Linear skill ratings deciding dice pool size is broken

[Christoffer Lernö]

Adding to the "broken" series here. Let's look at the popular dice pool decided on a skill rating. Most systems have a simple: skill level x gives x dice.

Similar to the Linear Die Roll Modifiers Are Broken thread. Let's have a look at the problems.

Basically, we have the problem of skill rating 0. Skill rating 0 doesn't work with the dice pool (although it can be tweaked). This gives us the requirement that the dice pool either always starts at 1 or 0 implies immeditate failure.

A consequence is that modifying number of dice rather than target number suffers from a cut-off similar to that of linear rolls with modifiers.

However, target number create their own problems, most of which are outside the scope of this discussion.

Beyond the skill 0 problem, we have the skill 1 compared to skill 2 problem.

IN SOME CASES it might be good that there's a difference between an opposed roll between "a person with skill 1 competing with a skill 2 person" and "skill 11 vs. skill 12". In other cases, this behaviour might be a "broken" feature (if we follow a similar judgement on the behavious of linear rolls).

One of the most redeeming features of linear skill systems is that it's possible to create tests where only the relative difference between ratings are relevant.

For example A has 90% in skill X, B has 95%. If they use an opposed test you can decide this with a single roll where the chance of B winning is (50%+skill of B-skill of A=) 55%. The same chance of B winning happens if A has 10% and B has 15%.

(Unfortunately, BRP uses individual rolls in many cases where this type of resisted rolls would be more appropriate. One such example is combat. If two persons with 10% weapon skills duke it out it's gonna take a long time to find a winner, whereas two with 90% will have a very quick combat, had BRP used the resistance roll method, both situations would have taken the same amount of time. This might not be 100% realistic, but it's an significant improvement over the original mechanic)

Anyway, this also means that the improvment from 11 to 12 is not the same as from 1 to 2 in terms of improving chances in many cases of dice pool mechanic.
To summarize:

* skill 0 cut-off
* skill 1 vs skill 2 is not the same as skill 11 vs skill 12
* skill improvement efficiency

It should be noted that these things might be interpreted as advantages and sought-after behaviour. I'd like to argue that the same is true of the linear modifiers.

/C

[Mike Holmes]

- Skill rating 0

This is a range problem, and hardly particular to die pools. I could make the same mistake and have a target number of 21 on a D20+Skill system. That would make skill zero ineffective as well. The "solutions" to this problem are simple, effective, and do not cause problems themselves (unlike the range problems with linear systems, which almost always have clunky solutions). So how is this even a problem?

- Skill 1 vs Skill 2 compared to 11 v 12 and efficiency

I think there definitely should be a difference, one player has twice the skill of the other in the forst case, whereas in the second they are very close proportionally. This is exactly the problem with linear systems, and exactly how die pools fix the problem.

And the player who has 12 dice is 12 times better than the skill 1 person in most systems. So I'm not seeing the efficiency drop off you imply. If you refer to the marginal efficiency, that should drop off as the law of diminishing returns states it should in all natural systems.

The linear system advantage that you point out has me confused. What advantage does it have over die pools? First, BRP is not a die pool system, it's linear. Second, die pools can easily do relative rating comparison, I made just such a system for a RPG/Wargame I was working on. The lower skilled player always rolled 2 dice, and the higher always rolled the difference between the skills plus 2. Seemed to work pretty well.

In any case, it's a not a very good argument to say that Linear systems are not broken because another method is broken.

I am very much of the opinion that the only potential drawback to die pools is rolling lots of dice. And since I rather like rolling lots of dice, actually, I like dice pools a lot.

Mike

[Christoffer Lernö]

- Skill rating 0

This is a range problem, and hardly particular to die pools. I could make the same mistake and have a target number of 21 on a D20+Skill system. That would make skill zero ineffective as well. The "solutions" to this problem are simple, effective, and do not cause problems themselves (unlike the range problems with linear systems, which almost always have clunky solutions). So how is this even a problem?

In some dice pool systems this becomes a problem. Consider a system where at least 1 "success" (to use the SR terminology) is a hit when you shoot at a target.

Let us start with a target number of 6 on a D6. Chance to succeed is then 1/6 (about 17%). Two dice, 11/36 (about 30%). Three dice, 91/216 (about 42%) and so on. This caps the lowest skill level at 17%. There is no way to have 6% chance to succeed unless we extend the target number range SR style. This behavious might be considered broken in some cases. (but more about "broken" later)

- Skill 1 vs Skill 2 compared to 11 v 12 and efficiency

I think there definitely should be a difference, one player has twice the skill of the other in the forst case, whereas in the second they are very close proportionally. This is exactly the problem with linear systems, and exactly how die pools fix the problem.

Skill 1 vs Skill 2 in the example above gives skill 2 as being 83% better than skill 1. Skill 11 has about 87% chance of success. Skill 12 has 89% chance of success, which is an increase of less than 3%.

On one hand this is the strength of the dice pool system (the asymptotic behaviour of the probability), however in opposed rolls - present in some systems, such as SR - may have problems.

For example in SR unarmed combat is resolved by having both fighters roll their skill against a target number of 4. The most successes wins.

Consider the case of skill 1 vs skill 2 and skill 11 vs skill 12. There will be less difference between the skills in the latter case. This can be considered broken OR a desired effect. I personally feel it's not working well, but again this is a matter of desired behaviour (this ties in to the broken stuff later)

The linear system advantage that you point out has me confused. What advantage does it have over die pools?

Only that you more easily can make use of the difference between two skills for contest resolution than you would in a dice pool system. See my original posting of an example. This is the common resistance roll (I think it's also part of BRP)

First, BRP is not a die pool system, it's linear.

Of course, I only mentioned this that BRP could have made use of the resistance roll principle in their combat system, but failed to do so and in that way missed out on a good opportunity.

Second, die pools can easily do relative rating comparison, I made just such a system for a RPG/Wargame I was working on. The lower skilled player always rolled 2 dice, and the higher always rolled the difference between the skills plus 2. Seemed to work pretty well.

Yes, that kind of would be grafting the linear resistance roll principle onto the dice pool. I note that you created an arbitrary zero point at 2 dice. For linear systems this can conveniently be set at 50%. I'm not saying that your solution is bad though. In fact I think that's a very functional way of incorporating relative ratings which I think ought to be used more with dice pools.

In any case, it's a not a very good argument to say that Linear systems are not broken because another method is broken.

I agree. But you make it sound that my argument ran along those lines, which is not the case.

To quote myself:

It should be noted that these things might be interpreted as advantages and sought-after behaviour. I'd like to argue that the same is true of the linear modifiers.

I brought up these examples, not to argue that "dice pools are broken, don't use em!". Rather I wanted to point out some common weaknesses which are incorporated in some dice pool systems. However, those weaknesses are not universally present, so keeping in mind that they really ARE THERE might help people from building in bugs into their systems.

That's one of the reasons for my post. The second is to illustrate a counter-point to the "linear modifiers are broken" post. The point being that "broken" behaviour can actually be desired behaviour, it's all a matter of what the original intention of the mechanic was.

[Walt Freitag]

Okay, I seem to have let this "broken" beast out of its cage, and now it's running amuck through various threads, chewing on the upholstery and leaving various unsightly messes. Let me see if I can tame it a bit, without having to put it down entirely...

When I call a mechanism broken, I mean either or both of two things:

1. It doesn't do what it's intended to do.

2. Whatever it does do has adverse effects on the system design as a whole, and/or on play.

In the case of additive modifiers, I pointed out and explained in detail what I believe are instances of both of these problems. However:

Very few game mechanisms include statements describing in detail what they are intended to do. Therefore, what a mechanism is intended to do is a matter of inference. Therefore, when I say a mechanism doesn't do what it's intended to do, I could be wrong. This is partly a subjective matter. But not entirely. It also depends on the quality of the inference. If I say a system is broken because it becomes difficult to run with fifteen players, I'm inferring that the system is intended to be able to run easily with fifteen players. This would be a poor inference. If I say a system is broken because it breaks down when trying to run player-characters of fifteenth level, I'm inferring that the system is supposed to be able to run with characters of fifteenth level. That would be a sound inference if the system's sourcebooks include rules and tables for player-characters up to twentieth level. Pointing out that most who play those systems don't attempt to play fifteenth level characters so it wouldn't be broken for them would not be a sound refutation, though it is probably a fact worth mentioning for overall perspective.

Adverse effects, even more, are subjective. Different people want different kinds of results from game mechanisms. Therefore, when I point out effects of a mechanism and call them adverse, I could be wrong. Whether an effect is adverse or beneficial, and whether the severity of adverse effects is singificant compared with the alternatives, is largely a subjective matter. But not entirely. Like all creative endeavors, some more or less consensual critical values do exist. Freedom of choice in character generation is usually considered a good thing. Additional complexity that does not "pay off" in additional depth or verisimilitude is usually considered a negative. Loopholes that greatly increase or decrease a character's effectiveness relative to what the situation being described appears to warrant are usually considered a negative. There are exceptions to every "usually," but those exceptions don't invalidate the notion of quality completely, otherwise no system would be better than any other.

So I accept, fully, the notion that when I say a mechanism is broken, others will say no it isn't: that I'm misinterpreting what it's intended to do; that effects I think are adverse are actually beneficial; that the adverse effects I see are not severe enough to be a problem; that the adverse effects have simple cures; that I'm overlooking worthwhile exceptions. All those specific issues I'm willing to discuss or debate, because there's content there.

What I don't see as productive is reacting as if someone at WoTC is going to see my posts and say, "Oh my god, Walt says D20 is broken. We'd better pull it off the market right away," and that if I can just be convinced to withdraw the word "broken" they might change their minds. I'm honored that some people apparently hold my opinions in such high esteem, but believe me, this isn't worth worrying about.

Christoffer, you appear to be arguing that "broken" is a useless description because one can point to any feature -- a condition designated with a number 1 that you think should be designated with a number 0 -- and arbitrarily call it broken. But doing so, without making an argument that points to either a mechanism not doing what it's supposed to, or adverse effects on play, is meaningless. I think you probably could describe an adverse effect from the one-die-minimum you describe (a player-character who wants a lower skill level, but has to settle for the minimum of 17%?), but you haven't done so.

Interestingly, Symmetry "fixes" this exact "problem," creating a dice pool mechanism that deals with small success chances with far greater precision. Why, then, did I not call dice pool systems that fail to do so "broken?" Because I'm not aware of any that has an implied intention to represent small success chances with precision, and I'm not aware of any for which that characteristic causes adverse effects on system design or play. But I'd be interested in any such cases anyone wants to call to my attention, because they'd be prime candidates for using the Symmetry mechanism instead.

- Walt

[Christoffer Lernö]

Sorry Walt. The "broken" was a consciously used teaser for the subject of the thread. I just throught it interesting to point out the limitations of the dice pools since we were dealing with the whole dice mechanic thing anyway. Nothing personal ;)

[Syphon]

Basically, we have the problem of skill rating 0. Skill rating 0 doesn't work with the dice pool (although it can be tweaked). This gives us the requirement that the dice pool either always starts at 1 or 0 implies immeditate failure.

There is a simple solution to the skill rating 0 problem with dice pool systems. Have any player with a skill rating 0 roll 2 dice and keep the lowest roll. Simple, and effective.

[Andrew Martin]

Basically, we have the problem of skill rating 0. Skill rating 0 doesn't work with the dice pool (although it can be tweaked). This gives us the requirement that the dice pool either always starts at 1 or 0 implies immeditate failure.

There is a simple solution to the skill rating 0 problem with dice pool systems. Have any player with a skill rating 0 roll 2 dice and keep the lowest roll. Simple, and effective.

I believe that 0 skill should be automatic not success. After all, it's implied by the level of skill being zero.

[Ron Edwards]

Hi there,

I think the last couple of posts are getting ahead of the basic topic of the thread. The assertions made by Syphon and Andrew are all well and good for specific designs that employ dice pools (by whatever definition), but there's nothing inherently "good" or "bad" about the suggestions themselves, out of the context of those specific designs.

In other words, I've reviewed the thread, and I'm reasonably convinced that Christoffer's points have been sufficiently discussed. Not much else is getting added - unless Christoffer expresses a desire for this to continue, then the thread needs to be left alone.

Best,
Ron