Einstein made “thought experiments” famous. “Suppose you are traveling on a beam of light,” that sort of thing, where you lay out a hypothetical situation and focus your attention strictly on the terms of the hypothesis, taking care not to put the situation into any context where you would be inclined to draw analogies between it and something about which you already have knowledge. I suppose thought experiments are just the thing when you’re trying to think about something like the relationship between space and time, where our usual habits of mind are not of much help. I’m usually skeptical about thought experiments as a guide to understanding anything that goes on among humans, but I wouldn’t dismiss them entirely as a tool for clearing our minds of prejudice in any field.

The other day, novelist William Gibson proposed a thought experiment on Twitter:

Before posting my reply, I scrolled through the thread to see whether anyone had already said what I wanted to say. Several came close. For example:

To this and other tweets making similar points, Mr Gibson answered with a reminder that the terms of the hypothetical specify that the abductors can be trusted. Since the first rule of thought experiments is that you have to accept the terms of the hypothesis, this was all that had to be said about the matter.

This poster came closer to what I had in mind:

Mr Gibson is a novelist. A very strange novelist, but a novelist nonetheless, and so it was quite reasonable of this person to expect him to have some kind of plot twist in mind that would develop from the psychological motives of his characters. Still, Mr Gibson’s occupation is not part of the terms of the hypothesis, and so it really isn’t playing the game to bring this information in.

This person is also thinking novelistically:

This person seems to have a happy-go-lucky attitude, though it is unclear how deeply s/he has engaged the hypothetical:

This person did stay within the rules:

Though perhaps it occupies too narrow a space within the rules. The terms of the hypothetical are not phrased so as to exclude the possibility that meta-analysis of their own use of words might be the best solution, though I think a minimal application of the principle of charity will incline us away from that approach.

Intriguing responses all. But I had something else to say, and it took me three tweets to say it. Here they are:

“If you accept that, however, there is nothing to stop you attempting to escape or to overpower them”- I thought of making an explicit allowance that what is preventing you doing those things is some threat that frightens you even more than the thought of being killed, but I was already posting three tweets on a thread to which no one else had posted more than one, so I decided against making it even longer.

That it is “only one of a list of conditions” means that it is not the case that they will kill you if and only if you win the lottery. It means only that your winning the lottery is one of the circumstances in which they would be prompted to kill you. By “indefinably long,” I mean that there is no rationally discoverable principle that will set a boundary to the range of conditions under which they will kill you.

Why am I playing the game by importing information about how characters like those in the hypothetical are likely to behave when others were not playing the game by importing information about that topic, or about Mr Gibson’s occupation, or about anything else? Because I showed that the hypothesis was incomplete. The hypothesis rests on the assumption that the captive has sufficient reason to fear the abductors that s/he must submit to the threat of murder, but does not specify what those reasons are. This opens the door to a substantial amount of outside information.