Quote from: JoeNuttall;854556Ideally you should decide  what you want the results of the system to be, then design a system that gives those results, and then use the system as an arbiter.

It should be possible for an expert to miss punching an amateur, and it should be possible for the amateur to land a punch.
Also it should be possible for the amateur to spot something hidden in the undergrowth which the expert missed.
However rolling every time a pianist performs Beethoven's Hammerklavier with the potential for an amateur to mostly fail abysmally, but occasionally play a stunning performance, would be ridiculous.
Similarly it might make sense for it to be impossible for a PC to climb something, but it should not be impossible (just very unlikely) for the expert to fail.

How to model such different situations though is tricky to do well.

Yup.

One solution is simple: just DO NOT use the same system for throwing a punch and for climbing. For D&D, for example, use 1d20 for throwing a punch and some other method (1d6, 2d6, 3d6 roll under, dice pools, etc) for skills, comparison of abilities, etc.

I know most "modern" systems believe that using a single mechanic is the right thing to do, but it just doesn't work, unless the GM modifies the rule on the fly (okay, he is twice as strong as you, so you have no chance of beating him, don't roll).

I wrote extensively about it here, although my examples are all D&D.

TBH, I cannot think of an easy way to make this work with d%.

And I like d%.

Quote from: Bloody Stupid Johnson;854463I think generally I don't want experts to perform >>> than non-experts,  because I don't want too much divergence in character abilities, realism notwithstanding. For me I'll say its probably most fun to have most rolls sitting at somewhere between 25% and 75% or the equivalent, say. (5E, Savage Worlds and Palladium being my go-to games currently).

It seems to me that in 5E and in Palladium, I think there is a lot of divergence in character abilities between different classes. For example, in 5E, there is a lot of divergence not just between a fighter and a wizard, say, but also between a fighter, a rogue, and a barbarian.

Do you think that divergence between classes is OK, but divergence between different skills chosen is a problem?

Quote from: Eric Diaz;854577One solution is simple: just DO NOT use the same system for throwing a punch and for climbing. For D&D, for example, use 1d20 for throwing a punch and some other method (1d6, 2d6, 3d6 roll under, dice pools, etc) for skills, comparison of abilities, etc.

The other methods you refer to are ways of making the bonus count for more, which are very similar in their effect to Bren's suggestion of doubling the % chance.
The issue with this method is that it's not clear which circumstances you should use which system (no roll, d6 roll, d20 roll). E.g.  in my examples above climbing is a case where skill matters a lot, but perception isn't, so I don't think it's clear cut when you should use which system.

Quote from: JoeNuttall;854608The other methods you refer to are ways of making the bonus count for more, which are very similar in their effect to Brett's suggestion of doubling the % chance.
The issue with this method is that it's not clear which circumstances you should use which system (no roll, d6 roll, d20 roll). E.g.  in my examples above climbing is a case where skill matters a lot, but perception isn't, so I don't think it's clear cut when you should use which system.

Doubling the % chance is good, and I guess it might be enough if an expert has at least, say, 48% chance to begin with (failing at an average task more than 4% sounds like too much for a specialist IMO). But what can you do when you face a hard task? If you halve the score, for example, the difference between expert and novice is diminished. And how do you compare two different scores, like in a car chase, for example?

If you use 2d6 or 3d6, you have the advantage of creating a bell curve, which is a bit harder to do with d%.

Circumstances are pretty clear to me: combat and non-combat, or "high action and no high action" if you prefer. Things that take a moment should be "swingy" (throwing a punch, shooting an arrow), things that take more than that shouldn't (playing a song, climbing a wall, etc). BTW, I think the character with the best perception should succeed more often, too, like the climber, the piano player, etc.

Quote from: Eric DiazOne solution is simple: just DO NOT use the same system for throwing a punch and for climbing. For D&D, for example, use 1d20 for throwing a punch and some other method (1d6, 2d6, 3d6 roll under, dice pools, etc) for skills, comparison of abilities, etc.

Quote from: JoeNuttall;854608The other methods you refer to are ways of making the bonus count for more, which are very similar in their effect to Brett's suggestion of doubling the % chance.
The issue with this method is that it's not clear which circumstances you should use which system (no roll, d6 roll, d20 roll). E.g.  in my examples above climbing is a case where skill matters a lot, but perception isn't, so I don't think it's clear cut when you should use which system.

Another way to approach this is to have it built into how the numbers for each skill work out.

For example, if maximum skill is 15, then if everyone starts out with a default of Perception 10 - then being an expert perceiver doesn't make as much of a difference. (Whereas, say, skill in speaking Chinese could start at zero - so a beginner has no chance to get a result like an expert.) Similarly, if you wanted, default for combat skills could be high so that an expert doesn't completely outclass a beginner.

In other words, you can change the range of the other numbers, rather than changing the number rolled on the die. Any 1d6 roll mechanics can be pretty trivially mapped to 1d20 roll mechanics.

Quote from: jhkim;854622Another way to approach this is to have it built into how the numbers for each skill work out.

For example, if maximum skill is 15, then if everyone starts out with a default of Perception 10 - then being an expert perceiver doesn't make as much of a difference. (Whereas, say, skill in speaking Chinese could start at zero - so a beginner has no chance to get a result like an expert.) Similarly, if you wanted, default for combat skills could be high so that an expert doesn't completely outclass a beginner.

In other words, you can change the range of the other numbers, rather than changing the number rolled on the die. Any 1d6 roll mechanics can be pretty trivially mapped to 1d20 roll mechanics.

Yes, this makes sense, although it can get complicated: not only you need different "base" numbers for each different skill, you also have to be careful that investing some skill-points to learn Chinese won't completely ruin all your chances on being a good fighter (TBH I don't play any game in which you'd invest skill points to learn Chinese, right now).

In any case, it might work, and it is a good fit to that versions of BRP with lots of skill defaults in the blank character sheet.

Quote from: Eric Diaz;854618Circumstances are pretty clear to me: combat and non-combat, or "high action and no high action" if you prefer. Things that take a moment should be "swingy" (throwing a punch, shooting an arrow), things that take more than that shouldn't (playing a song, climbing a wall, etc).

An interesting perspective. How would you count a careful measured arrow shot e.g. an ambush? Or climbing a wall during combat?
 

Quote from: Eric Diaz;854618BTW, I think the character with the best perception should succeed more often, too, like the climber, the piano player, etc.

I meant that the novice climber / piano player should never be able to perform an expert level task, whereas the better perception should *mostly* do better.

Silhouette system (by Dream Pod 9) had a system where characters skills were rated as Skill Level/Complexity. The Skill level was the raw talent and ability to get the job done while Complexity was breath of knowledge (experience and well rounded education). Thus you could have something like Negotiation 2/1 or Negotiation 3/3.

It worked on principle that every task would have a complexity rating and if it was more than character's skill's complexity, there would be negative modifier. If task had complexity less than character's skill's complexity there would be positive modifier to roll.

This was a neat idea but in my memory it was ignored by most people I know who played Silhouette based systems as too cumbersome.

I have toyed (for my homebrew RPG) a skill system where chance of critical failure is normally very low (natural 3 with 3D6) but it is much higher if character tries something very complex (say complexity of 3 and character's expertise of 1 would mean the natural roll for critical failure would be 3 + 3-1 --> 5 so all natural die rolls of 3, 4 and 5 with 3D6 would lead to disaster. However, a mountaineering expert would have 3 + 3-3 = 3 as the critical failure. This system is intended to make more skilled people more safe to work with in adversarial weather...

Quote from: JoeNuttall;854627An interesting perspective. How would you count a careful measured arrow shot e.g. an ambush? Or climbing a wall during combat?

I would still call  a careful measured arrow shot during an ambush "combat" (highly unpredictable on a second-by-second basis).

Climbing a wall is not a "combat" task IMO, even in the middle of combat.

I would save the d20 for actions that are resolved in a moment (maybe a few seconds); if I were to reduce combat to a single roll, I wouldn't use 1d20.

Outside of combat, I would only use it against other threats that are resolved in a moment, like dodging a fire ball trap.

I can see the argument for using 3d6 (etc) for a careful measured arrow shot, specially at an archery contest, stationary target, etc, but I don't think this applies to combat.

Quote from: jhkim;854586It seems to me that in 5E and in Palladium, I think there is a lot of divergence in character abilities between different classes. For example, in 5E, there is a lot of divergence not just between a fighter and a wizard, say, but also between a fighter, a rogue, and a barbarian.

Do you think that divergence between classes is OK, but divergence between different skills chosen is a problem?

In 5E I don't think there's that much divergence classes, actually. Characters pick up many of their skills from background (and sometimes one or two from race); all classes have the same base to-hit chance; a proficiency bonus is only a +2 (1st level) to +6 (at 17th), though occasionally a character may have Expertise and double that, or have some other ability that grants Advantage (roll twice and take the best).
So I can't answer the second part of your question for 5E specifically, but in general I think its best if all characters can contribute in some fashion toward achieving a majority of challenges, and have an adequate chance of succeeding an attack or challenge directed specifically at them (the sort of business that gets covered off under saving throws in D&D). Different classes may be able to get around a problem or contribute to a solution in different ways, which is OK.

On Palladium, I listed it above because if you have the skill its going to be, probably, at a range where rolling it is going to have a possibility of failing. The thing to note is that while it does have substantial divergence between classes (e.g. the old Vagabond vs. Dragon Hatchling problem in Rifts), thats not so much a case of individual skills differing as a case of the character's aggregate capabilities across dozens of skills (how many skills a character has, combat abilities, magic, psionics, etc). Mostly my solution to this to not to play Vagabonds.

Quote from: Eric Diaz;854636I can see the argument for using 3d6 (etc) for a careful measured arrow shot, specially at an archery contest, stationary target, etc, but I don't think this applies to combat.

Shooting in a contest seems like it should be easier than shooting in a combat situation or at least any combat situation where the shooter is also under fire or could come under other attack.

One option to avoid single roll luck and unluck is to use more than one roll. One could make the contest 2 out of 3. This decreases the odds that a good shooter will bomb out in the early rounds. Use multiple rounds to decrease the odds that a lucky shooter will win. I did this when I created rules for Shooting Tournament (with black powder muskets) for my Honor+Intrigue game. I talk about it and list the rules on my blog: Rules for a Shooting Contest.

Quote from: Phillip;853974That's why we don't give able seamen a mere 50% chance in the games of my acquaintance, regardless of dice used when a toss is called for.

Good for you, but that's obviously not how the mechanic works in a pressure situation in D&D, nor for a number of other games.

Isn't all this why we have a GM? To decide that, well, some things don't need a roll.

Or that if you're rolling, it's not for success or failure, it's to see how well you do. The locksmith will open that door, but will it take 1 minute or 10? The surgeon will save your leg, but do you need 1 week of physio or 10? And so on.

Quote from: Eric Diaz;854577I wrote extensively about it here, although my examples are all D&D.

TBH, I cannot think of an easy way to make this work with d%.

And I like d%.

3d34-2 instead of d100 I've heard d34s are out there somewhere - though I imagine they're expensive.

The equivalent of the 'double %' method (Bren's?) with d20, or a similar additive system, might be just to have both sides double or halve their bonus. That works to shift the relative balance of power on opposed contests, anyway. An additive system could also use a die size based on how uncertain the task is (a character might roll d4+bonuses for an arm wrestle, d10+bonuses for archery, d20+bonuses in combat...).
(EDIT: to be clear, I think that's easier with the lower-size dice, d4 to d12, so you might have a scale like
d4+bonuses: A foregone conclusion. Mostly determined by ability e.g. contests of strength or speed under controlled conditions (arm wrestling, foot races).
d6+bonuses: Slightly chancy. (?)
d8+bonuses: Not very random. Running in uncertain terrain, no-holds-barred grappling (in melee conditions).
d10+bonuses: Fairly random. Notable external factors (outdoor archery with varying winds).
d12+bonuses: Pretty chancy. (social situations?)
d20+bonuses: Wildly erratic. Games of chance where skill is only slightly useful, chaotic melee, fishing or other tasks where success is heavily influenced by external factors.  )

Non-opposed tasks might be vs. a target numbers of [average die roll + an extra difficulty modifier].

Quote from: Bloody Stupid Johnson;8547253d34-2 instead of d100 I've heard d34s are out there somewhere - though I imagine they're expensive.

The equivalent of the 'double %' method (Bren's?) with d20, or a similar additive system, might be just to have both sides double or halve their bonus. That works to shift the relative balance of power on opposed contests, anyway. An additive system could also use a die size based on how uncertain the task is (a character might roll d4+bonuses for an arm wrestle, d10+bonuses for archery, d20+bonuses in combat...).
(EDIT: to be clear, I think that's easier with the lower-size dice, d4 to d12, so you might have a scale like
d4+bonuses: A foregone conclusion. Mostly determined by ability e.g. contests of strength or speed under controlled conditions (arm wrestling, foot races).
d6+bonuses: Slightly chancy. (?)
d8+bonuses: Not very random. Running in uncertain terrain, no-holds-barred grappling (in melee conditions).
d10+bonuses: Fairly random. Notable external factors (outdoor archery with varying winds).
d12+bonuses: Pretty chancy. (social situations?)
d20+bonuses: Wildly erratic. Games of chance where skill is only slightly useful, chaotic melee, fishing or other tasks where success is heavily influenced by external factors.  )

Non-opposed tasks might be vs. a target numbers of [average die roll + an extra difficulty modifier].

For d100, you could also use 5d20, I guess... although 17+9+12+4+15 isn't something Id like to do often.

About the rest, yes, this could work. I prefer using 3d6 instead of 1d20 because I wouldn't need to change TNs, but 1d10+5 would work well too.