Restricting rerolls in D&D

[Macbeth]

If the character with the best attack bonus fails to hit a monster, do you let him try again? Do you let characters with lower bonuses try to hit the monster? Why is this different?

It's different because combat is a constantly shifting collection of moving parts. Since I tend to run with letting it ride until circumstances change, the circumstances in combat are, pretty much by definition, constantly changing, so you can always roll again in the next round.

If unlimited rerolls are allowed without consequence, it's like always taking 20, just with way more dice rolling. If I can keep trying to open the jar (which contains a jinn, of course) just by taking more time and rolling again, why don't we just skip all the rolling? If the amount of time is important, why not just make one roll and have that create consequences?

Expanding that out a bit more: I try opening the jar, fail, and get to keep rolling. Each roll takes time, so eventually there are consequences. Still, I sit at the table, rolling a dozen times before enough time has passed for the bad guys to find me. So, I rolled a dozen times, but there were only two possible consequences: I get it open, or I run out of time.

Now, with letting the roll ride, I roll once. If I succeed, I can get it open. If I fail, I can't, and the consequences can happen NOW. Instead of sitting around rolling for a few minutes, we make one roll and get to the interesting parts: what's inside, or what happens when I run out of time.


It's not a matter of player or GM benefit either: consider making a Stealth roll as opposed to opening the jar. If we don't let a Stealth roll ride (until circumstances change) the GM can just ask for rolls until you fail. If we don't let the Strength check ride, the player can just roll until they succeed.

 

[Janx]

It's not a matter of player or GM benefit either: consider making a Stealth roll as opposed to opening the jar. If we don't let a Stealth roll ride (until circumstances change) the GM can just ask for rolls until you fail. If we don't let the Strength check ride, the player can just roll until they succeed.

If you only get 1 chance to open the jar, you don't need to formulate an EXTRA consequence.

Failure means the contents of the jar remain unavailable to you.

if you get retries, lack of any extra consequence means eventually, the jar will be opened. Thus, the PC is never really denied the contents of the jar.

to sum up:
basic failure IS the consequence, you are denied the benefit that success gives
to be able to retry must have a price or consequence, in order to make initial failure be significant.

 

[Macbeth]

Yeah, Janx, we're pretty much on the same page. But basic failure is only a consequence if someone (the GM) or something (the rules) makes it so. Not being able to roll again (either as stated in the rules, or determined by the GM) is a consequence. Spending enough time that something interesting happens, like wandering monsters, is a consequence. Just wasting everyone at the table's time to roll again isn't.

 

[delericho]

It's different because combat is a constantly shifting collection of moving parts. Since I tend to run with letting it ride until circumstances change, the circumstances in combat are, pretty much by definition, constantly changing, so you can always roll again in the next round.

Well, fair enough. If we're desperately trying to escape a combat we can't win, but the only door is locked, do you allow our Rogue to retry his Thievery check every round?

If you do, you've just made it much easier to open that door while in combat than it is outside of combat! If you don't, you've stuck the party in a combat they neither win nor escape.

Or another example: If our Cleric is down and dying, do you allow us to retry Heal checks to stabilise him? Again, allow a reroll, and you make the check much easier. Don't do so, and a single failure is likely to doom the character.

If unlimited rerolls are allowed without consequence, it's like always taking 20, just with way more dice rolling. If I can keep trying to open the jar (which contains a jinn, of course) just by taking more time and rolling again, why don't we just skip all the rolling?

Well, indeed. I actually said as much in the post you quoted!

If the amount of time is important, why not just make one roll and have that create consequences?

If I have two minutes to open a door, I have a much greater chance to succeed than if I have two rounds! Now, this can either be modelled using multiple rolls, or by giving a bonus to a single roll, but it really should be reflected in the system one way or another.

(Incidentally, it was precisely this that was at the heart of the flaws with the original Skill Challenge system - using multiple rolls changes the probabilities of success in a manner that is not immediately obvious. WotC's system made Skill Challenges too hard. Enforce a single roll, without some sort of correction for time available, and you are doing essentially the same thing.)

Expanding that out a bit more: I try opening the jar, fail, and get to keep rolling. Each roll takes time, so eventually there are consequences. Still, I sit at the table, rolling a dozen times before enough time has passed for the bad guys to find me. So, I rolled a dozen times, but there were only two possible consequences: I get it open, or I run out of time.

If the DC of the check is 20, and I have a +5 modifier, then I need a 15 or better on the die - a 30% chance of success.

If you give me one roll, then, I have a 30% chance of getting the jar open.

If you give me a dozen rolls, my chance of success is (1 - (0.7 ^ 12)) = 98.6%.

It's not a small difference - it's the difference between probable failure and near-certain success.

If we don't let a Stealth roll ride (until circumstances change) the GM can just ask for rolls until you fail.

He could, in the same way he could secretly add 20 to the DC. He shouldn't for the same reason.

If we don't let the Strength check ride, the player can just roll until they succeed.

If time isn't a factor, a sensible group will apply the "take 20" rule, and skip rolling entirely. (Although, I've just noticed that "take 20" isn't in my 4e PHB. Hopefully just an oversight by WotC. In any case, if time isn't a factor, and it's within the abilities of the character, the DM should just "say yes".)

 

[Umbran]

When in doubt, steal from another game.

I first ran into "frustration penalties", in the context of White Wolf games - if you tried to do something, and failed, you could generally try again. But, each time beyond the first added a +2 to your target number (usually until you'd had some significant break or rest - your penalties usually went away by the next day).

So, you can try again, but if you keep failing, you eventually drive the target number up so high that you cannot complete the task. The mechanic allows for some retries, but not an infinite number.

 

[Plane Sailing]

I like that 'frustration penalties' idea.

 

[Macbeth]

Well, fair enough. If we're desperately trying to escape a combat we can't win, but the only door is locked, do you allow our Rogue to retry his Thievery check every round?

Well, ideally, the people who design the game have figured this out and the game tells what a roll means, when it can be retried, etc.

In by-the-book D&D, I'd let them retry. In my D&D, I'd not let them retry, but take that into account when setting the DC/making the skill challenge. In Burning Wheel, I wouldn't let them retry without significant change in circumstances. In Apocalypse World, they can retry, but failure has consequences, which might block a retry.

If you do, you've just made it much easier to open that door while in combat than it is outside of combat! If you don't, you've stuck the party in a combat they neither win nor escape.

Woah, wait a second, you're assuming a lot of other factors here. An otherwise unwinnable fight, for one.

I'm also not 100% sold that it's easier. 'Easy' is a tough word to use here. Yes, they are more likely to succeed on the roll, but they're also more likely to suffer the consequences. Ignoring a fight to pick a lock is a big deal.

Or another example: If our Cleric is down and dying, do you allow us to retry Heal checks to stabilise him? Again, allow a reroll, and you make the check much easier. Don't do so, and a single failure is likely to doom the character.

The rules tell me to allow multiple rolls, so I do. At the risk of being Forge-y: system matters.

I also think that makes sense with changing circumstances: something significant has changed between rolls, they're now closer to death.

Well, indeed. I actually said as much in the post you quoted!

I know!

If I have two minutes to open a door, I have a much greater chance to succeed than if I have two rounds! Now, this can either be modelled using multiple rolls, or by giving a bonus to a single roll, but it really should be reflected in the system one way or another.

[...Math...]

It's not a small difference - it's the difference between probable failure and near-certain success.

Yeah, totally. There's different approaches to represent the benefit of having time to work: Taking 10/20 in D&D, working patiently in Burning Wheel, etc. There's also the GM judgment call, of course: I often have things that require rolls during combat (say, opening a door) but don't once you have time. I basically assume that someone has the bonus and can take 10 and get on with the game. Only make rolls when there's something to lose.

He could, in the same way he could secretly add 20 to the DC. He shouldn't for the same reason.

Well, okay, you and I both know that, but does some other DM? From a quick flip through the PHB and DMG (I don't have essentials on hand) a specific skill challenge can say a skill can only be used once, but there's nothing else about retries. There is, however, talk about fudging, setting DCs, etc. So the game doesn't really say I can't keep asking for Stealth checks whenever I feel like it, outside of combat and skill challenges.

If time isn't a factor, a sensible group will apply the "take 20" rule, and skip rolling entirely. (Although, I've just noticed that "take 20" isn't in my 4e PHB. Hopefully just an oversight by WotC. In any case, if time isn't a factor, and it's within the abilities of the character, the DM should just "say yes".)

I would take it a step further than that (and the skill discussion in the PHB and DMG seems to agree): I wouldn't make them roll when there aren't interesting consequences. Time is an obvious one, with being found out and so on, but it's not the only one.

 

[The Human Target]

I used to be pretty lenient on allowing multiple attempts on skill checks and allowing people to aid each other a lot.

I think my game has improved since I've cracked down on it.

 

[delericho]

I would take it a step further than that (and the skill discussion in the PHB and DMG seems to agree): I wouldn't make them roll when there aren't interesting consequences.

Really good advice, this.

 

[S'mon]

3e assumed unlimited rerolls, hence "take 20". I find that only 1 roll makes many things too random. My compromise is to take a hint from Skill Challenges and typically allow rerolls, but an accumulated 3 failures means the task is failed, whatever it is. I find this works VERY well, and can cause a lot of tension when eg the Rogue has failed twice to pick the must-be-opened lock - one more failure & that's IT!