One risk when you make a plan is that it is too inflexible, and, while the plan looked fine when you made it, as conditions change in the future, as edge cases appear you didn't foresee, it is no longer a good route to push on, or it needs serious adjustments.
Now we have two possibilities. If the person following the plan is stubborn and keeps pushing with the original plan, despite becoming clear at some point that the direction is wrong, it is very likely the plan fails or it produces lots of damages toward the end when it is followed despite going in the wrong direction.
If the person is flexible and adjusts the plan to the new reality, or if the plan itself is flexible from the start (much better), there is a chance it will still be carried on despite set backs or edge cases, without major issues.
Now, if a person adjusts a plan that is under way, how do they know they are thinking straight and not making decisions based on their emotions? What if the initial plan was sound and they can't see it anymore with the same eyes as when the plan was drafted?
That's a good question, I believe. Normally, the initial plan should remain the focus, if it was made when things were "cool" and judgement unclouded by emotions.
But the way I see it, if a medium-long term plan is changed because it was poorly made, is too inflexible or doesn't correspond to the existing reality, the change should not be taken lightly, in the rush of a moment. This can be different for trading, but we are not talking about this.
I believe at least a good night sleep is required before confirming the decision to change a plan that is on-going. Sometimes a longer period of analyzing and pondering the various options can be useful.
Of course, the best thing would be to have a great plan in place from the start. That's why it is important to take your time when you make your plans initially. But even when you take your time, you will rarely account for all the variables that are involved.
A little unrelated to the bulk of the post, I wonder how do AI handle edge cases nowadays? For example in self-driving. Have they gotten any better at it? There may come a day when AIs could plan and execute better than most humans on their domains of expertise. On the planning part I have my doubts. What if they hallucinate when making the plan?