We have so many options of what we could do for every second of every day of our life, that the potential outcomes are almost unthinkable, from ending up considerably better off (these ones are tricky to find) to ending up ‘on a list’, in prison or dead (there are lots of obvious ways for these ones).
The problem is that there is an opportunity cost to thinking about every single decision, so for almost everything we just act on autopilot. Some things we deem important though, we take the time to really evaluate what our options are, so that we can make sure we are taking the optimal decision, at least given the information we have at the time. Sometimes the effort we spend here can actually get in the way of actually just trying out some of the options, especially when the downside to trying things out is low.
The problem with just trying everything, of course, is that lots of the options are completley useless, if we’d just taken a few moments to stop and think about it. Some obvious heuristics on when to stop and think are if there is a potential big downside to getting the decision wrong, or if the potential upside to a correct decision is very high. Something less obvious is when the branching factor around certain options are particularly high, trying too many options will cover such a small fraction of the search space that the chance of a close-to-optimal option being discovered doesn’t move the needle.
So, beyond the obvious high-stakes situations, the other place is when there are so many potential options, leading to such varied outcomes, that it will be impossible to test a reasonable fraction. Here you need to model your problem, and use that squishy stuff between your ears to pare down the options into something that will be tractable to actually find a better outcome.
I thought of this as a human-behaviour analogy to robotic motion planning. Even when you plan for just a few dimensions and limited time horizon, there are still so many options for a motion planner to explore that you have to use very strict heuristics to limit the sampling space.