I think it's very difficult to implement a system featuring both a DC (target number) and a level (unt/com/exp/mas) unless they have distinct and well-defined meanings. Even then, as a DM, it adds a further dimension to have to consider in designing/throwing out a skill check. The only clear separation I can come up with is innate aptitude vs. learned aptitude, which works for example with Climb (being strong and dextrous will make you a natural talent, but you can't get up a mountain without training). Setting task difficulties based on this distinction is still quite a challenge, if I had to advise on how to do it I would recommend considering breadth and depth. The breadth of a skill is akin to how many things need to work out for you to succeed (and thus partial success can exist), whereas the depth of a skill is how hard it is for any one thing to work out.
Going along with this, I think it might be easier to understand your system if you described it differently. Rather than saying "it takes two successes to get an Expert result, but Experts only need one success," you might explain it similarly to how you did your Master scholar making his checks, which really illuminated the system for me. "Competent ranking lets you roll one die for Competent outcomes. Expert ranking lets you roll one die for Expert outcomes and two dice for Competent outcomes. Master ranking lets you roll one die for Master outcomes, two dice for Expert outcomes, and three dice for Competent outcomes. Any single bonus (through one time increment, Aid Another, magic item use, etc.) gains you one additional die and access to the next tier up." (Like your apprentice taking extra time to roll one die for the Master outcome and one for the Expert outcome). You know, maybe it's just me. But that makes more sense somehow...
I'm in agreement with both of you here, rules without a clean presentation (and unifying concept) don't reach their potential. Besides just the presentation of the mechanics, there is the implied or explicit presentation of the concepts behind those mechanics. After looking at your ideas, a thought that crossed my mind is that two different presentations might make the most sense, one geared for DMs and the other for players. Most RPGs make that distinction already, of course, but it's not something I usually think about when I'm wearing my game design hat. In game design my thoughts are usually dominated by "How can I make this sytem work?" with relatively little thought to "How can I enable people to make this system work?" and "How can I make people want to make this system work?" That perhaps I should intentionally pursue them all as integral parts of design feels like a small epiphany to me.
The "resolution-oriented" perspective I described in the last post or CN's "breadth/depth" distinction both seem to me more suited for the DM side of things, while KZ's description seems to me to be much better for a player or as a tutorial. As a DM the thing I probably most need to understand about the skill system is how to construct a scenario as a skill check, resolve it, and roll with the punches. For the player the need is usually more functional. That is, knowing how to resolve checks, how training affects a character, and how to roughly judge their abilities in relation to a proposed action.
So here's the tension. I've been describing the system primarily in terms of multiple successes where the single-success effect of training is essentially an exception to the rule. Although this emphasizes the system's flexibility, it obscures continuity with historical D&D and risks making single-success play seem like a second-class citizen (despite my goals). Whatever its merits, a D&D mechanic that encourages that perception invites all the baggage associated with other mechanical upheavals. On the other hand, emphasizing the single-die style of play and presenting multiple successes as more like an exception risks making multiple successes seem like an afterthought, something to be forgotten or ignored until it is absolutely necessary to do otherwise.
So, how can we conceptualize the system, at least in a perspective useful to the DM? The definition of "breadth" as the number of parts needed to succeed (i.e. number of successes) and "depth" as the difficulty (DC) for those parts is interesting. I actually found myself gravitating to the opposite labels for almost the same concepts. In fact, it was this difference which, after confusing me for a while, lead me to realize how deeply (and implicitly) I had adopted a results-oriented approach. That is, when I think of depth on a skill I tend not to focus on the "depth" of the character or the process of the task, but on the depth (detail) of the result. Likewise, for "breadth" I think of the different ways someone could achieve the result (using a different skill to aid, for example). There is some commonality between both conceptions of breadth because both relate to the number of skills, but I don't necessarily associate that with the number of successes. I don't bring this up to quibble about terminology because I'm sure we could figure out a name we'd both find acceptable for these concepts, but this little difference in understanding highlights the unexpected issues that can crop up. I do think, however, the reason I came to have my results-oriented focus might shed some light on not just the terminology, but why I approached the whole skill system the way I did.
While previously designing a homebrew success-based system the other designer and I really struggled with what DC, skill bonus, and number of dice thrown "meant" in a simulationist sense, and especially how to represent modifiers. (For example, should "combat advantage" be like a bonus to a skill, a bonus to an ability score [determined number of dice thrown], or a reduction in the action's inherent difficulty [DC].) After a good number of weeks trying and failing to come up with a coherent game-wide interpretation that made sense, and in fact seeing several areas where conflict was actually unavoidable, we decided to stop thinking about the simulationist meaning. We focused instead on how each of those things affected the results of a check, i.e. what they actually do for the math. (In that game the primary effects are these: a larger skill bonus makes the first success more likely, a reduction in DC makes all successes more likely, and changing the number of dice changes the maximum potential number of successes.) Almost overnight the game ran more smoothly, in part because we had consensus on how the game should work, but also because the DM had a tool which addressed his main concern: how should some modifier affect what is actually likely to happen? There was still the notion of base DC as being somehow "inherent" to the difficulty of the task, for example, but adjustments to the DC were made whenever the DM thought all successes should become more or less likely to occur, regardless of the "source" of the adjustment in the game world.
That same sort of logic is what I'm aiming to achieve for DMs here as well, but a really coherent explication of the idea obviously still eludes me. The "determine appropriate training for the minimal result, set DC for a person of equivalent training" rule-of-thumb is succinct, and for me feels natural after being immersed in a success-based system. Even so, I had to give the Track example in the last post quite a bit of thought before I came up with a version I really liked. I'm still thinking about this issue.
So now you know the method to my madness.
Let's return now to the Decipher Script check you presented.
Returning to your initial example, Deciphering the entire scroll is a broad task - you've got to get the language, history and arcana all correct. Translating it is narrow, you just need the language. The depth depends on how obscure the text and references are.
The "distinct" skills to use for any given check is highly dependent on the skill list for the game, or whether there even is a skill list. In a game where Language, History, and Arcana are all distinct aptitudes your example works great. In a game, however, where a catch-all "Decipher Script" skill is designed to cover all those aspects we are left with a single check again, and there isn't really a way to represent a more involved situation without putting in multiple Decipher Script checks, which more-or-less returns us to something more like my original proposal. And leaving it as a single die roll loses the mathematical richness I think we're both trying to achieve.
In addition, I think a very important aspect of most skill checks is that they involve only a single skill. Compound checks (i.e. separate checks related to the same basic goal) have always been available for DMs, and indeed the Skill Challenge is a structured way of making compound checks. (In a loose sense the entire game is one big compound check.) I like having that separation because it keeps things focused on the specifics of what a given skill is accomplishing, and its scope is always the same as the scope of the skill itself. If multiple skills are considered a default part of skill checks, I think the very definition of what counts as a single check becomes very nebulous, especially if there are game mechanics that interact with checks. Consider, for example, a spell that grants an extra die to roll on every skill check for a few minutes. A DM that construed checks narrowly might give a character under that spell the bonus 5 times, while a DM that construed them broadly might give it only twice, even though in both cases the character is performing the exact same thing. (The Aid Another examples I gave where a different skill is used to grant an extra die doesn't break this structure, in my opinion, because the Aid Another part makes it clear the additional skill is subordinate to the main purpose of the check.)
As you noted, breadth/depth doesn't fit that well with the mechanics I proposed. Looking for alternate mechanics that would work well with them is a good idea, so let's take a look at the alternative you suggested.
Chris_Nightwing said:
Ok, now I would tweak your system a little to make this work. The DC (or depth) of the check is a target for your innate ability, and so your d20 rolls should be modified by the relevant ability score (you can change it according to the task..) and by equipment/magic that makes it possible for you to do more difficult things (or achieve narrow, 1 success, tasks easier). Now for the levels of expertise:
Untrained - You roll 1 die and cannot aid another
Trained - You roll 2 dice, aid another 1
Expert - You roll 3 dice, aid another 2
Master - You roll 4 dice, aid another 3
(You can only give 1 dice to the person you are aiding, however many you roll)
The breadth of a skill check needn't be limited to these levels, just decide how many successes are needed (even 5 is reasonable, a Master must take longer or get help). Don't make it require fewer successes for a Master to complete an easy check, that makes each advance too great and breaks the 'breadth' argument I've tried to fit the system with. Excess successes can be used for skill tricks (or indeed, perhaps you use some as tricks if you're not going to make the check).
I'll try to consider this proposal on its own terms, without biasing myself with the more general reservations I have about breadth/depth. The main thing I like is that every level of training behaves the same way, so the number of dice rolled depends only on the level of training, and the number of successes required depends only on the "breadth" of the goal. That is simple to describe and consistent, qualities I admire in mechanics. I also like that broad ranges of possible results are truly embraced, and how the better-trained can more effectively utilize aid another. With respect to a d20 system, I like that there is a bit more freedom (compared to my proposal) to play around with the bonus modifiers or the DC because training has no effect on the skill bonus. Finally, I like that there is no ambiguity as to whether modifiers should be included by changing the bonuses/DC or the number of successes. In that respect you absolutely met your goal.
I don't like that a check with a single skill always means requiring a minimum of exactly one success (if I've understood your tweak correctly), because that means all checks involving a single skill can never have a probability of success below 5% unless it's exactly 0%. On broad checks it's not clear to me how various levels of training in the different skills interact. To take the Decipher Script example again, does a Master of Arcana, Expert in History, and Trainee in Translation roll 9 dice and need 3 successes? And if so, does it matter if all three successes came from, for example, the dice granted for Arcana? If not, then Mastery at one thing in a broad check may make weak training in another part effectively irrelevant.
I also don't like that estimating (or setting) the probability of success becomes difficult on essentially all broad checks compared to the standard D&D check. I'll grant that estimating the expected number of successes (=n*p) is dead simple, and it is a standard tool in many success-based systems, but that kind of thinking is foreign to D&D, where the probability of succeeding on the d20 roll is king. Here's the comparison:
Mathematically the probability of getting at least k successes on n rolls with probability p per roll is a simple application of the binomial distribution:
For example, taking something like the decipher script example again, the probability of getting at least 3 success on 9 rolls with probability p per roll 84p^3-378p^4+756p^5-840p^6+540p^7-189p^8+28p^9.
At the table I'd have no idea how to ballpark that unless p is fairly close to 0 or 1, and it doesn't help that p might be different because several different skills are involved. Using the expected number of successes can help, of course. Roughly half of all outcomes occur above and below the expected number of successes, so given both n and p one can easily guess if the probability of getting at least k successes is better than or worse than 50%. But if one wants to set up a situation where the PCs are 80% likely to succeed, knowing the expected value is less useful. For example, in the decipher script example with 9 dice, if p=.5 then the expected number of successes is 4.5 and the players clearly have a better than 50% chance of success. How much better isn't obvious. In fact, it is just over 90%. As this shows, trying to put some of the advantages of success-based systems into D&D while hewing closely to that standard d20 system feel is not trivial.
A less major concern I have is that your tweak might involve too many d20 rolls, principally because even the lowest level of training grants an extra die, and everyone will have that level for several skills. The larger number of dice for higher levels of training is probably not so bad since those levels should be relatively rare. (Subjectively I find 4d20 to be too much for a default roll, mathematically and at the table, but since I advocated 3d20 for Masters it would stretch credibility for me to pretend that is some vast difference.) Nevertheless, if my proposal already strains what counts as a "tolerable level" of multiple success influence in D&D, this tweak stretches it further.
I think there is room for doing interesting things with the breadth/depth concept, but I think it's missing a compelling reason to "package" all the skills together the way it does. The Wushu RPG, for example, adds a die for every detail added to the description of an action, and something like that could be used to package the breadth/depth check in a way which makes it clearly distinct from a string of multiple checks with single skills. In a hypothetical stunting system where each element of the stunt must succeed for the whole check to succeed (or something close to that), adding new elements might lead to greater potential rewards, but also increase risk because adding only 1 die to a check makes it less and less likely that all the dice will succeed. Certain kinds of stunt elements might allow the player to add more than a single die, representing great skill with that particular element. Thus players could build their own risk/reward structure while describing their actions.
Going along with this, I think it might be easier to understand your system if you described it differently. Rather than saying "it takes two successes to get an Expert result, but Experts only need one success," you might explain it similarly to how you did your Master scholar making his checks, which really illuminated the system for me. "Competent ranking lets you roll one die for Competent outcomes. Expert ranking lets you roll one die for Expert outcomes and two dice for Competent outcomes. Master ranking lets you roll one die for Master outcomes, two dice for Expert outcomes, and three dice for Competent outcomes. Any single bonus (through one time increment, Aid Another, magic item use, etc.) gains you one additional die and access to the next tier up." (Like your apprentice taking extra time to roll one die for the Master outcome and one for the Expert outcome). You know, maybe it's just me. But that makes more sense somehow...
As I said at the beginning, I think you've got a really good idea here, especially for explaining the system to players. Following your example, maybe an outline of what the PHB could say is something like this:
A skill is an ability which a creature can use to perform specific kinds of tasks. Each skill a creature uses has a ranking that measures the degree of training in that skill: Untrained (the default), Competent, Expert, or Master. Additional training usually increases the odds of successfully using a skill, and in many cases provides more powerful results as well.
A skill check determines whether using a skill succeeds or fails. Skill checks may have different degrees of success, corresponding to one or more levels of training. The components of a skill check are one or more rolls of the form d20 + skill bonus vs. DC. (Unless otherwise specified every skill check uses only a single d20). The number of rolls that equal or exceed the DC is the number of successes on the skill check. On one or more successes the result of a skill check is all outcomes with ratings less than or equal to the user's training in the skill. Additional successes may allow one to use skill tricks for special effects, which are covered in each skill's description.
The benefits of training are listed below. The benefits are not cumulative.
Code:
Training Benefits
-------- -------
Untrained Access to universal skill tricks.
Competent +5 bonus to checks.
Access to universal skill tricks.
Expert +5 bonus to checks.
Roll one additional d20 for Untrained and Competent outcomes.
Access to expert and universal skill tricks.
Master +5 bonus to checks.
Roll one additional d20 for Expert outcomes.
Roll two additional d20s for Untrained and Competent outcomes.
Access to expert, master, and universal skill tricks.
Skill checks with multiple outcomes may sometimes require a trained user to roll different numbers of dice for the different outcomes. In these cases do not roll the additional dice until necessary. For example, a Master Acrobat who attempts a check that has outcomes for Competent, Expert, and Master should first roll a single d20 to determine if the Master outcome is obtained. If it is successful no further rolling is necessary. If not, the master rolls a second d20 to see whether the Expert outcome is obtained instead, and so on. If desired a creature may forfeit a higher outcome to gain a lower one instead. For example, if the master acrobat succeeded on the first check, but then decided that using a specific skill trick is more important than obtaining the Master outcome, the master may choose to accept the Expert outcome instead and roll an additional die to try and power the skill trick.
Some skill checks have outcomes that normally require more training than the user of a skill may actually possess. Nevertheless, a user with at least Competent training can, if it rolls enough dice, obtain outcomes above its own training by getting multiple successes on the check. In such cases two successes are required to obtain Expert outcomes and three successes to obtain Master outcomes. At the DM's discretion even more successes may be required for truly incredible outcomes for which training alone is always insufficient. Various ways to roll more dice on a skill check are described below...
I think that reads a lot better than the initial version.
Thank you both again for your insightful comments! Working through all these things in detail has been a lot of fun.
But for someone who didn't like that, it would be simple enough to say that a trained person using multiple dice for a series of rolls can only get a complication on the first roll, same as someone with less dice.
)
