I am always curious how others use tool rules.
In most situations (IE when not in an encounter), couldn't the character just keep trying until they get a perfect result, assuming they have time?
Seems to me if there is a locked door, unless "noise from the check" is relevant to alerting something on the other side, or "time spent which allows wandering monsters to come by" is relevant, the check would always (eventually) succeed, right?
How do others do this? Do you allow multiple checks until it finally works, and if not, why not?
In 3e we just used the Take20 rule. But honestly, it delivers a sour feeling. Because that maximum DC you can hit only with a 20 and maybe get a "hurrah!" moment, now it's a guarantee. With a skill allowing Take20 (not all do), everything is either certain success or certain failure (I know, I know... "not if you are under pressure", but the truth is those skills most often than not happen when you are not under pressure).
There's another, very different approach: just allow
one check per challenge, and rework the narrative backwards. E.g. when a lock has DC at which you'd succeed with 30% probability, don't think of it as "if I try to pick
this lock 100 times, I would succeed ~30 times", but instead think of it as "if I try to pick
100 locks similar to this, I would succeed with ~30 of them". In other words, you have ONE attempt, if you fail it means
this lock is not one of those 30 locks you can pick.
But then again, I never liked "take 20" because I've never known someone who failed at something 19 times in a row to attempt it a 20th time. Most people get frustrated at about the 5th time they do something and give up, assuming it is too hard for them.
Definitely a good point... a D&D character would stubbornly repeat the attempt 20 times,
only because the player knows that she will succeed. In a sense it's an instance of metagaming.