Assumptions about the rationality of the participants can become extremely complex.
But that is my entire point I didn't have to make those assumptions because the conditional probability of losing 1% to the unknown is so much greater than the conditional probability of losing 50% to the unknown, so that is clearly the inferior risk in an ambiguous specification (which might actually be the captain's information set).
Ah, that's one way to look at it. I assumed from the way the puzzle was presented that any interpretation that involved probabilities was not be considered. Just because that seems to be the cultural norm for such questions. I think IQ tests are more about culture + a decently high bar for IQ, than a direct measure of anything. That is, over a certain number what they're really measuring is how clued in you are to the tacit assumptions of academic/mathematical/mind-puzzle culture. Kind of like I can usually predict how highly a LessWrong.com comment will be upvoted almost solely on how LessWrongian it sounds, regardless of actual content.
By the way, it seems that I, too, never sleep:
Some of us just have very irregular sleep schedules. I've been up all night and it is now 8am in Japan, and I'm about to do 4 hours a work if I can just get break out of the Bitcoin news cycle.