Teach me about your favorite skill system

[PhoenixPhyre]

I want to learn more about the wide variations in systems out there. Right now, I'm focusing on "skill" systems. Please restrict yourself to answering about systems you like so we can remain positive. Please refrain from bashing other systems .

Things I'd like to know:

Name of system (including edition if that matters)
Basic mechanic: Roll under (% or dX), roll vs TN (fixed? varying?), dice pool (counting successes), ??
How specific are the skills? very general (eg Fight) to very specific (Knowledge (specific region))
Is combat done using the skill system? (a Shoot skill, for example)
How do skills improve? With level (if such exists), by spending XP, ???
How are different difficulties distinguished? Bonuses to skills, changing TN, changing dice pools, ??
Can I use skills untrained?
What part do you think it handles best?
How frequently do you have to refer to tables (either as a player or as a GM)?

[NichG]

I've found that my preference for skill systems has gravitated towards discrete 'X ranks lets you do Y' type things as opposed to 'roll vs DC' type things. There isn't a specific system I'd point at for it, but you see elements of that design in piecemeal in a variety of systems - 7th Sea has it for some of the sorcerous knacks, L5R has special threshold values for some of its skills, World of Darkness tends to do it for supernatural abilities, even D&D has its skill tricks and things like the Epic Skill Usages chart (which might as well be fixed thresholds since the numbers are much greater than the range on a d20 anyhow).

So when I design systems now, if I'm going to have a skill, I ask myself 'can I think of 5-7 distinct types of things that either should be impossible for an untrained person to do consistently or have clear numerical demarcations that are a function of expertise, but would become possible as someone masters this skill, to assign to each rank?'. If I can't think of such things, I tend to fold the uses of the skill into being associated with a generic attribute instead, or rethink how that interaction works in the game entirely.

So for example, while anyone could theoretically move quietly or hide (so it initially fails the 'clear demarcation' question), maybe I can instead say that each rank of Stealth gives you a single extra chance to remain hidden even if you do something that should have given you away. So anyone can hide, but someone trained in Stealth can recover even if they e.g. knock over something or turn a corner and run into a guard. So now each rank can correspond to a distinct thing (number of recoveries you get). Or, maybe everyone has some ability to hide, but someone with the first rank of Stealth can know absolutely when they've been spotted, with the second rank can attack from hiding without giving away their position. with the third rank can pass through even closed doors and other directly observed points without breaking Stealth, etc. So each rank provides a specific type of 'thing you can do'.

Skill ranks in such systems can still also fold into rolls or other random factors, but the distinct per-rank or threshold points end up being sort of inspiring when thinking about character advancement in a way that just improving success rates doesn't quite manage to be.

[PhoenixPhyre]

I've found that my preference for skill systems has gravitated towards discrete 'X ranks lets you do Y' type things as opposed to 'roll vs DC' type things. There isn't a specific system I'd point at for it, but you see elements of that design in piecemeal in a variety of systems - 7th Sea has it for some of the sorcerous knacks, L5R has special threshold values for some of its skills, World of Darkness tends to do it for supernatural abilities, even D&D has its skill tricks and things like the Epic Skill Usages chart (which might as well be fixed thresholds since the numbers are much greater than the range on a d20 anyhow).

So when I design systems now, if I'm going to have a skill, I ask myself 'can I think of 5-7 distinct types of things that either should be impossible for an untrained person to do consistently or have clear numerical demarcations that are a function of expertise, but would become possible as someone masters this skill, to assign to each rank?'. If I can't think of such things, I tend to fold the uses of the skill into being associated with a generic attribute instead, or rethink how that interaction works in the game entirely.

So for example, while anyone could theoretically move quietly or hide (so it initially fails the 'clear demarcation' question), maybe I can instead say that each rank of Stealth gives you a single extra chance to remain hidden even if you do something that should have given you away. So anyone can hide, but someone trained in Stealth can recover even if they e.g. knock over something or turn a corner and run into a guard. So now each rank can correspond to a distinct thing (number of recoveries you get). Or, maybe everyone has some ability to hide, but someone with the first rank of Stealth can know absolutely when they've been spotted, with the second rank can attack from hiding without giving away their position. with the third rank can pass through even closed doors and other directly observed points without breaking Stealth, etc. So each rank provides a specific type of 'thing you can do'.

Skill ranks in such systems can still also fold into rolls or other random factors, but the distinct per-rank or threshold points end up being sort of inspiring when thinking about character advancement in a way that just improving success rates doesn't quite manage to be.

Would that include combat? And would those be just blanket permission (as in "with X ranks in Spot you can see invisible things") without consideration for the fiction?

[Max_Killjoy]

Specific to Skills, I still like the general concept of how HERO does it (or at least did it in 4th/5th edition, some changes in 6th turned me off).

* You have Characteristics, 0 to 20 for normal human beings, average default for 0 character points is 10.

* Skills usually cost 3 character points, and have an associated Characteristic.

* A successful roll on a Skill is <= a value determined by 9+{Characteristic/5} on 3d6, (bonus or penalty aside).

* So someone with the relevant Characteristic at 10 starts the Skill at "11-" (eleven or less), someone with the Char at 20 starts at 13-.

The difference between 11- and 13- is bigger than it looks because of the curve on 3d6.

There's also a diminishing return built in where raising your skill from 15- to 16- costs the same number of points, but gets less benefit than raising it from 14- to 15-, which also gets less benefit than raising it from 13- to 14-.

[oxybe]

Depends on the style of game.

For the Epic (by which I mean it shows leaps and bounds of character growth/power as you play), I tend to prefer 4th ed D&D's take: a small, concise skill list that focuses on the themes of the game (in this case adventuring). As you level, you get better at all things related to adventuring (either due to exposure, talking about it with other PCs, independent study, etc...), but your class/specializations/items/feats allow you to further specialize in your field.

For the more grounded game (where there i growth, but the scope rarely changes power-wise), I've grown fond of FFG's Genesys/star wars.

you have d6s, d8s and d12s, both for your total skill & penalties: d6's are simply success/failures, d8's add narrative advantage/disadvantage, while d12 further add the equivalent of a crit success/fail.

to figure out your dice pool, find the skill & associated stat. the highest is the number of d8's you roll, the lowest is how many of those d8's you upgrade into d12's. d6's are more incidental, like from taking time to aim or another PC granting you a boon to the action you're taking.

difficulties are using d8's for the base, with being upgraded to d12's for particularly disadvantageous conditions, and d6's as incidentals, like some minor cloud cover from dust.

so in addition to simply deciding success/failure by comparing success pips VS failure pips, it can allow you to gain or grant beneficial boons to your allies or on future checks. of course this requires special dice so the whole of it is YMMV, but it's interesting in play, at the very least.

[Quertus]

Wow. It's actually amazingly difficult for me to describe a skill system purely positively. But here's as positive as I can get:

I love 2e D&D, overall, as a game, the most. Its skill system is... acceptable. In particular, it's just a d20 roll under your stat with appropriate bonus or penalty, and generally encourages characters to get broader as they level (unlike 3e). And, very importantly, there are approximately zero social skills. So, the skill system largely gets out of my way, and gives a concrete answer when one is needed. It's acceptable.

I love how Mutants and Masterminds makes skills cost way less build points than most anything else (d20 system, 4 skill points for 1 build point (compared to, say, one attribute point for 1 build point)). I love that you don't feel like you failed at the game by making your character actually skilled at something, or, heaven forbid, trying to build a skill monkey (which is just an epic fail in many point buy systems).

I've found that my preference for skill systems has gravitated towards discrete 'X ranks lets you do Y' type things as opposed to 'roll vs DC' type things... World of Darkness tends to do it for supernatural abilities,

Huh. I'd never thought of WoD supernatural abilities as "skills" before. So, yeah, while their general skill system rubs me the wrong way (I don't know why, but the various dice pool skills, including talents, knowledges, and combat skills just never did it for me), the "x level of skill means y capabilities" in their powers is my cup of tea.

One homebrew system I rather liked ran much like this. Skills had definite levels. Want to perform surgery? You need Medic III. (You can technically attempt it at Medic II under the correct circumstances, but failure is more likely, and failure results are worse).

EDIT: I think I know why I hate WoD skill systems in general - I like being an expert in something to mean something. But, in WoD, the dice pool math means that the expert can fail or botch almost as easily as the noob - and both can usually attempt the same complexity of tasks. I like skills to include definite benchmarks and/or notable, meaningful differences. In D&D, "Quertus knows more than the gods / understands magic better than the god(desse)s of magic" is almost an interesting benchmark, but the skill system is often lackluster in providing meaningful differentiations otherwise.

[PhoenixPhyre]

I'm seeing a philosophical split here:

Some systems use a permission slip model: pay your dues (build points, etc) and you can do X. Don't, and you can't. This allows you to infer backward--can do X => has Y "ability". These also have fewer "ranks", but each rank is more significant. Handles more difficult tasks by requiring higher ranks to do. No partial success possible(?)

Others use a capability improvement model: anyone can attempt X; people with more ranks have a better chance of doing X or can get better results when doing X. More granular--low "rank" people may be able to do harder tasks--but also more prone to calibration issues. Doesn't provide quite as clear a signal about capabilities (being less binary).

Is that an accurate (or at least not horribly inaccurate) grouping? I recognize that there are systems that share traits of both philosophies.

[Quertus]

I'm seeing a philosophical split here:

Some systems use a permission slip model: pay your dues (build points, etc) and you can do X. Don't, and you can't. This allows you to infer backward--can do X => has Y "ability". These also have fewer "ranks", but each rank is more significant. Handles more difficult tasks by requiring higher ranks to do. No partial success possible(?)

Others use a capability improvement model: anyone can attempt X; people with more ranks have a better chance of doing X or can get better results when doing X. More granular--low "rank" people may be able to do harder tasks--but also more prone to calibration issues. Doesn't provide quite as clear a signal about capabilities (being less binary).

Is that an accurate (or at least not horribly inaccurate) grouping? I recognize that there are systems that share traits of both philosophies.

Very much this, I believe.

However, while I'm primarily in the permission slip camp, I can enjoy a degrees of success style game (which largely falls under the "better results" that you put in the capability improvement section) - especially when it more accurately models how I perceive reality working.

[Cluedrew]

Spoiler: Roll for Shoes

Show

[LIST][*]Say what you do and roll a number of d6s.[*]If the sum of your roll is higher than the opposing roll (either another player or the DM), the thing you wanted to happen, happens.[*]The number of the d6s you roll is determined by the level of skill you have.[*]At start, you have only one skill: Do anything 1.[*]If you roll all sixes on your roll, you can get new skill one level higher than the one you used for the action. The skill must be a subset of what happened to you in the action (Say, Athletics 2 if you were climbing a wall, or Teeth of Biting 2 if you were eating a cake).[*]For every roll you fail, you get 1 XP.[*]XP can be used to change a die into a 6 for advancement purposes but not for success purposes.

I just went ahead and dumped the entire system in there for context. It is not the best system overall by any means, but it does some really cool stuff. Most of which actually come down to the skill system.

Mostly in that the dynamic skill system. It makes you better at the things you do more often, it also rewards your just running with it what you get instead of trying to plan everything out. (I didn't know that my character played soccer, but they must of because they some how ended mastering the use of knocking out enemies using a globe they found.) It also encourages taking risks. The times I have played it I have often ended up with a really strong character late game because I just try random things. When they go badly I get XP, when they go well (which happens more often with lower level skills) I get a new skill.

The fact there are no skill lists, it can sometimes be hard to pin down exactly what a skill should do or what a new skill can be can be problematic. I have considered trying to make a rules medium version of the system that solves some of that. Although if you are playing role for shows, don't worry about it too much.

[Max_Killjoy]

EDIT: I think I know why I hate WoD skill systems in general - I like being an expert in something to mean something. But, in WoD, the dice pool math means that the expert can fail or botch almost as easily as the noob - and both can usually attempt the same complexity of tasks. I like skills to include definite benchmarks and/or notable, meaningful differences. In D&D, "Quertus knows more than the gods / understands magic better than the god(desse)s of magic" is almost an interesting benchmark, but the skill system is often lackluster in providing meaningful differentiations otherwise.

Both of these were addressed to some degree in the oWoD system, but the changes seem to have been often ignored by the playerbase, even across multiple editions (that is, many seem to have learned the original botch rule and never unlearned it no matter how often WW published or pointed out the new botch rule).

* The botch math was changed such that 1s could still offset successes, but the only way to actually botch was to gets one or more 1s, and NO successes at all.

* A roll can be called for with a minimum number of successes required, greatly reducing the chances for someone with a small pool. The number of successes can also be treated as the degree of success, making the expert capable of achievements the novice can't reach.

* The ST can require that the character has the requisite Ability (skill) in order to even attempt the roll, locking out the unskilled.

[PhoenixPhyre]

Very much this, I believe.

However, while I'm primarily in the permission slip camp, I can enjoy a degrees of success style game (which largely falls under the "better results" that you put in the capability improvement section) - especially when it more accurately models how I perceive reality working.

I can do either (despite preferring a capability improvement game because it allows everyone to attempt everything and stay involved), but I strongly prefer if the system is clear about which model it's using and relatively consistent. Clarity lets me feel like I'm playing as intended (something I care about strongly); consistency lets me predict and adjudicate fairly.

[Quertus]

Both of these were addressed to some degree in the oWoD system, but the changes seem to have been often ignored by the playerbase,

* The botch math was changed such that 1s could still offset successes, but the only way to actually botch was to gets one or more 1s, and NO successes at all.

True. That's arguably one of my favorite changes. But that doesn't change the experiences I had back before that "errata" came out.

* A roll can be called for with a minimum number of successes required, greatly reducing the chances for someone with a small pool. The number of successes can also be treated as the degree of success, making the expert capable of achievements the novice can't reach.

True, somewhat. For one, the probability of the expert achieving that level of success is quite rare - quite unlike how successful at surgery I'd expect an actual surgeon to be compared to, say, me. Second, a 1 medicine 8 int character has a significantly larger dice pool than a 2 int, 4 medicine expert.

* The ST can require that the character has the requisite Ability (skill) in order to even attempt the roll, locking out the unskilled.

True. And the fluff strongly implies this. But I've never seen it at an actual table. What are your experiences in this regard?

I can do either (despite preferring a capability improvement game because it allows everyone to attempt everything and stay involved), but I strongly prefer if the system is clear about which model it's using and relatively consistent. Clarity lets me feel like I'm playing as intended (something I care about strongly); consistency lets me predict and adjudicate fairly.

Good points. Something that's good at what it does, and clearly states what it does, is vastly superior to the alternative.

[NichG]

Would that include combat? And would those be just blanket permission (as in "with X ranks in Spot you can see invisible things") without consideration for the fiction?

The general rule of thumb I used was that ranks 1 through 5 in a skill correspond to things which are exceptional but still fit within the general meaning of things in the fiction, whereas ranks 6 and 7 allow literally impossible things (with rank 5 being the natural cap for characters, but with various means existing to raise cap on single skills under certain circumstances). Rank 6 effects tend to be 'supernatural', while Rank 7 effects tend to be 'conceptual'. So a Spot 4 character might for example cause Stealth-users to pay double cost to sneak past them and triple cost to take overt actions in their vicinity, but invisibility is still invisibility. However, Spot 6 character would be able to see things like magical fields, vibrations, air currents, etc, and could just see straight through invisibility. And a Spot 7 character might be able to literally look into someone's soul and see scenes from their past.

For combat, I tended to use a system where roll-offs determined the cost of defense (coming out of pools that characters could voluntarily tap to no-sell consequences), while the discrete abilities determined the kind of consequences you could levy. So a character with a high Command skill might be able to force a defeated enemy (one who has no more points left to defend with) to change sides and fight on the party's side for example. Weapon skills tended to give things sort of like D&D feats - Bow 5 granting a free attack on anyone entering your line of sight from outside of it, Bludgeon 4 causing your blows to Confuse targets who partially resist them, etc.

I'm seeing a philosophical split here:

Some systems use a permission slip model: pay your dues (build points, etc) and you can do X. Don't, and you can't. This allows you to infer backward--can do X => has Y "ability". These also have fewer "ranks", but each rank is more significant. Handles more difficult tasks by requiring higher ranks to do. No partial success possible(?)

Others use a capability improvement model: anyone can attempt X; people with more ranks have a better chance of doing X or can get better results when doing X. More granular--low "rank" people may be able to do harder tasks--but also more prone to calibration issues. Doesn't provide quite as clear a signal about capabilities (being less binary).

Is that an accurate (or at least not horribly inaccurate) grouping? I recognize that there are systems that share traits of both philosophies.

Yeah, at least that's the distinction I was aiming at. For the 'permission slip' model, I prefer ones in which 'everything a normal person could do' is free, and you're only paying for literally exceptional things which in a default game one wouldn't actually assume could be done just by trying - generally tied to some kind of mechanic which partially overrides the fiction. So for example, a couple mechanics I like for social skills are things like, 'you get to know how this NPC would react to an offer', 'you get to take back something you said (or something another PC said)', etc.

That way you avoid the feeling of having to pay for what you already have. Otherwise it can be quite oppressive.

In terms of partial success, I think the permission slip systems also work better when the thing in question is not success but rather everything is thought of as transforming the game state in some way. So it's not 'I try to convince this guy, do I succeed?' but rather 'do we have some way to get from the current state to a state in which he is convinced?'. The permission slip abilities are guaranteed transformations - ones where how the game state changes is spelled out in mechanics - but you can also have 'non-guaranteed' transformations by engaging directly with the fiction or context.

[PhoenixPhyre]

The general rule of thumb I used was that ranks 1 through 5 in a skill correspond to things which are exceptional but still fit within the general meaning of things in the fiction, whereas ranks 6 and 7 allow literally impossible things (with rank 5 being the natural cap for characters, but with various means existing to raise cap on single skills under certain circumstances). Rank 6 effects tend to be 'supernatural', while Rank 7 effects tend to be 'conceptual'. So a Spot 4 character might for example cause Stealth-users to pay double cost to sneak past them and triple cost to take overt actions in their vicinity, but invisibility is still invisibility. However, Spot 6 character would be able to see things like magical fields, vibrations, air currents, etc, and could just see straight through invisibility. And a Spot 7 character might be able to literally look into someone's soul and see scenes from their past.

For combat, I tended to use a system where roll-offs determined the cost of defense (coming out of pools that characters could voluntarily tap to no-sell consequences), while the discrete abilities determined the kind of consequences you could levy. So a character with a high Command skill might be able to force a defeated enemy (one who has no more points left to defend with) to change sides and fight on the party's side for example. Weapon skills tended to give things sort of like D&D feats - Bow 5 granting a free attack on anyone entering your line of sight from outside of it, Bludgeon 4 causing your blows to Confuse targets who partially resist them, etc.

Ah. I see. Thanks.

Yeah, at least that's the distinction I was aiming at. For the 'permission slip' model, I prefer ones in which 'everything a normal person could do' is free, and you're only paying for literally exceptional things which in a default game one wouldn't actually assume could be done just by trying - generally tied to some kind of mechanic which partially overrides the fiction. So for example, a couple mechanics I like for social skills are things like, 'you get to know how this NPC would react to an offer', 'you get to take back something you said (or something another PC said)', etc.

That way you avoid the feeling of having to pay for what you already have. Otherwise it can be quite oppressive.

That last sentence is one reason I am less fond of the permission slip models I've used/read--they seem to presume incompetence. That means that someone with a 0 rank is incapable of doing relatively ordinary things. If normal people have to have at least one rank in everyday things, that makes 1 the new 0, and makes 0 a trap.

I'd rather have a "skill" system that covers a much narrower range but with broad skills, with normal things just happening with no check needed. For a capability upgrade model, this works out to setting the +0 breakpoint at "a normal person can do this about 50% of the time under pressure." Scaling things to the average person (with explicit understanding that PCs tend to deviate strongly from average) lets everybody do something. The illiterate barbarian can hear oral histories about magic (but isn't nearly as good at it as a specialist), the adventuring wizard has been around enough to be ok (but not great) at normal climbs, the rogue might know about religion from hanging out in church (to steal the offering plate ), etc.

[Florian]

I love how Splittermond handles it.

Core is 2d10 + primary ATT + secondary ATT + Skill, either vs. TN or comparative roll.

Critical failure is always "double one". You have the choice to roll "standard", "safe" (1d10) or "risky" (4d10).

Everything is skill-based, from weapon skills, defensive skills, magic schools and tasks, so your "rolls" feature into everything, up to how a spells works/your shield block reduces damage based on your roll.

Skills, not levels, give you your feats. For every three ranks (out of a maximum of 12 ranks), you can pick a feat either from the generic list or from the feat-specified list, up to a maximum of 5 feats per skill.

[NichG]

Using costs rather than chances of success can help a lot with avoiding the incompetency trap. Numenera (where characters are almost expected to spend 'effort' on most rolls to boost then) can play more like 'if I'm good at something, it will cost me less of my pools to do it' but anyone could succeed at a single challenge if they were willing to drop most of their points. So skill is a measure of staying power rather than peak ability or consistency.

[PhoenixPhyre]

I love how Splittermond handles it.

Core is 2d10 + primary ATT + secondary ATT + Skill, either vs. TN or comparative roll.

Critical failure is always "double one". You have the choice to roll "standard", "safe" (1d10) or "risky" (4d10).

Everything is skill-based, from weapon skills, defensive skills, magic schools and tasks, so your "rolls" feature into everything, up to how a spells works/your shield block reduces damage based on your roll.

Skills, not levels, give you your feats. For every three ranks (out of a maximum of 12 ranks), you can pick a feat either from the generic list or from the feat-specified list, up to a maximum of 5 feats per skill.

How are the TN determined? Fixed, Static (by task), level dependent, etc? I'd assume they'd have to scale somewhat if your ATT and Skill numbers go up during a campaign.

[Pex]

I suppose I have to get in on this.

Name of system (including edition if that matters)

Pathfinder

Basic mechanic: Roll under (% or dX), roll vs TN (fixed? varying?), dice pool (counting successes), ??

Roll d20 add total skill modifier to be equal or greater than a target number.

How specific are the skills? very general (eg Fight) to very specific (Knowledge (specific region))

A bit specific. Each skill provides different target numbers to achieve various tasks within the skill. Knowledge skills are different categories.

Is combat done using the skill system? (a Shoot skill, for example)

No. A particular skill can help in combat, but combat uses its own system. Knowledge (Something) can let you know in character strengths, weakness, and abilities of a creature you are fighting if it falls into that Something category. Acrobatics can allow you to move freely about the battlefield without provoking attacks of opportunity from the enemy.

How do skills improve? With level (if such exists), by spending XP, ???

Characters get skill points to spend on skills to improve them. Maximum total points spend on a skill over all equals your current level. You get bonuses to the skill from ability score modifier, +3 bonus for being on your class list, feats if you choose, and the optional trait system which gives your character small quirks at character creation.

How are different difficulties distinguished? Bonuses to skills, changing TN, changing dice pools, ??

The harder the task the higher the target number you need to reach. Depending on skill instead of a target number you make an opposed roll. You roll a d20 and add total skill modifier. DM does the same for the NPC. Who ever is higher wins. Ties go to the active participant, the one doing the thing that forces the other to roll.

You can Take 10 when there are no distractions. That means presume you rolled a 10 on a d20. Add your modifier. That's your total. Target numbers you reach means you autosucceed. Combat is a distraction and thus you cannot Take 10. However, Rogues may choose as a high level class feature the ability to Take 10 on some skills during combat.

You can Take 20 when there are no distractions and nothing happens if you were to fail. This means treat the die roll as if you rolled a 20. However, the task takes at least 2 minutes to do, longer depending on task and DM adjudication.It also presumed you rolled a 1 before you got the 20. If a task has consequences upon failure you cannot Take 20. For example, if rolling a 1 means you fail to climb a cliff and thus fall, you can't Take 20. You are allowed to Take 10 when you can't Take 20 if Take 10 would let you succeed. Searching a room is a common Take 20 event. You find whatever it is in the room the party is meant to find if anything.

Can I use skills untrained?

Depends on the skill. Some skills require you spent at least one skill point into it, such as all Knowledge Skills. Other skills require no points spent.

What part do you think it handles best?

Because of Take 10 and Take 20 it helps players (including the DM) know when a character is just that good he autosucceeds. There doesn't always have to be a chance of failure. It saves time having to roll a die and move on with the game.

How frequently do you have to refer to tables (either as a player or as a GM)?

Sometimes. You couldn't possibly memorize every detail. You can leave the book open to skills section for easy look up. It takes a few seconds. It's not a bother to the people I play with. However, common events that happen often are eventually memorized no look up is needed. When it's an opposed roll no look up is needed. Some target numbers use formulas instead of specific numbers. Formulas used a lot are memorized. For example, to identify a spell being cast the target number is 15 + spell level using the skill Spellcraft.

[Floret]

The most convoluted (But still quite fun once you have grasped it) skill system i know is from The Dark Eye (4th and 5th edition).

Basically, each skill has both a skill rank, and three associated attributes (sometimes those 3 involve the same one twice); and you try to roll under in sequence. If you roll over, you pay skill points in height of the difference.

If you have any (or exactly 0) skill points left after compensating for each of the three attribute dice (if necessary), you succeed. How many points you have left determines how well you succeeded, should that be relevant. Skill points are only lost for that one skill check, every check starts you off with points in height of your skill rank.

Challenge modifiers take or give skill points (Though you can't keep more than your skill rank, 4th ed.) or treat your attributes (all three) as lower or higher for the purposes of rolling under (5th ed.). In 4th edition, the question of untrained rolls depends on the skill in question; in 5th, the answer is yes. Skill points cost way less than attribute points, in 5th edition 1-4 points per level vs. 15; in 4th... too complicated to remember.

Does that sound complicated? Because it is. It also accounts for both inherent ability and learned skill, manages to let untrained people succeed (with luck); while giving trained characters the ability for both more consistent and better success. It also makes the math of the system nigh-incomprehensible, though.

Combat skills are calculated differently and have a simple roll under (once) mechanic; checking tables isn't really a thing... Though in 5th edition can happen for spell effects. Spells are done with the same system, though never usable untrained (And with each spell counting as a different skill).

...My favourite, though, are pool systems, especially L5R 4th ed. With their unlocking of mastery abilities at certain skill ranks. Wish they had done that more consistently though. And I don't really like the fact that attribute points are just so, so much *better* than skill points, making taking more than 2 ranks in a skill a questionable decision unless you want those mastery skills. Here combat and noncombat skills are rolled the same way.

[Max_Killjoy]

I'd rather have a "skill" system that covers a much narrower range but with broad skills, with normal things just happening with no check needed. For a capability upgrade model, this works out to setting the +0 breakpoint at "a normal person can do this about 50% of the time under pressure." Scaling things to the average person (with explicit understanding that PCs tend to deviate strongly from average) lets everybody do something. The illiterate barbarian can hear oral histories about magic (but isn't nearly as good at it as a specialist), the adventuring wizard has been around enough to be ok (but not great) at normal climbs, the rogue might know about religion from hanging out in church (to steal the offering plate ), etc.

Something I forgot to mention about HERO's setup: depending on setting details, characters get free "everyman" Skills at 8-, which is presumed to cover basic competence in that skill and allow the "nah, no roll needed" everyday tasks that fall under those skills.

E: one of the things I like is that the scale is known, and the results are known. No additive die pools or exploding dice to create huge results, no TNs off the chart.

[Jama7301]

I liked skill groups in Shadowrun. A block of similar skills that I didn't have to buy individually was nice, even if advancing the block was more expensive, XP wise, than individual skills. It was nice to say "Yeah, my character has spent time with all sorts of weapons, but never really specialized".

[Florian]

How are the TN determined? Fixed, Static (by task), level dependent, etc? I'd assume they'd have to scale somewhat if your ATT and Skill numbers go up during a campaign.

That system acknowledges three tiers of play, standard, heroic and epic. Roughly speaking, that is translated for skills on a 1-6, 7-9, 10-12 scale, and attributes on 1-4, 5 and 6.
TN are fixed on task but take tier (and luck) into account when up- or downscaling results.

As a side-note, tier also handles how many buffs you can handle. Instead of breaking it down into type, like d20 did, you can handle 3,4 or 5, pick the ones with the highest bonus.

[Firest Kathon]

I really like the skill system in Das Schwarze Auge (The Dark Eye).

Name of system: Das Schwarze Auge, 4th edition
Basic mechanic: 3d20 roll under (% or dX)
How specific are the skills? moderate (e.g. Perception, Athletics, Spellcraft)
Is combat done using the skill system? Physical combat: no, magic: yes
How do skills improve? by spending XP. Actual cost depends on the skill (e.g. bodily skills such as swimming are more expensice than mental skills such as spellcraft)
How are different difficulties distinguished? Penalties on the skill roll
Can I use skills untrained? Some skills can be used untrained (e.g athletics), others must be trained (e.g. opening locks)
What part do you think it handles best? No dependency on a single die roll, i.e. less variance than single D20; and dependency on multiple attributes. Also raising skill ranks is handled quite well.
How frequently do you have to refer to tables (either as a player or as a GM)? Never

Spoiler: A more detailed explanation

Show

Your character has eight attrbutes with a value of 1-19 (average human: 9-10 in each). Each skill is assigned to three attributes, e.g. Athletics (Agility, Constitution, Strength) or Streetsmarts (Wisdom, Intelligence, Charisma). Usually these are three different attributes, but the same may be used twice, e.g. Spellcraft (Wisdom, Wisdom, Intelligence). Attributes may also be different depending on usage, e.g. finding tracks is Tracking (Wisdom, Intelligence, Intelligence) but following tracks is Tracking (Wisdom, Intelligence, Constitution), the GM may also adjust it on the fly for uncommon usages of skills.
You also have a rank in each (trained) skill.

When doing a skill check, you roll 3d20 in order, comparing each roll to the attribute value. If you are under the attribute value (or equal to it), all is good. If you are over your attribute value, you must use your skill ranks to compensate. If any skill ranks are left over at the end, you succeeded.
For example, Kosimo has Bravery 14, Intelligence 13, Agility 14. He wants to sneak into some building, using his sneaking skill (Bravery, Intelligence, Agility) of 6.
Example 1: The player rolls 6, 17, 15. 6 is for Bravery, as it is less than 14 this is no problem. 17 is for Intelligence, more than 14, so he has to deduct 4 from his ranks (2 left). 15 is for Agility, so one more point is lost. 1 is left over, so Kosimo barely manages to get in without drawing attention.
Example 2: The player rolls 6, 18, 16. Bravery is still OK, Intelligence costs 5 and Agility costs 2, bringing the result to -1. Kosimo fails to be silent and might alert some guards.

2x20 is always a failure, and 2x1 is always a success. Either case grants a "special experience" with that skill, which makes the next skill rank increase cheaper. "Special Experience" can also be gained as an adventure reward. Costs for raising skills rise depending on how many skill ranks you already have, so raising a skill from 1->2 may cost 5 xp, while raising a skill from 10->11 may cost 70XP.

The GM can modify the difficulty by assigning a penalty (or a bonus), which is subtracted from your ranks before compensating roll-overs. If your skill ranks are lower than the penalty, the remainder is subtracted from your abilities when comparing rolls.
Example 1: Skill ranks: 5, Penalty: -2: You only have 3 points to compensate your rolls.
Example 2: Ride (Charisma [12], Agility [13], Strength [8]), 2 Ranks, Penalty -3: After deducting from your ranks you still have a -1 penalty left over. Your effective scores for this check will be Charisma 11, Agility 12, Strength 7.

You can also specialize in a specific usage of a skill (e.g. Spellcraft [Demons]). If you do that, your ranks will considered 2 higher for that specific skill usage.

The new 5th addition mainly keeps the same system, but adds two more aspects:
1) if you have enough ranks in a skill, you do not need to roll but succeed automatically if no penalty is applied
2) The result is ranked in degrees of success, e.g. 0-2 points left over: normal success, 3-5 points left over: better success, etc.

The most convoluted (But still quite fun once you have grasped it) skill system i know is from The Dark Eye (4th and 5th edition).
[...]
Does that sound complicated? Because it is. It also accounts for both inherent ability and learned skill, manages to let untrained people succeed (with luck); while giving trained characters the ability for both more consistent and better success. It also makes the math of the system nigh-incomprehensible, though.
[...]

Damn, ninja'd. But what you describe is exactly the reason why I like this skill system.

[Beleriphon]

I'm a big fan of Mutant and Mastermind 2E and 3E system. Not-with-standing the cost of skills a character combines their associated ability score with the associated skill to get a total, which can't exceed a maximum value based on the power level of the game.

Since the rules generally allow a character trained in a skill (that is to say has purchased ranks in it) to use routine checks (assume a roll of 10 on a d20). So a sufficiently skilled character can pass most checks.

[2D8HP]

As a Gamemaster/"Keeper":

Name of system (including edition if that matters)

Call of C'thullu, RuneQuest, and other "BRP" games, 1e, 2e, and probably more.

Basic mechanic: Sometimes there's "mods" but mostly players roll under their skill percentage that's on their character record sheet

How specific are the skills? Very specific

Is combat done using the skill system? Yes.

How do skills improve? Success equals a chance to improve skill some.

Can I use skills untrained? Sometimes, a little, but not usually.

What part do you think it handles best? Felt more realistic than D&D, sometimes less work to GM

How frequently do you have to refer to tables (either as a player or as a GM)? Not much.

---

As a player:

Name of system (including edition if that matters) original Dungeons & Dragons;

Basic mechanic: I look the DM/referee in the eye tell 'em what my character tries to do and 'e decides my chances, sometimes I'd bother to first ask what my character guesses are the odds of success first.

What part do you think it handles best? I played a character not a sheet

[SimonMoon6]

I want to learn more about the wide variations in systems out there. Right now, I'm focusing on "skill" systems. Please restrict yourself to answering about systems you like so we can remain positive. Please refrain from bashing other systems .

Things I'd like to know:

Name of system (including edition if that matters)
Basic mechanic: Roll under (% or dX), roll vs TN (fixed? varying?), dice pool (counting successes), ??
How specific are the skills? very general (eg Fight) to very specific (Knowledge (specific region))
Is combat done using the skill system? (a Shoot skill, for example)
How do skills improve? With level (if such exists), by spending XP, ???
How are different difficulties distinguished? Bonuses to skills, changing TN, changing dice pools, ??
Can I use skills untrained?
What part do you think it handles best?
How frequently do you have to refer to tables (either as a player or as a GM)?

1. My favorite is the skill system of Mayfair's DC Heroes RPG. It's the same system from 1st to 3rd edition.

2. You have a skill rating which can be anything from 1 to, well, anything, but you don't often get much above 12. Many skills can default to your relevant ability score (though some skills are "trained only" meaning that you can't use them without actually having the skill); if you are using your ability score instead of the skill, you get a penalty of -2 "column shifts" (more on that later).

Every action involving dice in this game uses the same system: roll 2d10 that are "exploding" (rolling doubles means you get to roll again and add to the previous result, which can continue to happen as long as you roll doubles). A roll of double 1's is an automatic failure. You then compare your results to a table that tells you not only if you succeeded but how well you succeeded. For some skills (such as Weaponry), the difficulty is obvious (how hard is your opponent to hit? Look at his DEX) but for other skills, you may have to use a table that tells you the difficulty based on whether the task should be Easy, Difficult, etc, all the way up to something like Legendary difficulty. So, you can always easily eyeball the difficulty of any skill check.

3. Skills are very general (like Weaponry, Martial Arts, Artist, Charisma, Scientist, Gadgetry, Vehicles, etc), but they also have subskills (like Weaponry (melee), Vehicles (land), etc) that can be more specific. But if you want to be really good at something, you may have to buy a Scholar advantage that gives you a bonus on a specific topic within a skill.

4. Skills improve like everything else. You can spend Hero Points (experience points, sort of) to improve any part of your character, including skills.

5. Different difficulties are distinguished as mentioned above. You cross-reference your skill level with the difficulty of the task on the table to see what number you need to roll for even a minimal success.

6. Yes, you can use skills untrained, but not all of them.

7. It handles combat really well in that you don't have to guess if Batman can do a foot sweep or a cleave or anything else. If you're good at fighting, you're good at fighting period. You *can* specialize, but you're never gonna go, "Oops, I forgot to take the Disarm feat, I guess I can't disarm anyone."

8. You use one table (in two parts) for everything in the game. And that's pretty much it (not counting the "here's the generic difficulty for stuff" table).

[Mr Beer]

Name of system (including edition if that matters): GURPS4e

Basic mechanic: Roll under (% or dX), roll vs TN (fixed? varying?), dice pool (counting successes), ??: Roll under skill on 3d6 to succeed.

How specific are the skills? very general (eg Fight) to very specific (Knowledge (specific region)): Extremely specific. There's an option to "clump" a bunch of similar skills together e.g. "Guns!" covers every gun and gun-like weapon in one skill instead of separate ones.

Is combat done using the skill system? (a Shoot skill, for example): Yes

How do skills improve? With level (if such exists), by spending XP, ???: By spending XP.

How are different difficulties distinguished? Bonuses to skills, changing TN, changing dice pools, ??: Skill penalties.

Can I use skills untrained? Yes for any skill that you might reasonably succeed without relevant training at e.g. Swimming, Climbing and No for ones you won't e.g. Nuclear Engineering.

What part do you think it handles best? Passes the realism 'feel' test, avoids the swinginess of d20 and % based skill systems, is consistent, is comprehensive.

How frequently do you have to refer to tables (either as a player or as a GM)? Generally only for combat skills because Range penalties and Criticals are a thing.

[Knaight]

I've got two systems here, plus my half finished game that I'm very happy with the skill system for. In order:

Fudge: The skills are fundamentally roll and add, using 4dF+x (dF=d3-2). The reason I like this is that the skills being centered around 0 keep the difficulties and skills on the same scale, and it's small enough that every level could be named. The numbers fade to the background, and a Fudge sheet is immediately understandable without knowing the system (Swords 2 is meaningless, Great Swordsman says something about the character). Combat is generally done with the skill system, skills can be used untrained but specialized skills start at a very low level and will probably fail miserably. Most of the rest varies highly; Fudge is somewhere between toolkit and system and different implementations can be very different.

ORE: ORE's system is quirky, so I'll get that across first - you roll a pool of d10s, ranging in size from 2-10 and with options to make them behave in certain ways (setting a die before rolling, setting a die after rolling). Instead of counting successes though you look for matches. Any match is usually a success, the higher the number the better you do and the more dice in the match the faster you do it. This produces a very interesting success curve, and more than that it feeds beautifully into various subsystems, particularly combat (which uses the skill system). Difficulty can be handled in three ways, minimum number on the dice, minimum number of dice, or just flat reduction in number of dice you get, representing different types of difficulty. Skills have middling specificity, improve with XP, can be used untrained, and don't really involve tables much, though combat manuevers and magic can bring them in. It's the sheer versatality of this core mechanic, and the way small variations allow for so much so intuitively that makes ORE good, with specific ORE implementations also having other major upsides.

Legacy: This is my half finished homebrew, to be finished eventually. The skill system is a reverse dice pool system, where difficulty is determined in number of dice and skill is represented as how many of those dice can fail with you still succeeding. There's some slight variations to that as well (e.g. opposed checks involving accumulating dice until you just fail with more dice being better), and combat uses the opposed skill system. Skills are fairly specific, improve with experience with some restrictions (the system is built around the idea of a master and apprentice, with the members of this changing through the generations, and thus the masters' skills affect what the apprentice learns), can be used untrained but routinely fail even easy checks, with the way everything is always theoretically achievable for everyone but failure can be completely proofed against being what sets the system apart.

[Selene Sparks]

My favorite skill system belongs to Shadowrun 4th edition.
Basic mechanic: Dice pool, specifically skill+attribute+modifiers, counting hits.
How specific are the skills? Generally, the skills are reasonably specific, but not too much. For example, computer usage is split up into five skills, the Computers skill for general usage, such as editing documents or the like, Hacking, for, well, hacking, Cybercombat, for punching people in the brain through your computer, because this is a game based on 80s cyberpunk, Electronic Warfare, for encryption, and Data Search, for googling things, but you don't need to invest heavily in all of these to be a good computer user. Combat skills, however, are too specific, in that both firearms and melee combat are split up into three skills each with no relation to each other(IE, being crazy good with automatics has literally no impact on how you handle a semiautomatic weapon at all), but there are at least "skill groups," so you can buy up several related skills cheaper than you could raise them each individually, so the problem isn't too significant.
Is combat done using the skill system? Yep. Guns are a skill, stabbing is a skill, dodging is a skill(although unless you waste an action, you're on flat Reaction against firearms, rather than the Dodge+Reaction).
How do skills improve? You spend XP to buy up skills, either individually or in skill groups, as Shadowrun is an unleveled skill-based system.
How are different difficulties distinguished? By requiring more hits to succeed.
Can I use skills untrained? Yes you can on most skills, although not all. For example, you can't roll to develop or install cyberware in people unless you're trained in it. Additionally, you do suffer a one-die penalty on skills you default on.
What part do you think it handles best? Expertise. In Shadowrun, you can walk out of character generation extremely competent in your field, but this, due to the nature of the game, is unlikely to infringe on the rest of the party's fields. Additionally, the system is rather sturdy. You can push the numbers pretty darn far without the system falling apart. But, really, the system is extremely versatile and handles almost anything very well.
How frequently do you have to refer to tables (either as a player or as a GM)? Very rarely. Most skills that show up in play are opposed or based on a readily visible number, such a flat stat or rating of an item.

[Floret]

Damn, ninja'd. But what you describe is exactly the reason why I like this skill system.

Oh, you used the form, I didn't, so your post does add extra benefit since I skipped some. And yes, if it wasn't clear from my post, while the system looks a bit ridiculous (And if explaining your math requires charts and a trn-page essay, that's certainly something), I do like the skill system.

I do take issue with you describing 4.1 skills as "moderately specific", though. With physical stuff, maybe, but once you get to knowlege and craft skills, skills like Bowyer, woodworker, Instrument maker, cartwright and carpenter all being entirely separate skills; every language and script (That you never even roll on) being a separate skill, the existance of a "writing" skill distinct from those for being able to phrase things properly... It does get a bit out of hand in places.

(Abd, just to nitpick: level of success in 5th edition are 0-3/4-6/7-9/etc. Three points per group is correct, but the first level is off )