This seems like one of the less credible Umeshisms.
Suppose that at the margin you can get to the airport one minute later on average at the cost of a probability p of missing your plane. Then you should do this if U(one minute at home instead of at airport) > -p U(missing plane). That’s not obviously the case. Is it?
The argument might be that if you never miss a plane then your current plane-missing probability is zero, and therefore you’re in the “safe” region, and it’s very unlikely that you’re right on the boundary of the safe region, so you can afford to move a little in the unsafe direction. But this is rubbish; all that “I never miss a plane” tells you is a rough upper bound on how likely you are to miss a plane, and if you don’t fly an awful lot this might be a very crude upper bound; and depending on those U() values even a rather small extra chance of missing the plane might be (anti-)worth a lot more to you than a few extra minutes spent at home.
(A couple of other fiddly details, both of which also argue against arriving later at the airport. First, you may be happier when you leave more margin because you are less worried about the consequences of missing your plane. Second, even when you don’t actually miss your plane you may sometimes have to endure lesser unpleasantnesses like having to run to avoid missing it.)
What about “power actions” in board games? Here I mostly agree with Daniel, but with a proviso: If there is uncertainty about when the game is going to end, then on any given occasion you may find that it ends before you use a “power action” but that you were still right (in the sense of “executing a good strategy” rather than “doing things that turn out in hindsight to be optimal”) not to use it earlier. Because maybe there was an 80% chance that the game would go on longer and you’d have a more effective time to use it later, and only a 20% chance that it would end this soon.
On the real-life application, I think Daniel is probably right that many people are too conservative about spending social capital.
This seems like one of the less credible Umeshisms.
Suppose that at the margin you can get to the airport one minute later on average at the cost of a probability p of missing your plane. Then you should do this if U(one minute at home instead of at airport) > -p U(missing plane). That’s not obviously the case. Is it?
The argument might be that if you never miss a plane then your current plane-missing probability is zero, and therefore you’re in the “safe” region, and it’s very unlikely that you’re right on the boundary of the safe region, so you can afford to move a little in the unsafe direction. But this is rubbish; all that “I never miss a plane” tells you is a rough upper bound on how likely you are to miss a plane, and if you don’t fly an awful lot this might be a very crude upper bound; and depending on those U() values even a rather small extra chance of missing the plane might be (anti-)worth a lot more to you than a few extra minutes spent at home.
(A couple of other fiddly details, both of which also argue against arriving later at the airport. First, you may be happier when you leave more margin because you are less worried about the consequences of missing your plane. Second, even when you don’t actually miss your plane you may sometimes have to endure lesser unpleasantnesses like having to run to avoid missing it.)
What about “power actions” in board games? Here I mostly agree with Daniel, but with a proviso: If there is uncertainty about when the game is going to end, then on any given occasion you may find that it ends before you use a “power action” but that you were still right (in the sense of “executing a good strategy” rather than “doing things that turn out in hindsight to be optimal”) not to use it earlier. Because maybe there was an 80% chance that the game would go on longer and you’d have a more effective time to use it later, and only a 20% chance that it would end this soon.
On the real-life application, I think Daniel is probably right that many people are too conservative about spending social capital.