Skill checks which are difficult even if you have a lot of time

[jack]

Background

It varies, but in several editions of DnD, the official way skills work is: if it matters if you succeed on a task (say, picking a lock) first time, the task has a difficulty assigned, and you roll a d20 and add your skill, and if your number is as big or bigger, you succeed.

If it seems just obvious your character can probably do this, the GM is encouraged to just let it succeed. If you care about mechanics, there's a very similar concept "take 10" which allows you to not roll and just assume you got 10 (provided you're in an everyday not hectic situation), ie. just assume you got an average level of competence most of the time.

Likewise, if you could do it *eventually* by exhaustively trying everything, the GM is encouraged to just assume you can do that, even if it takes a while. The mechanical equivalent is "take 20", ie. assume you take 20 rounds, but eventually get 20 on your check. That's what you'd guess, but it's what the rules actually say in several editions.

Since you can usually figure out what the "obvious" thing that happens is, there's not that much benefit to having those specific rules (they're only half there in 5e). They're far from always realistic. But they do the right thing some of the time, while giving a default to use if you're not sure if the mechanics matter or not.

Shortcomings

Implicitly the way it works is you're *either* quite high variance, or perfectly consistent.

It's not quite like that, if you roleplayed a mundane day-to-day activity repeatedly N times, you would assume you'd SOMETIMES screw up, even if it was something you could usually do, even if not as often as when people are attacking you.

But some situations, I'm not really satisfied with it.

If you're trying to pick a lock, or checking to see if you spot a hidden door, etc, then:

* If you roll every time, the party effectively have the value of the character with the best skill rolling perfectly, because they'll get there eventually
* If you assume you use 10, then the result is always the same: the character who's best at it ALWAYS spots it first, and the GM knows in advance exactly which locks and doors they'll succeed at.

Both are too deterministic, which takes away some of the fun for me: I *want* it to be a bit random. This lock is a bit harder than it looked from the outside. You get a bit of lockpick jammed in it. The party wizard just happens to remember an obscure lecture on magically concealing doors and notices a clue the rogue missed.

I've sometimes thought of this as "you roll, but if it doesn't make sense you'd succeed better a second time, then you don't get to reroll". But that's confusing and hard to track. Has the rogue tried this door already? If they have, why can't they try again?

My idea

It turns out, I actually want the take 10 and take 20 rules to work as they are, but maybe the difficulty should be randomly determined on the spot. Eg. instead of saying "all these doors are difficulty 15", say, "are difficulty 10+d10". Or a smaller variation, or occasionally a larger one.

That way there's a random element. Which doesn't matter if they're in the heat of combat, but does mean, there'll be some where "I can't get this one, we have to smash it" or "would one of you useless louts like to try to help?".

I could even generate a slightly different difficulty for different characters or different parties, although I'd probably only do that if it seemed to matter (and might simulate that in a more streamlined way, eg. dropping the difficulty for each party member, but choosing randomly who succeeds, the expert or someone else). That way you'd occasionally get a meaningful variation. And most of the time they'd deal with the encounter right then, only if they came back to it and it mattered that it was consistent would you have to track what you generated the first time.

Comments

[ptc24]

One neat solution: re-interpret the dice roll. So Bob tries to pick a lock, he rolls, not quite enough. What does the roll represent?

The particular merits of this particular attempt?
The real difficulty of the lock?
The match between the lock and Bob's fine-grained skill set?
Whether Bob is having a good day?

A dice roll, by the usual mechanics, but non-rerollable, could be interpreted as points two and three - essentially, the roll fills in details about the world on the fly, just like a roll on the wandering monster table.

If you go for point three, then maybe if Bob can get a skill boost somehow, then he gets a re-roll. If Bob goes and fights some kobolds and dings and then returns to the lock, well, maybe that lock he couldn't pick is something that he's been wondering about at the back of his mind and now he feels all accomplished and ready for another crack at the problem with genuinely fresh inspiration.

Also, point three makes it meaningful for different party members to have a go - the bigger the party, the more the people who might have the relevant knack.

[(Anonymous)]

The whole point of this kind of roll is that it's the equivalent of the bit in improv where the director rings the bell and tells you 'New idea!' so you have to come up with something that will hopefully send the story off in a new, more interesting direction.

Therefore the most important thing to remember is the primary rule of improv: It's never 'No', it's 'Yes, and'.

A character tries to pick a lock. What's the most boring thing that can happen? They fail. What's the next most boring thing? They succeed. What's more interesting? Practically anything else: a bit of their lockpick snaps off in the lock, now they have to get it out or the guards will know they've been there. A bit of their finger gets stuck in the lock, now they have to work out how to get free before the next patrol. The lock opens… but unknown to them it's set off an alarm in the control room and every guar din the place is converging on them right now.

Don't go down the interesting spectrum, always go up. Never just, 'No, you fail': always 'Yes, and…'

(And if you can't think of anything more interesting, then don't even bother with the rules, just skip over it, just like a film doesn't show every lock being picked during the heist but cuts straight to the bit where something interesting happens. The point is the story, not the rules.)

[jack]

Who are you? If you're not signed in to dreamwidth, could you sign your comments somehow?

This is a very good description of dnd, i was thinking exactly of the "yes, and", and yes, as much as possible, I seize opportunities for failure to be "success but with complications" not "you can't do that".

But I think there has to be *some* background. When I improvise, I seize on *some* interesting hooks, but I need a variety of stuff some of which is just straightforward, in order for some them to stand out.

I think a big part of the DnD experience is experiencing these things as actually happening to a character. If every single lock jams in an entertaining way, it's more slapstick, if any lock *might* do, but most are dealt with ok, the players are more involved in what happens, making implicit decisions on what risk trade-offs to take, and it's interesting when something unusual happens.

For that to work, there needs to be some underlying consistency, hence rules beyond "everything has a flat 75% chance of success" or "there are no rules, we just decide what would be most interesting". And if there are rules, without *over* complicated them, it's useful if the rules reflect things that are interesting most often.

[simont]

I *want* it to be a bit random. This lock is a bit harder than it looked from the outside. You get a bit of lockpick jammed in it.

The interesting thing for me is that these two failure modes are very different sources of randomness.

In the first one, the underlying model is that locks vary independently at random, rather than attempts; so you can try as many times as you like at the same door, but you'll get the same result every time, so there's no point trying more than once. You could imagine dealing with this using a sort of random-oracle model in which the first time you find a given door the GM rolls to generate the stats for its particular lock, but once they're generated, they won't change on future encounters with that same lock.

But in the second, the model is that the attempt actually makes matters worse – if a mediocre lockpicker breaks off a piece of pick in the lock, then not only do they fail to open it, not only is it pointless for them to try again, but they've now messed up the lock to the point where perhaps even a very skilled lockpicker would fail to open it, even if they could have succeeded if the mediocre person had never touched it.

Put another way, in the second case, the Markov chain of what happens in repeated attempts has more than one absorbing state. The 'take 10' rule makes sense if the only absorbing state is the 'success, lock is now open' state; in that case it really is true that with enough repeated attempts you'll inevitably get there in the end, and it makes sense to simply have a crude model of about how long it will take you (which might be done by a mathematically justifiable model of probabilistic expectation, but might be the really simple 'just don't try it in the middle of mortal combat' which is all the gameplay requires in practice.) But if there are two absorbing states, then what needs to be modelled is not just the expected time to hit one of them, but also, the relative probabilities of which one you hit. So you get to do a single roll that models your whole series of attempts at the task, and the outcome of that roll is that you either eventually end up opening the lock or eventually end up jamming it beyond recovery. And again, you could do the actual Markov-chain theory to figure out a mathematically justifiable account of what the distribution ought properly to be for the length of 'eventually' and the choice of absorbing state, or you could go with a crude approximation if that seems more sensible.

[jack]

Hah! Now you've inadvertently tied this to stuff I was thinking about QM.

In DnD, how much do die rolls determines *what happens* against a fixed pre-existing reality, and how much do they determine what was there but not mentioned?

That is, how much chance is quantum-true-randomness or chaotic-unpredictability? And how much is "discovering what was true all along but as yet undetermined".

That last is like baysean probability of things that only happen once, like, how likely is it the fundamental constants are a particular value. Or quantum probabilities in many worlds where every outcome is guaranteed, but you find out which one you are in the same universe as).

Both clearly happen sometimes. Most combat rolls are "do you hit them or not". GMs often use dice to decide what's there, when many answers are plausible and variety is more interesting than any fixed answer (eg. the guard patrol is coming round at SOME point, is it now? Wandering monsters are an oft-abused form of this).

What I'm wondering is, are rolls sometimes a hybrid? It happens with knowledge rolls "oh yes, I remember a lecture about that" vs "oops, I slept in and missed all those lectures". And part of what I was reaching for is maybe it happens with locks -- do you fail *that attempt*, or do you determine if the lock is pickable by a technique you know? Or both?

I like having both sorts of randomness.

Modelling it as absorbing states makes sense for some things: that's what would happen if you tried repeatedly. The official take 20 rules used to say if there are success outcomes and bad outcomes, assume both happen.

But it's not always what I want -- it's hard to fail a "do you notice a secret door" so badly that you physically prevent yourself spotting it in future. You could have that mechanic (you put something in front of it, or you just subconsciously discount the possibility) but it doesn't really fit.

Maybe I can combine both, more chances of lockpicking screwing up, but also, some chance of "nothing goes wrong but it's just too hard".

Or I could fold those into one thing. For other reasons I've toyed with mechanics for determining *sort* of success. I borrowed this from a different system on the alexandrian, but basically, in addition to success/failure, a random roll for "due to brute force, due to finesse, due to external circumstances, due to style, due to luck" etc. Sometimes you don't care, but when you score a critical hit, you can choose to describe it as (external influence) "a weak point on the orc's armour give way, your sword sinking deeply into their flank. roll damage" or (style) "dodging their blow, you spin completely around into a perfect fencer's lunge into its unarmoured belly. roll damage". If you had that system already, you could interpret the answers as the sort of failure, eg. brute force failure = you broke it; external circumstances failure = just too hard; style failure = "you roll your tools back into the little black cloth and tuck them into your belt. As your teammates jabber at you, you stalk away, proclaiming you will sleep on it".

[gerald_duck]

Have you ever looked at Rolemaster? It's been several decades since I played it, but back in the day it had a far more rational system for this kind of thing than D&D.

One core concept was open-ended rolls: you roll d% but on a 96-00 you roll and add, on 01-05 you roll and subtract. This process repeats as long as you continue to make outlying rolls.

This did several neat things:

• It provided a framework for attempting extremely implausible things, giving them a non-zero chance of success and characterising the severity of the consequences if/when you failed.
  
• Conversely, it also provided a framework for attempting utterly routine things. Normally, you wouldn't bother rolling to check the player walked from one side of the room to the other without stumbling, but what if they're carrying a Ming vase?
  
• Within that framework it was easy to apply modifiers for superhuman dexterity, having your legs tied together, or whatever.
  
• It was also easy to apply modifiers for how much time and care was taken over an activity.

[jack]

Hm. I haven't, but I've read similar things.

I like the open-ended result concept. But I don't think it particularly fixes anything. Some tasks have a very band of success, someone with more training succeeds basically 100% of the time and someone without doesn't. Others are much more variable (eg. might one lucky blow fell a giant?). Picking either "bounded" or "open ended" and assuming that's always the case doesn't really fix any of the inconsistencies, what's missing is some standards for where on the spectrum to evaluate a particular task.

I might occasionally tack on an "unusually lucky result" or "an unusually bad result" if I wanted a particular test to allow extreme success or extreme failure by doing basically the same thing (roll d20, if you get 1 or 20. roll again and add/subtract 20). But you can also do that by having a couple of successive tests (for skill tests not combat):

P1: I flap my arms and try to fly!
GM: *sigh* OK, roll.
P1: Nat 20!
GM: OK, one flap seems to maybe work. Roll again.
P1: Another nat 20!
GM: Oh my god it seems to be working. Roll again.

And basically to do something godly, you need to roll three nat 20 on the trot.

That's occasionally useful but if you're still rolling results, it doesn't seem to answer the question of "which rolls does it make sense to try again and which doesn't it".

Is there something other than "it's unbounded" which I should be paying attention to?

[damerell]

To a degree I think these offset each other. If the party all take 10 on something, sure, the best skill will always come out top, but we already agreed it was a routine sort of task, so perhaps we don't really care who gets it done - and we were willing to forgo the chance of an exceptional result by not rolling. If we take 20, well, it makes sense to have the most skilled person attempt the task if we're all going to hang about for 20x the normal duration.

It is a _bit_ odd that in the take-10 take-20 cases the best skill always comes out top, but I don't find it very odd. That said, when I was last running d20 the players almost never chose these options; mostly taking-10 I regard as a convenient fudge to let the players do something I wanted to let them do anyway without the chance of rolling an inconvenient failure.

For exceptional results I've never gone beyond copying the combat critical rules: on a 20 roll again and if the second roll is a success you get an illustrious success; on a 1 roll again and failure on the second roll means hilarity. I have no idea if this is canon (and if so, in which editions/d20 variants/Pathfinder...)