There's a concept that I don't really have a word for, but that lies at the heart of a lot of gaming (and writing). I don't see it talked about much, I think because it's a little bit too big to see. It's such an essential part of making anything happen that it's easy to look. For lack of a better word, I'm going to call it "expectation" but that does not quite convey the whole of the idea, but let me drill into it a bit.
The core of it is this: people have a natural sense of what should happen next. We're not all totally in tune on it at all times, but there's a lot of commonality in our take on things. This sense is an essential part of storytelling because stories depend on violating that expectation. Something must take a turn away from the way things are supposed to go, or else there wil be no story, just a logical series of events. This expectation is also essential for great things like humor and irony, since they hinge upon the friction between that expectation and what actually happens.
This expectation is also essential to the mechanics of most roleplaying systems. More than anything, it is essential to the decisions around when to roll the dice and when not to roll the dice. A player describes what his character is doing, and the GM consults his internal expectation - if a clear outcome suggests itself, then there's no need to consult the dice. If there's uncertainty, then the dice might be turned to for an answer.
This tends to happen kind of naturally - it's just the process that we go with, but it can be very interesting to stop and think about this step since so much hinges on it. Enjoyment of play can hinge on the players and GM having similar expectations - if they don't, then it is easy for the other party's decision to seem capricious or even malicious. Even more, the mechanics of almost every game out there depend in large part on this expectation.
This is a hard space to call out mechanically. If you stop and think too much about it you run into the same problems you have when everyone stops and negotiates potential consequences of a scene. It might be very thorough, but it breaks flow pretty hard.
Instead, it is merely something to keep in mind as you run. It may be the single most important tool in your arsenal for deciding when not to engage the system. It's easy to fall into the habit of trusting the rules to provide the cues for when to use the system - if there's a rule, then you should use it right? That's not necessarily a terrible thing, but it's not a reliable yardstick for when it will be fun to bust out the dice versus when it's smarter to just keep things moving.
If, instead, you can keep a firm grasp on your sense of what should happen next, then it's much easier to tell when to turn to the dice. You can trust your own sense of uncertainty, and you can focus on those situations where you want things to take a drastic left turn. That ability to select when to engage the system helps make sure that you do it when it matters, and it's amazing how many headaches you can save yourself.
The vast majority of problems that come from the dice going askew are rooted in a misjudgment before the roll, such as when a failed roll will result in an undesirable outcome. It's possible to to try to wiggle out of such a roll, but it's easier still to not make the roll in the first place. If only one outcome is tolerable, then that speaks directly to your expectations for the roll and the action - just go with what you expect, and save the dice for sometime they'll actually help.