What's Your "Sweet Spot" for a Skill system?

[pointofyou]

And so, agreeing with @Thourne that a reason for many action declarations is to propel the narrative, we see that at least prima facie it is possible to declare an action without trying to avoid consequences. Does anything exclude this prima facie possibility?

Given that I have experienced, I very confidently infer that nothing does!

Seems to be so.

 

[Crimson Longinus]

I don't care who does or doesn't like things. I do take a modest degree of exception to being told my play is illogical and incoherent.

It is not illogical to design a dice-based game so that, if a player loses on their roll, the "game state" changes in a way contrary to what they were hoping for.

When the "game state" in question is essentially a shared fiction, which the player is engaging via their control of a particular character (let's call it a PC) then it it not illogical that the change that is contrary to what the player was hoping for be a new state of the fiction that sets back their PC's interests.

When the fiction is one of being lost in a swamp near a moathouse populated by bandits, it is not illogical that the new state of the fiction, which sets back the PC's interests, is one in which the PC is discovered by bandits and threatened with capture by them.

When the player's action declaration that triggered all this was "I preserve some of the meat of the giant frogs I killed", it is not illogical that the dice roll be structured by reference to the character's skill as a cook.

There is no illogicality or incoherence to be found. All there is is the fact that some people do not like a game in which consequences for failed rolls can be matters that fall outside a very narrow domain of the action the character was attempting, that generated the need for the roll. That domain is one of not only causal proximity (after all, if everything else is held equal then spending a long time cooking does increase the likelihood of being found by nearby bandits) but of "proximity of topic" - that is, the preference is one in which the consequence is about the same thing as the action declaration (in this example, the topic in question is the cooking of frog meat).

So are you now arguing that the test was to establish how long it took to cook the meat, and that the causal connection for the bandits arriving is that it took long? Would the same have occurred had the PCs took the same time to do something else? Perhaps something that did not involve a check. Furthermore, you earlier argued that the test was not to establish how well the character cooked, but like I have said already, if cooking fast is the goal then cooking slowly is cooking badly.

But I doubt this is what you mean, as I remember from our previous "documents in the safe" discussion that you don't care whether a causal connection exists.

In any case, the comment about lack of logic is not judgement about your playstyle, but about the specific resolution approach. It seems illogical to me to use values that represent character competence and difficulty of the task in the setting to establish the odds of failure unless that failure is related to the aforementioned things. Using them to establish the likelihood of causally unrelated thing is illogical.

And that the resolution framework might contain such lapse in logic, doesn't necessarily mean it is bad. For example here the goal is to have mechanism for introducing complications in a manner where their timing does not rely on GM judgement. And if one considers this a worthwhile trade, then the resolution framework is fine. But I think it is valuable to recognise that such a trade-off is happening. For example if one wanted to have causally logical resolution framework and disclaim GM decision making at the same time they would need to search for other approaches.

That preference for proximity of both causation and topic is moderately widespread among RPGers, albeit somewhat loosely applied (eg I think at least some of those objecting to the cooking example would not object to a failed roll to hit being narrated as a shield block by the opponent, although that does not satisfy the joint proximity requirement, and has basically the same structure as the cooking example - hence why really serious simulationist RPGing such as RuneQuest introduces a roll to block/parry). But it has not distinctive claim to logic or coherence.

It is not same thing at all. AC represents person's capability to defend themselves, so of course failing to hit it can be narrated as the defender blocking.

 

[Mustrum_Ridcully]

The cooking discussion feels kinda like one of those EN World endless discussion cycles that stopped adding meaningful to the discussion on page 2 of 20. Every relevant argument has already been exchanged, anyone without an opinion or idea on the topic has enough material to read and learn about the concept, and anyone that already had an opinion and hasn't changed it at page 2 isn't going to on page 20 either.
---

Back to the topic in the title.

My first roleplaying game was Shadowrun 3E. I remember making my first character, which was inspired thematically be Léon the Professional. Then I had to make some kind of Stealth roll and I realized I hadn't even trained that and didn't understand beforehand that this skill was kinda important - both for the character I had in mind, and for typical Shadowrun games.
Another thing I realized over time was that skills that you didn't max out were kinda useless.

Now, this was my first RPG experience, so I was really just a noob and overwhelmed by a complex rule system and a new paradigm of playing, so probably no system could adequately prepare me for this.

But, anyways, my lessons from this and later games.

I like a list of skills that isn't too long, with each skill covering sufficient ground that if you imagine your character good at certain things, you know which skills to pick and don't get any surprises that some aspect isn't covered by the skills you have. So I'd rather have Perception than Spot and Listen or Stealth than Hide and Move Silently.
What also help if there are purely descriptive "meta" groups. Like "these are social skills" "these are combat skills" "these are technical skills". They help navigating the system and make it easier to create characters as you envision them.

I don't want too much granularity in skill ratings, especially if low ratings don't do much to differ from being just clueless about a skill.

Things I learned later:
I kinda aesthetically prefer combat "skills" to be part of the skill system, and not seperated, but I can accept it either way. D&D since 3E at least for example uses the skill system for "non-fighting" skills, and proficiencies for fighting skills. You don't have a Swordmanship skill, you are proficient with Longswords.

I like characters to be able to rely on skills they are good at. They really shouldn't fail common tasks, within reason. A soldier will miss shots in a firefight, but if he's well trained, he won't miss a stationary target at target practice at lall, and it's really more about how many bull's eyes he's hitting than anything else. And an experienced climber isn't going to break his neck while trying to scale a 10 ft wall. (so definitely no fumble rules for me.)
And I learned from Gumshoe system that sometimes its best if a skill can't be failed at all. Investigative skills or skill uses should probably always reveal clues, and if there is any failure at all, it should be about the cost and risks, because not giving the players the clues is typically going to make things less interesting than more.

 

[Crimson Longinus]

The cooking discussion feels kinda like one of those EN World endless discussion cycles that stopped adding meaningful to the discussion on page 2 of 20. Every relevant argument has already been exchanged, anyone without an opinion or idea on the topic has enough material to read and learn about the concept, and anyone that already had an opinion and hasn't changed it at page 2 isn't going to on page 20 either.
---

You are not wrong.

But, anyways, my lessons from this and later games.

I like a list of skills that isn't too long, with each skill covering sufficient ground that if you imagine your character good at certain things, you know which skills to pick and don't get any surprises that some aspect isn't covered by the skills you have. So I'd rather have Perception than Spot and Listen or Stealth than Hide and Move Silently.
What also help if there are purely descriptive "meta" groups. Like "these are social skills" "these are combat skills" "these are technical skills". They help navigating the system and make it easier to create characters as you envision them.

Yeah, agreed. What skills do should be clear.

I don't want too much granularity in skill ratings, especially if low ratings don't do much to differ from being just clueless about a skill.

Fair. I always found Storyteller system's five point rating pretty sufficient.

Things I learned later:
I kinda aesthetically prefer combat "skills" to be part of the skill system, and not seperated, but I can accept it either way. D&D since 3E at least for example uses the skill system for "non-fighting" skills, and proficiencies for fighting skills. You don't have a Swordmanship skill, you are proficient with Longswords.

It always bugged me in the earlier D&D editions how combat stuff and non-combat stuff was handled differently. Nowadays it the same proficiency+ability mod for both, so it is symmetrical. Though it low key bothers me that there is expertise for non-combat proficiencies but not for combat ones. (I get why this is from balance perspective, but system aesthetically I'd like the symmetry.)

I like characters to be able to rely on skills they are good at. They really shouldn't fail common tasks, within reason. A soldier will miss shots in a firefight, but if he's well trained, he won't miss a stationary target at target practice at lall, and it's really more about how many bull's eyes he's hitting than anything else. And an experienced climber isn't going to break his neck while trying to scale a 10 ft wall. (so definitely no fumble rules for me.)

I feel this works better with systems where there is a degree of success and it actually matters. Because then you still get the excitement of the dice roll, even though you're mostly just rolling how well you succeed rather than whether you succeed at all.

 

[Committed Hero]

What skills do should be clear.

Agreed, although I am happy if skills overlap at various edge points.

To delve into the bandit example, having a cooking skill should mean that the act of cooking is important - possibly the designer meant the game should feel realistic, or wanted to emphasize resource management. Or it's tied into the creation of potions.

I am entirely OK with the example, because the consequences were suggested/approved by the player, and because the check led to something meaningful. When a player chooses to invest points in a particular ability (assuming it's not part of a template or another way that doesn't involve choosing), I take that to mean that they want to use it, and I am on the lookout for ways to make this happen.

 

[Celebrim]

I feel this works better with systems where there is a degree of success, and it actually matters.

I earlier defined the attributes I prefer to see in a skill system without making mention of "degree of success". That was an oversight in retrospect.

Personally, I'm agnostic on whether a system benefits from degree of success compared to simple pass fail, but I do prefer that if a system considers degree of success that they consider it as a separate issue in each type of skill check rather than having a unified system that assumes for all skill tests that degree of success exists and is meaningful. If you design "degree of success" into your system, I believe you are obligated for each and every possible skill test in your system to define what it means to succeed at every degree of success your system defines.

In other words, in my opinion, skills are not actually completely congruent and interchangeable. Many skills are actually orthogonal to each other and assuming a unified system for skills that is completely generic and abstract just does not work. Two attributes of a skill that exemplify this orthogonality are:

1) whether or not the success represents some quantity that can be measured ("How far did I jump?") or some measureless quantity ("Was I seen or not?")
2) Whether or not successes in the skill are linearly distributed or highly normalized. For example, "How much do you know?" or "Can you pass/fail hit a target" might be linearly distributed, but "Can you lift this object?" is highly normalized.

Systems that assume all skills can be abstracted away in the same manner - everything pass/fail or everything degree of success, everything dimensionless or everything measurable, or everything linear or everything normalized (implied by tying all skills to the same fortune mechanic) - IMO tend to be eventually problematic in play and require the GM to make ad hoc rulings in order to get around the limitations of the skill system and the nonsense that applying it without judgment calls would regularly throw up.

An example of this is that while I in general love 3e D&D skill system, it's inclusion of "Jump" as a skill is in my opinion a mistake. Because D20 is inherently a pass/fail linear fortune mechanic skill system, and "Jumping" as a skill is orthogonal to both design decisions in that it's inherently about a measurable quantity and inherently highly normalized. In this manner, 1e D&D which just defined absolute distances that you could jump is actually a stronger design.

 

[Mustrum_Ridcully]

One thing that occured to me also: Sometimes you kinda want skills to cover certain things, but not all skills are really equally important. (Sometimes based on the game itself, sometimes based on the specific campaign).

It would be nice if the game has a way to highlight such skills, possibly making the less important ones cheaper or them being put in a seperate "silo", to be paid with a different character creation and advancement "currency" so that being good at a skill of less importance doesn't mean you accidentally suck at something everyone is expected to participate in.

I mentioned that combat skills in D&D tend to not work like "normal" skills, and that can be an example - because everyone needs an attack bonus and an armor class since combats are expected to happen often or at least are fairly important for most campaigns.

In a subterfuge and infiltration game or campaign, lip reading or deciphering encoded scripts might be a much more potent skill, while armorcrafting is of minor importance. In some other game (or camapgin), armor crafting might be crucial because players are expected to build custom-made armor for their party and can't just randomly find or purchase them, while deciphering encoded scripts comes up never.

 

[Celebrim]

One thing that occured to me also: Sometimes you kinda want skills to cover certain things, but not all skills are really equally important. (Sometimes based on the game itself, sometimes based on the specific campaign).

It would be nice if the game has a way to highlight such skills, possibly making the less important ones cheaper or them being put in a seperate "silo", to be paid with a different character creation and advancement "currency" so that being good at a skill of less importance doesn't mean you accidentally suck at something everyone is expected to participate in.

Oh that's a good point as well. When I outlined my main considerations for a skill system one of them was that I wanted the system to be "space spanning" - which I defined as you did as "whatever the players attempt should have some sort of skill that covers it". But merely being discrete, space spanning, and independent isn't enough to make a skill system good and precisely one of the ways that space spanning can go wrong is if the slice of the skill space were highly inequitable so that a skill that covers a vast range of common tasks is weighted the same as a skill that covers only a rare and narrow specialty. I can think of examples of that in so many published rules sets, and it's a horrible bit of dysfunctionality that just becomes a trap.

A good example would be "photography" in the Call of Cthulhu game whose use almost would never come up in any game and yet the rules treat it as if it were a skill of equal rank to everything else - even making it a common professional skill. And of course there are a lot of other examples in BRP as well. GURPS - a system whose skill design I have mocked elsewhere - is filled with that sort of crap. It shows up in the WEG Star Wars system as the fact that every single thing you might want to fix has its own separate repair skill. It even shows up in "skillless" systems like 1e AD&D in the form of things like a dwarf's percentage chance of detecting sloping passages taking up space on a character sheet but basically doing nothing.

So yeah, ideally when you start cutting up the space of tasks into separate skills that cover those tasks you try to make all the pieces of the skill pie somewhat equal.

 

[Crimson Longinus]

One thing I consider paramount in a skill system is that you can easily compare one roll against several different difficulties. Meaning that if you have a room with three things worth noticing and one is really easy to notice and the second is kinda hard to notice and the third is super difficult, we can just make one awareness roll to see which of them the character notices. The original Storyteller system where the difficulty was expressed as the target number for the dice in the pool was bad with this, whilst the later version where the target number was fixed and the difficulty was expresses as the number of successes you needed didn't have this issue. And of course basic skill+die to beat a target number handles this easily as well.

 

[Pedantic]

So yeah, ideally when you start cutting up the space of tasks into separate skills that cover those tasks you try to make all the pieces of the skill pie somewhat equal.

Ideally, this concern should apply generally to character building resources. Costs should be weighted by utility, and/or options players can select should have relatively equal impact. The most obvious place this goes wrong is attributes. There should be a compelling argument for a player to want average scores in all of them.