You've probably heard of the "trolley problem": Five workers are working on some train tracks. An out-of-control trolley is barreling toward them, and they are all sure to be killed by it. If you hit a switch, the trolley will be diverted onto a different track, where it will hit and kill only one worker. If you could save five persons' lives by killing one, would you?
I hold a strong antipathy toward this problem. People like to use it as an example of testing how people would respond when asked to make "fraught" ethical decisions. But I think it fundamentally fails to simulate how such decisions are made.
A better trolley problem would be like this: The workers have put up a curtain on a scaffolding around their work area; it is the sort of barrier one sees erected around construction sites, to deaden noise and contain dust. From where you are standing, you cannot see into the work area. The trolley will burst through this curtain and kill the workers on the other side if there is anyone there.
Note that the workers may be on break right now, safely away from the tracks. They may be working off to the side of the track, not on it. They may be far enough from the curtain that, once the trolley bursts through, they will be able to dodge out of the way in time. There may be one worker in danger, or five, or twenty. You have no way of knowing any of this.
But you are near a switch, and if you press it, the trolley will be diverted onto a side track, and a single worker there (not concealed by any curtains and thus plain to view) will be killed.
Why is this version better? Because the original trolley problem, and others of the same kind, presume foreknowledge of all possible outcomes. Ethically difficult decisions in the real world do not work this way. Put another way, a decision made with such foreknowledge would be significantly less difficult in ethical terms.
When people pose you a "take a time machine and assassinate Hitler" hypothetical scenario, they are guilty of the same sloppiness. We have no time machines, and a preventive assassination or preemptive war will always be committed on the basis of an unprovable assertion: that if it isn't done now, we'll wish we had a time machine once we've seen the result. The best that can be said of such decisions is that, if successful, the erase any future evidence that they were justified. The worst that can be said is that they were unnecessary—like diverting a trolley onto a side track and killing a man when, in truth, there was no one in danger behind the curtain.
So, a (lengthy) postscript. One common variant of the problem is this: You're on a bridge above the tracks, and the trolley is coming and will kill the five workers. Instead of diverting the trolley with a switch, you're given the option of pushing a fat man who is standing next to you, so that he falls onto the tracks; the trolley will hit and kill him, but he's fat enough to stop the trolley. Kill one, or let five die? This variant is supposed to be "interesting" because people who were ready to hit the button in the first version, often claim that they couldn't bring themselves to push the fat fellow in the second.
But from my perspective, if the first version was bad, this one is even worse. Aside from the offensive supposition that the protagonist in the story must be rather slender themselves—positively willowy, no doubt—so that simply casting oneself onto the tracks, stopping the trolley at the cost of nobody's life but one's own, isn't a feasible option, a more fundamental flaw is that this story treats the idea that the fat man will actually stop the trolley as a given. How could one really know this? I mean, of course, in the real world, not in a smug hypothetical exercise?
(And if I may be permitted a quick oh-please moment... If the trolley is heavy enough, and moving fast enough, to run over and kill four workers, before running over and killing the fifth, then the fat man must have the stopping power of more than five normal men. If you've got the muscle to leverage a person that enormous up and over the rails of the bridge, you must be extraordinarily muscular... and thus pretty heavy yourself...)