Partial success in d20 systems

edited March 2013 in Play Advice
How would you go about implementing indie-style fail/partial success/success resolution into a d20 system (D&D 3.5, Pathfinder, etc). Looking for quick, simple, elegant house-rule type solutions, requiring little-to-no modification of other rules and stats.

A target range rather than a target number (in the range partial success, beat the range success), maybe? How to determine the appropriate range? Should the partial success range be below the DC/AC as written, or above it?

This is mostly just out of curiosity, but I am getting a bit sick of binary results in my Pathfinder game. I want some compromised successes and hard bargains!

Comments

  • Rolling a second die that would qualify the yes or no d20 result with an and or but?

    Or, even rolls are ands, odd rolls are buts?
  • edited March 2013
    Allow the player to declare post-roll, "yes, but". They can then add +2 to their existing roll and turn a possible failure into a success with consequences. This way clear success and failure are still on the table but you add "you almost got away with it" to the die roll.

    I don't know how powerful criticals are in Pathfinder, so I don't know if you would want to allow this to push a normal success to a 'tainted' critical.
    --
    TAZ

    // Then again, isn't this what Gandalf did to the Balrog? :-)
  • Adding a way that the player can take on personal or narrative costs in exchange for a success. So, a player could, in exchange for doing something dramatic or story serving, gain a +1d6 to a roll they just made or something. These could be based on the PC's personality or given by the GM or other players to encourage dramatic play. Success/failure ranges are baked into a lot of d20 products (like the Craft games and Mutants and Masterminds), so you could always check those out for inspiration.
  • Because there is no bell curve, the success/failure result is less tied to a number range as it is a percentage of the d20 yes? So I like Zircher's post roll choice. I also like the idea that the partial success is baked in, but not so much as having to adapt the TN to fit. Hmmmm. What if the player could decide to take a token on a success to endure a complication, ugly choice or hard bargain (partial success) that is suitable to the relative difficulty of the test. Then on a future failure (their's or another player's) , they could spend a token to turn a failure into a partial success?
  • edited March 2013
    I am getting a bit sick of binary results in my Pathfinder game. I want some compromised successes and hard bargains!
    I think that it could be perceived as unfair if you converted a full success into a partial so zircher's suggestion above is really good. I'd be tempted to expand the +2 to +5, after all that is still just a 25% chance of complications ie the interesting stuff!
    IMO a bigger, related problem under D20 etc is when failure = "nothing happens".

    Interestingly all of this is baked into the d20 roll by Matt Finch in his quick primer for OSR gaming where he suggests that the DM add in complications/consequences for high/low rolls in combat to keep things interesting. Look what he says:
    It’s also your job to inject events from outside the rules during combat. “You rolled a 1.
    Your sword goes flying.” “You rolled a 1. You trip and fall.” “You rolled a 1. Your
    sword sticks into a crack in the floor.” “Hey, you rolled a 20. You spin around and gain
    an extra attack.” Hey, you rolled a 20. You slay the orc, kick his body off your sword,
    and blood spatters into the eyes of one of the orcs behind him. He’s not getting an attack
    this round.” “Hey, you rolled a 20. You knock his sword out of his hand even though
    you didn’t do enough damage to kill him.” That’s just a set of examples for the various
    ways you could handle natural rolls of 1 or 20. Each result is different, and none of them
    were official – you just made them up out of nowhere. You’re being consistent – the
    high and low rolls always generate a good or bad result – but exactly what happens is
    pretty much a matter of you deciding what seems realistic, or really fun.

    Also, flavorful combat isn’t just in the naturally high and low rolls. A character leaps
    onto a table, but the table breaks. Swinging into combat on a rope succeeds – but the
    rope breaks and the character ends up swinging into the wrong group of monsters. A hit
    by a monster causes one of the characters to drop a torch. The feathered plume on
    someone’s helmet is chopped off by a missed stroke. All these little details add to the
    quality of old-style combat, and change it dramatically from a sequence of d20 rolls into
    something far more alive and exciting. This doesn’t mean, of course, that every swing of
    a sword blade and every step into combat must generate lavish descriptions and details
    from you. It’s a matter of pacing, and frankly I can’t explain how to do it well other than
    to say you’ll get the hang of it.
    So, this is kinda what AW etc are doing, in a more explicit way.

  • There was a thing kind of like this that went with E6. It was called "Raising the Stakes."
  • I'm throwing out some quick solutions.

    1. Grade the d20. 1-5 = bad failure (if failed roll); 15-20 = good success (if successful roll).

    2. Decrease the difficulty level by 5. If the roll is five or below the difficulty level, it's a bad failure. If it's five above, it's a good success.

    3. Roll an extra d20. If both dice are below the difficulty level, bad failure. If both dice are successful, good success.

    Everything else in all three suggestions above are partial failure (No, but) and partial success (Yes, but). Bad failure and good success are "No, and" and "Yes, and".

    4. All failed rolls creates a new obstacle instead. Examples: you have lock picked the door, but there's a guard behind the door. You knock out the guard, but it's too late because the guard had time to alert other guards.
  • edited March 2013
    Mouse Guard has a pretty simple solution: A failed roll = GM's choice between failure with a twist or success at a cost. It's all three results but still a binary roll.
  • Cool, thoughtful suggestions all around. It's really the "Yes, but" compromised success that I'm missing - as indicates above, critical hits and botches work well as "Yes, and" and "No, and."

    It would be interesting to apply MG's fail+twist/success+cost to d20 - I feel like I would more frequently choose the twist though (that's prett much what I do now), so I'd like to have a clearly defined situation in which a partial success occurs.

    I really like the idea that the player can choose to "risk" a compromised success. Maybe the player gets to choose +2 or +5, for lesser or greater potential risks (softer/harder moves, basically).

    I'll have to check out the E6 rules, too.





  • The simple fix we did in high school was to eliminate cold failure. Either you succeed, or you got a complication to work through. Only a "botch" led to something bad. Target number was 17. Roll a 15? "You get this close... you can hear the first few tumblers click in to place, but the guard is rounding the corner."

    It's not ideal, but it helped a bit. We needed to break out of a bad habit we had, which was "even if you miss the TN by 1 you are still a bid idiot and your character does something stupid and embarassing."
  • You could have a chart of ten different qualifiers (like "a condition is attached", "something surprising comes to light", "the situation gets more complicated", etc.), and then use the ones digit of the die to pick one of the elements from the chart, if the roll is a failure.
  • edited March 2013
    Talislanta uses the action table (subtract DC from skill level, roll with resulting modifier, consult table):

    0-: Mishap
    1-5: Failure
    6-10: Partial Success
    11-19: Success
    20+: Critical success

    But I'd probably use the Mouse Guard method for minimal interference with the rest of the system.
  • Thanks, folks! Again, the issue I have with the MG approach (this is just a personal GM hurdle) is that I default to fail+twist on a normal failed d20 roll - arbitrarily deciding that a failed roll is a success+cost just doesn't sit well, and it's that "yes, but" that I feel is missing from my current style.

    E6's raising the stakes is cool, but a bit too cumbersome for my taste, as it requires a player-GM negotiation that predetermined two possible outcomes before the stunt. I'd prefer to shoot from the hip. I think I'm going to try a version of zircher's player-activated method. Will report back.
  • Thanks, folks! Again, the issue I have with the MG approach (this is just a personal GM hurdle) is that I default to fail+twist on a normal failed d20 roll - arbitrarily deciding that a failed roll is a success+cost just doesn't sit well, and it's that "yes, but" that I feel is missing from my current style.
    There must be some times when you don't have a good twist, right? What do you do now in that situation. Plus, in d20 success+cost is easy: they have hit points!
  • I rather like the approach where you roll 2d20. If both dice meet or beat the target number, it's a Full Success. If one die meets or or beats the target number, it's a Partial Success. If both dice roll below the target number, it's a Failure.
  • Don't GMs do this normally? I mean, even when running Rifts, if you miss the result by 15 points, I tend to describe a different outcome than if you miss it by 1.
  • Don't GMs do this normally?
    I don't think it's an intuitive and self-evident technique, no.

  • Really? If you miss a roll by 1, it's still a total failure? And not, say, you barely missed in the fiction as well?
  • edited March 2013
    I've certainly used that technique. When I was kid, it was all or nothing - a miss by one is still a miss. Just sayin' it's not something that's a GM standard (if such things might be considered to exist).

    Particularly when most game texts from the trad/90's basically said things like "If you roll equal to or above the TN, it's a success. If you roll below the TN, it's a failure."
  • I don't use partial success in D&D, but I do use degrees of success, which is ultimately the same thing as long as you do task resolution (and not stakes-ful conflict resolution). My system is simple: every five points over the target number is an extra degree of success, every five points under the target number is an extra degree of failure. A natural '1' is an extra degree of failure regardless of the result (so you could still succeed if your bonuses are high enough and the task easy, but you also get a single degree of failure - sort of a mixed success outcome), a natural '20' explodes (usually providing you a relatively large degree of success).

    (The obvious variant is that if you don't like wildly swinging degrees of success, then make a '20' just give an extra degree of success in addition to the calculated outcome - the opposite of a natural '1'. It's still a lot of success, unless you were making a desperation play of some sort.)

    As for what the degrees of failure and success do, they work as a currency, sort of like in The Mountain Witch - you can buy all sorts of things with them. The first success or failure always goes for establishing the task result that was under consideration, but either the GM or the player can allocate the rest. For example, in combat this principle turns into "stunting", where a player declares additional facts of positioning or status attacks (disarming, tripping, pushing opponents) on the strength of their "extra" success. In information gathering the GM might declare that OK, you get one question per degree of success, choose them well.

    My house rules also organically develop special package deals. For example, a fourth-degree success in combat is usually a "critical hit" that fells ordinary foes in one blow regardless of hitpoints. The player could also request to split that into four lesser effects, which is sometimes important tactically, if they need to e.g. defeat the foe and do something else as well at the same time.

    When there are both success and failure involved (due to rolling a fumble, or for some more exotic reason), the currency is not neutralized if at all possible, but rather the success will be mixed with lateral failure modes. You get what you wanted, but you also get an extra complication you have to deal with now, or whatever. Resisted checks of various sorts might be resolved in this manner: give both sides as much success as their degree of success indicates, and once they've gained all the success they can lateral of each other (that is, forms of success that are not also failures of the opponent), then deduct the rest of the success from each other and award the remainder to the winner (the one with the more degrees of success) as a victory, and the opponent as a defeat.

    What makes this a "partial success system" is the task-resolution logic: because an individual conflict can be resolved by one or more tasks, but it usually does require more than just the simple completion of a single task, the issue is often not whether you succeed in some task, but rather whether you succeed well enough to achieve conflict victory out of it. For example, if the conflict can be resolved by climbing over the wall and escaping into a crowd, then a "partial success" might be one where you "only" got a first degree of success in your climbing check; you'd have needed to get that +5 points to achieve second degree, in which case we could have established that you managed to disappear into the crowd. As it is, that partial success only buys you a venue change and perhaps a superior position from where to continue resolving the situation.

    If nobody has a subgame in mind when a task is rolled for, the degrees do nothing - the binary success model is the fall-back. However, if somebody rolls really well or badly even in a minor task, there is a vested interest for either the player or the GM in at least quickly thinking through the situation in case some natural follow-up occurs. So basically, if you fuck up a roll spectacularly, that'll only have consequences if the GM can invent something on the spot. If not, then apparently it wasn't a very interesting topic to begin with. Sometimes the consequences of extra success or failure are much more important than other times: if the situation is complex and there are lots of interesting moving parts, it's much more important to fail or succeed big than when the stakes are small and the situation straightforward.

    I don't know off-hand whether my system would do that well in d20. It's a pretty complete system in what it does, so I'm sure that something of this sort would fuck it up one way or another. But on the other hand, you could do interesting things with the currency. For example, my D&D doesn't use skills, but if it did, like d20 does, then support skills wouldn't just give you flat bonuses; you'd instead get bonus dice equal to your degree of success, like TSoY does.
  • As a house rule, I like the idea of having a second die roll as espoused by Felan and others.

    As a philosophical/design question, the linearity of a single-die roll is always going to model binary success-or-failure better than it does partial results, simply because any successful roll is as likely as any other successful roll, and ditto for failures. Some people like to phrase this as "with a d20, a 20 is just as likely as a 1", but what they really mean is "with a d20, a 20 is just as likely as a 19."

    In D&D, Complex skill rolls in 3.x and Skill Challenges in 4e attempt to add more options to the inherent binariness of the roll, by requiring multiple rolls and having each one turn into a partial success or failure. Skill challenges still typically rely on the result of the last roll, though. You could easily modify them so that, in addition to the final yes-or-no success roll, small penalties are incurred along the way; what those penalties are would depend on the situation.

    Alternately, you could say that unsuccessful die rolls succeed at the task, but with some kind of consequence or penalty. So the rogue unlocks the padlock, but just as he does, some bugbear guards round the corner; or a player's reflexes really do save him from a trap, but he slips and drops a potion container. This is the basic indie-game "yes, but..." pattern as applied to saves. The best way to handle this as a GM is to negotiate for the penalty before rolling, so players know the stakes they're up against.

    I go that way myself, but I also try to find options for the players where they have clear, unopposed actions which may themselves be successful or not. What I mean by this is, in addition to simply rolling to Detect Traps, I would allow them the option of, say, rolling a barrel down a corridor and setting off the poison darts or whatnot without being in the line of fire. That would always succeed, although deliberately springing the trap may have unintended consequences as well. But it wouldn't rely on a die roll at all: the roll is only there as a back-up.

    Finally, consider this: basic, simple combat is a d20 system that has partial success built into it: you hit, but only do 2 damage. The orc hits and does 3 back. After 3 turns, you've succeeded at killing or driving it off, but you've taken damage yourself. The basic pattern here is success plus a point tracking system, and a chance to lose something. I don't know how to apply this to other rolls, but it's useful to think of it in this light to see that the d20 isn't always straight success or failure.

  • edited March 2013
    There must be some times when you don't have a good twist, right? What do you do now in that situation. Plus, in d20 success+cost is easy: they have hit points!
    Don't GMs do this normally? I mean, even when running Rifts, if you miss the result by 15 points, I tend to describe a different outcome than if you miss it by 1.
    Alternately, you could say that unsuccessful die rolls succeed at the task, but with some kind of consequence or penalty.
    I guess what I mean is, I don't like having to make the call entirely by myself. My earlier posts might be too broad - I definitely narrate different kinds of outcomes for different rolls. But I think the "yes, but" success with a cost or compromise is a very specific kind of outcome that isn't really accounted for with d20 vs target number, even if the GM is interpreting the results relatively. It covers "yes, and" and "no, and" just fine - but not "yes, but."

    I would prefer to have a specific "trigger" for the "yes, but" GM action. I'm also uncertain how allowing d20 failures (ie: below the target number) be occasionally treated as compromised success in the fiction might mess with the math of the game - I feel like, according to the logic of the system, not meeting the target number should never result in a success. Likewise, I don't feel comfortable treating a roll that beats the target number as a compromised success, which just doesn't seem fair - if you beat the number, it's gotta be a full success, you've earned it.

    Hence why I think a player-bid method of some kind works well - the player knows in advance that only a compromised success is possible, and accepts the risk. What I like about the +2/+5 player-bid method described above is that it's a clearly-defined, fairly narrow threshold that both players and GM understand and are aware of, just as clearly as >DC = success. In case it's gotten lost in the shuffle, this is what I'm planning on trying:
    After a failed roll, the player can choose to add +2 or +5 to the total. If the new total bumps the roll over the target number, it is treated as a partial success - the degree of cost or compromise is determined by the how much the player "bids" - +2 is minor risk, +5 is a major risk.

    In a combat situation, a minor risk might simply result in an attack of opportunity; a major risk could be a disarm, or a bull rush off a platform. Out of combat, say lockpicking, a minor risk could result in broken tools; a major risk could result in a trap being triggered.

    Thanks for the run-down, @Eero_Tuovinen - a much more elaborate hack than I'd be willing to experiment with (I'm trying to leave the game as unaltered as possible), but you point to some interesting techniques.

    The various multi-dice methods suggested are cool, but for me personally, I'm trying to minimize the extra math. I feel like calculating modifiers on two dice for every roll would slow down the game too much.
    Finally, consider this: basic, simple combat is a d20 system that has partial success built into it: you hit, but only do 2 damage.
    That's a really good point - in and of itself, that's a multi-dice approach to degrees of success and failure. It's strange to imagine what that would look like outside of combat - roll your lockpicking skill, beat the target number and succeed; then roll your lockpicking tools' "damage" rating to see exactly how successful you are.


  • Here's a "solution" that totally doesn't work: if you hit your DC with just the die roll, you succeed. If you only hit the DC by adding whatever modifiers you add (skill ranks, etc.), you get a "yes, but" result.

    (This would basically mean that, the higher level you got, the more "yes, but" results you would get)

  • As a philosophical/design question, the linearity of a single-die roll is always going to model binary success-or-failure better than it does partial results, simply because any successful roll is as likely as any other successful roll, and ditto for failures. Some people like to phrase this as "with a d20, a 20 is just as likely as a 1", but what they really mean is "with a d20, a 20 is just as likely as a 19."
    IMO this is not totally true. All you need for non-binary results is a distribution of possible outcomes. Hence fail the roll, make the roll, make the roll by 5, etc is one way of specifying a distribution. Because the d20 is linear, changes to target number make linear changes to the probability of each outcome in the distribution. The nice thing about bell curves (such as the sum of two or more die rolls) is that you get non-linear adjustments to the probabilities for linear adjustments to the target numbers but this is not necessary for there to be a range of outcomes.

  • Here's a "solution" that totally doesn't work: if you hit your DC with just the die roll, you succeed. If you only hit the DC by adding whatever modifiers you add (skill ranks, etc.), you get a "yes, but" result.

    (This would basically mean that, the higher level you got, the more "yes, but" results you would get)
    No this is totally great with magic swords. When you need to use the sword the sword gets to use you too.
  • Here's a "solution" that totally doesn't work: if you hit your DC with just the die roll, you succeed. If you only hit the DC by adding whatever modifiers you add (skill ranks, etc.), you get a "yes, but" result.

    (This would basically mean that, the higher level you got, the more "yes, but" results you would get)
    No this is totally great with magic swords. When you need to use the sword the sword gets to use you too.
    Plus 5. This exchange is full of win on all sides.
  • Here's a "solution" that totally doesn't work: if you hit your DC with just the die roll, you succeed. If you only hit the DC by adding whatever modifiers you add (skill ranks, etc.), you get a "yes, but" result.

    (This would basically mean that, the higher level you got, the more "yes, but" results you would get)
    Yeah, that was my first thought! I think it would be great for an E6-type campaign, but my current group is at level 12, with DCs rarely below 20, and I don't think a switch-over now would be well-received.


  • As a philosophical/design question, the linearity of a single-die roll is always going to model binary success-or-failure better than it does partial results, simply because any successful roll is as likely as any other successful roll, and ditto for failures. Some people like to phrase this as "with a d20, a 20 is just as likely as a 1", but what they really mean is "with a d20, a 20 is just as likely as a 19."
    IMO this is not totally true. All you need for non-binary results is a distribution of possible outcomes. Hence fail the roll, make the roll, make the roll by 5, etc is one way of specifying a distribution.
    While I agree with that, I still think it is clunkier than generating a bell (or bell-like) curve, especially when you start moving the target numbers back and forth with situational modifiers or whatnot. Also, in any case where you need to roll a d20 more than once, it skews the probability more than I suspect most people realize.
Sign In or Register to comment.