that attributes are difficult or impossible to improve, simply for game balance; if we have 6 attributes and 120 skills, and attribute levels affect the total as much as does the skill level, obiously players will try to improve attributes. So the attribute/level balance affects how xp are awarded or spent, and so on.

What then should be the relative weight of attributes and skills? Should one be harder to improve than the other?

I've always held that skills are more imnportant. You know my games and how I handle it, so no surprise there!

-clash

My instinct is that talent is more important for amateurs, and skill more important for the experienced.

You see this in any kind of learning environment. There are some people who get the idea right away and really shine, but if everyone sticks to the course, a couple of years later you can't tell the talented ones from the untalented who worked hard.

In game terms you'd simulate this by having the range of attributes and skills be the same, so that attributes are more important to low-skilled characters, and skills rise to match them.

Or else you could have it that when people get a new skill, rather than starting it at level 1, it starts at a random 1d6 or whatever.

By the way, Clash, isn't there a draft rpg of mine you were going to look over about four months ago? I'm thinking about it again, thus this thread.

Quote from: Kyle AaronMy instinct is that talent is more important for amateurs, and skill more important for the experienced.

Yes, but the amount that talent matters for beginners pales in comparison to how much skill matters for the trained.  In other words, talent might differentiate amateurs because they have no skill but all of the natural talent in the world isn't going to help them compete against someone who is trained, even if the trained person has only moderate talent.

Would you rather get in a car with a dextrous child who has never driven before or a person of moderate dexterity that's been driving a while?  If you had a question about a historical event, would you ask a smart kid who never took a history class or a person of only modest intelligence who has studied history?  And untrained combatants usually get their heads handed to them by soldiers and police officers who have actually had combat training.

Personally, I prefer the Fudge model, where attributes and skills are separate.  The way I deal with your untrained but talented vs. trained division is to look at the task being attempted.

If the task is something that an unskilled person can do, then they get to roll against their attribute but a skilled person doesn't need to roll.  They just succeed.  

If the task is something that requires skill, then the person without the skill rolls at an unskilled level on their skill roll and their attribute doesn't factor in.

For example, an unskilled person might be able to drive a car without crashing in a simple situation so they'd get an attribute roll but a person skilled in driving doesn't need to roll to not crash in a simple situation.  But an unskilled person can't normally do heart surgery so they'd roll will an unskilled penalty on their surgery skill.  In combat, two unskilled people can smack away at each other making attribute rolls but once a skilled person gets involved, then they need to make a skill roll at an unskilled level.

Quote from: Kyle AaronYou see this in any kind of learning environment. There are some people who get the idea right away and really shine, but if everyone sticks to the course, a couple of years later you can't tell the talented ones from the untalented who worked hard.

I've seen it argued (and have experienced) that sometimes the talent actually creates problems because people will get by with their talent for a long time without really learning the skill.  But I think the more telling example is the one that I made above.  The people in the learning environment will be blown away by those people, years later, who have the skill and experience.

So you're saying that if it were Attribute + Skill in a system, you'd make it something like 25% attributes and 75% skill?

As I see it, you could accomplish that in two ways. You could have attributes on a smaller range than skills, for example 0 to 3 vs 0 to 9. Or you could have them on the same range, but make improving attributes several times harder than improving skills, for example both range from 0 to 9, but to improve an attribute costs the new level in xp, while improving the skill always costs 1xp.

Oh, and no-one is going to accept a system where if your attributes are higher you learn more slowly than someone whose attributes are low. It might be realistic, but no-one would accept that.

I think you can't make a generic statement that one way is inherently better than the other. It largely depends on what feel you want for the game in question.

For instance, imagine a pulpy, heroic game of high adventure. Dr. Wisconsin Jones in engaged in a firefight with Nazis, and he runs out of ammunition for his revolver. He grabs a Glock or a submachinegun off a dead Nazi and carries on fighting. Now, a system which emphasises skill over attribute would result in him becoming totally incompentent, since he doesn't have the appropriate skill for these firearms. However, that's not pulpy, not heroic and not adventurous, so in this case the system is built wrong for the genre. Similarly, if Dr. Jones is once established to be good at flying fighter planes, it can be expected in this genre that he'd also be good at flying large passenger liners or driving cars, even if he never really had any formal training in it.

On the other hand, you might have a highly realistic, gritty spycraft game, where you might want to place much greater emphasis on skills - realistically, bugging an embassy with microphones, installing a Trojan Horse on a computer and restoring data integrity on a heavily damaged data carrier would require completely different kinds of training, and no one could have enough affinity with electronics (as represented by some ability) to be competent in all this without actual experience.

Quote from: PremierIt largely depends on what feel you want for the game in question.

That's obvious. My questions here really are two: what's realistic? what makes sense?

Realism doesn't always make sense. I knew a guy who broke his ankle stepping off a curb, and there have been people who won the lottery two weeks in a row. Anything which has really happened must by definition be "realistic", nonetheless those sorts of things just don't make sense to people.

Quote from: PremierDr. Wisconsin Jones in engaged in a firefight with Nazis, and he runs out of ammunition for his revolver. He grabs a Glock or a submachinegun off a dead Nazi and carries on fighting. Now, a system which emphasises skill over attribute would result in him becoming totally incompentent, since he doesn't have the appropriate skill for these firearms.

Not necessarily. It's possible to have a system based on attribute + skill, but allow that skills default to one another; if you have X in one skill, you might have X/2, or attribute +1, in another related skill.

The question then becomes "what's a related skill?" How close do they have to be for you to say they're related? In most games their relation is inversely proportional to the number of unique skills there are. If there are just 20 skills, often there won't be relative defaults, since they already describe quite broad ranges of things; if there are 200 skills, there'll often be many defaults, since each skill covers a narrow area and it's not likely the party will have all of them.

The point is that skills can default from other skills as well as from attributes, so the issue of characters being able to try things they don't have direct skill in isn't necessarily that serious.

Quote from: Kyle AaronMy instinct is that talent is more important for amateurs, and skill more important for the experienced.

You see this in any kind of learning environment. There are some people who get the idea right away and really shine, but if everyone sticks to the course, a couple of years later you can't tell the talented ones from the untalented who worked hard.

In game terms you'd simulate this by having the range of attributes and skills be the same, so that attributes are more important to low-skilled characters, and skills rise to match them.

Or else you could have it that when people get a new skill, rather than starting it at level 1, it starts at a random 1d6 or whatever.

By the way, Clash, isn't there a draft rpg of mine you were going to look over about four months ago? I'm thinking about it again, thus this thread.

Yes... four months ago... where I was preparing to release IHW:Aces And Angels (203p) and Sweet Chariot 2 (303p) within three weeks of each other? That four months ago? Possibly... :O

Sorry, Kyle! I was in Feverish End-Mode writing, editing, and painting right about then! I will surely look it over now that I'm in Feverish End-Mode on IHW:Wild Blue.

-clash

Quote from: John MorrowIf the task is something that an unskilled person can do, then they get to roll against their attribute but a skilled person doesn't need to roll.  They just succeed.  

If the task is something that requires skill, then the person without the skill rolls at an unskilled level on their skill roll and their attribute doesn't factor in.

For example, an unskilled person might be able to drive a car without crashing in a simple situation so they'd get an attribute roll but a person skilled in driving doesn't need to roll to not crash in a simple situation.  But an unskilled person can't normally do heart surgery so they'd roll will an unskilled penalty on their surgery skill.  In combat, two unskilled people can smack away at each other making attribute rolls but once a skilled person gets involved, then they need to make a skill roll at an unskilled level.

That's the method I use, to a T, John.

-clash

Quote from: Kyle AaronThat's obvious. My questions here really are two: what's realistic? what makes sense?

Realism doesn't always make sense. I knew a guy who broke his ankle stepping off a curb, and there have been people who won the lottery two weeks in a row. Anything which has really happened must by definition be "realistic", nonetheless those sorts of things just don't make sense to people.


Not necessarily. It's possible to have a system based on attribute + skill, but allow that skills default to one another; if you have X in one skill, you might have X/2, or attribute +1, in another related skill.

The question then becomes "what's a related skill?" How close do they have to be for you to say they're related? In most games their relation is inversely proportional to the number of unique skills there are. If there are just 20 skills, often there won't be relative defaults, since they already describe quite broad ranges of things; if there are 200 skills, there'll often be many defaults, since each skill covers a narrow area and it's not likely the party will have all of them.

The point is that skills can default from other skills as well as from attributes, so the issue of characters being able to try things they don't have direct skill in isn't necessarily that serious.

You also have the strange "Lots of broad skills that overlap" concept...

-clash

I think the best compromises I have seen are where the skill is made up of a base level formula from your attributes added to a skill points.
My favourite example of this and one that always most heavily influences my own games is the old FGU system ( Davedevils, Aftermath, Flashing Blades etc)  So handgun base skill is say (Dex x 2 ) % (dex ranging say 2 -20). More complex skills say First Aid might be (Int+ Dex) and hard skills might come out like Computer Use (Education + Int)/2 etc etc. FGU games also had talents which were a range of secondard attributes ranging I think from +5 to -5 which were added into formulae.

Now I think it is useful to look at use of skills in reality. Take a skill like Computer Programming. Someone with zero training in Computer programming simply can not program zero chance. Someone can learn to be a competant programmer in a year. People who have a more logical mind (not necessarily higher inteligence but for game purposes an attribute that could be considered intelligence) quickly outpace those that are less able.
We could create this in a game sense by saying a skill increases in blocks a point 'spent' in a skill give you so much benefit in it. Ideally the more logical mind should get more skill for the expenditure of lesspoints. Its like in college 2 guys can get As on a JAVA course but one guy has been studying 5 hours a day for 6 months and the ohter guy spends 3 hours a week. (this is also true for sports although the base attribute 'sportiness' is often put into Dexterity and Stength, largely do to the attributes Gygax chose at the outset) .
The next point is at eh top end of the scale to be come a competent programmer is reasonable easy to become a great programmer (or althete) is much harder. The difference in real skill between competant and great in terms of the spectrum from none to great is actually small, You could probably use the 80/20 rule for it. A competant 100 meter runner might do 12 seconds a world class 100 meter runner is 20% better. In the same way i can give a competent coder a task, write an API to do x and they will deliver it in a week. A great programmer will stil take a week but the code might be faster, better error checking, self healing whatever ...
Then there is the topic of failure. Its usually quite hard to fail stuff. In a technical situation it usually just takes longer (if you are competant) or its new to you.in a physical situation you need to define failure. Its hard to miss a tree trunk with a baseball bat. Its hard in a fight to miss a guys head with a punch. Its easy to miss if he is moving or if you are trying to hit a baseball fast ball.
Finally you have the idea of links skills. Skills exist in trees. A java coder can probably debug some C++ code. A guy who can pilot a MIg can probably keep a 747 in the air (may be not land it). If you can use a rapier you can certainly use a Small sword.

So the 'perfect' balance for me would be one where Skills were arranged in trees; skill progression was based on attributes; Skills levelled off at about 80% such that further increase was slow; standard skill checks were automatic; conflict/resistance would be the main result of skill failure.

I would say from a Game balance perspective that if you randomise Attributes you have to weaken the link into skills or its a double whammy for the unfortunate player that rolls 3 dex...

Quote from: Kyle AaronSo you're saying that if it were Attribute + Skill in a system, you'd make it something like 25% attributes and 75% skill?

What I'm saying is that I prefer a system where attributes and skills are distinct things and if attributes need to contribute to skills, I prefer that they do it like in at least some versions of GDW's House System where it's cheaper to buy levels in a skill that's at or below your attribute level than above.

If a system does need to combine attributes and skills, then, yes, I think that 25% or even less is the way to go.  The Hero System divides attributes by 5 before adding them to skills such that attributes normally add only +1 to +4 to a skill base for a normal person.  In Traveller, certain skills had a +1 bonus only for exceptional attributes.  That's the sort of thing I prefer.  

Simply adding attribute to skill creates all sorts of problems, especially dealing with min-maxing.  Years ago, someone called it the "double point effect" meaning that if there are two ways to buy a value in something and the two components don't have the same cost, there will always be an optimal and suboptimal way to create characters.  And that leads to the complaint in GURPS that gifted amateurs are cheaper than skilled normals, or at least they were before 4E.

I think it's not only "realistic" in many cases (though different skills can work differently enough that it's often possible to find an example that supports just about any model) but it helps with niche protection by giving an advantage to characters who actually take the skills rather than letting gifted amateurs do as well.

Quote from: Kyle AaronAs I see it, you could accomplish that in two ways. You could have attributes on a smaller range than skills, for example 0 to 3 vs 0 to 9. Or you could have them on the same range, but make improving attributes several times harder than improving skills, for example both range from 0 to 9, but to improve an attribute costs the new level in xp, while improving the skill always costs 1xp.

Well, Hero System divides and d20 essentially does, too, by giving a +1 for every 2 attribute points.  The problem with using different scales is that it becomes difficult to use the same mechanics for an attribute or a skill roll if that becomes desirable for some reason, because the odds will be different.  The problem with the same scale without division is that if the attributes get high enough, they make it possible to create incredibly powerful characters spending only a modest amount of attention on skills.

Quote from: Kyle AaronOh, and no-one is going to accept a system where if your attributes are higher you learn more slowly than someone whose attributes are low. It might be realistic, but no-one would accept that.

Well, like I said, the House System does the opposite (it's harder to learn a skill once you exceed the governing attribute) as a substitute for combining the to together and I think that's a pretty nifty solution if the goal is to make them relate to each other.

Quote from: jibbajibbaMy favourite example of this and one that always most heavily influences my own games is the old FGU system ( Davedevils, Aftermath, Flashing Blades etc)  So handgun base skill is say (Dex x 2 ) % (dex ranging say 2 -20). More complex skills say First Aid might be (Int+ Dex) and hard skills might come out like Computer Use (Education + Int)/2 etc etc. FGU games also had talents which were a range of secondard attributes ranging I think from +5 to -5 which were added into formulae.

You need to be careful about averaging multiple attributes into a skill because the more numbers you average (starting at 2, but some games do more), the more likely the result of that averaging will be some medium value every time unless players can afford to peak both attributes (which often isn't the case and raises other issues).  So if I average Dex and Int, then the Wizard with the 2 Dex and 10 Int averages to 6, the Warrior with a 10 Dex and 2 Int averages to 6, the Rogue with the 8 Dex and 6 Int averages to 7, etc.  Not enough variance, in my opinion, to justify the overhead.

Quote from: jibbajibbaSo the 'perfect' balance for me would be one where Skills were arranged in trees; skill progression was based on attributes; Skills levelled off at about 80% such that further increase was slow; standard skill checks were automatic; conflict/resistance would be the main result of skill failure.

I think that's a pretty good list though I think that in practice trees create a lot of headaches because the relationship between two skills often isn't clear or easy to represent in a tree and trying to force things to fit into a structure that don't really fit can create worse problems.  There is a middle ground between trees and no relationship in things like the synergistic bonuses that D&D 3.5 grants for having related skills at a certain level.

As background, my group used to play games almost exclusively with homebrew rule systems and in the course of designing and tuning those systems, we tried all sorts of approaches to these things, including a few that we laugh about now (e.g., when we went through our attributes and sub-attributes phase, we broke out Perception into the 5 senses).  And in the course of doing that, I learned that creating a structure for things like attributes and skills can create a tail wagging the dog situation where instead of picking and naming attributes and skills because the game needs them and they'll work well in play, it becomes an exercise in making it fit and filling in the blank.  For example, if you break your attributes into Mind, Body, and Spirit, you might naturally need 2 Mind attributes, 3 Body attributes, and 1 Spirit attribute but when you go for symmetry, you'll wind up making up 1 Mind and 2 Spirit attributes that you really don't need and will have to shove things like Perception, which has a mental and intuitive component, arbitrarily under either Mind or Spirit.  So I'm very wary of trees, hierarchies, and symmetries because I've seen what they did to our homebrew design efforts.

To address the OP, mechanically, you can obtain any balance you want via abstracting your modifiers from your stats. You probably know this but it just seemed the easiest solution.

For Iridium, this is the way I went. Essentially, levels add a +3 to those skills in your profession reflecting they are used more often. You receive a +2 for those outside your profession. At lower levels, the bonuses from stats are much more important. At higher levels, the level mods become the only wayto become truly exceptional. It breaks down like this.
1. Skill ranks: each rank someone takes gives diminishing returns. Progression is: 40, 60, 70, 75, 90,82, 84, 86, 88, 90, 91, 92...

2. Each level gives you +3 for in class skills, +2 for outside class. For a 4th level fighter, Targeting (Short Sword) with 3 ranks would be 82%. For a thief, it would be 78%.

3. Stats usually give -20 to +20 to a skill. So, the same above example would be 102% vs 98%.

The gap widens slight with higher levels. I have been toying with the idea of making the level bonus +4 for in class skills with Iridium V2. Still, you get the idea.

Part of it is what your interpretation of "Realistic" is. Genre emulation or modeling of every possible real world outcome or modeling studies of learning behavior? Personally, I think any true modeling of real world learning methods would also need a method of "forgetting" unpracticed skills, learning by practice, incorrect learning by practicing the wrong methods, and instruction education. It would be a fantastically complex system that tracks actions, the number of times an action is performed, the time since you last used a skill, a series of overlapping skills (I am a decent carpenter but have never installed made a bow, a specialized skill of working with wood). Not saying it is impossible but at some point you have to ask if it is worth it.

Bill

Quote from: Kyle AaronNot necessarily. It's possible to have a system based on attribute + skill, but allow that skills default to one another; if you have X in one skill, you might have X/2, or attribute +1, in another related skill.

True, but mathematically, I think this just works out to the equivalent of having an intermediate quality, let's call it aptitude, between attributes and skills. Narrower than attributes, wider than skills, easier to raise than attributes, harder than skills.

I pointed more or less in this direction when I started a thread on skill growth by cluster.

If you're looking for realistic, I also think you should consider that some things we think of as "attributes" in the RPG world are really only a measure of the correlation of groups of related skills--"strength" is a raw quality, dexterity and intelligence less so. So you might simplify the issue by first trying to take "attributes" out of the equation--not permanently but as an exercise. (They're a useful abstraction.)

E.g. ask, if someone has a given learning curve for studying/practicing skill X in isolation, what does that say about their learning curve for skill Y in isolation? And if a person actually has skill P at a certain level, what would their ability with skill Q be, approximately, even if they're never studied it?

On another tack, attribute + skill isn't the only way to consider the interaction. E.g. in Harnmaster skills are skills, but if you have high related attributes you'll start at a higher skill and progress faster.