That is the title of this paper by Brian Christian, Evan M. Russek, and Thomas L. Griffiths, all of whom are affiliated with Princeton University. Here is their introduction:
“In the late 1970s, the physicist Richard Feynman sat down for lunch with his friend Ralph Leighton at a Thai restaurant called Indra in Glendale, California. Leighton was trying to decide whether to order his running favorite (the ginger chicken), or try something new that had a chance of being even better. Feynman turned the dilemma into a math problem—and solved it. Unfortunately Feynman never published his analysis. All that remained from the conversation were his handwritten notes, which Leighton happened to keep. The notes remained inscrutable for decades, until we managed to decipher them and reconstruct Feynman’s original problem and solution.”
More details here and here. Alas, I have to confess that when I first heard about the “Feynman restaurant problem” back in June of this year (see here), my first reaction was: “What’s the big deal? Feynman’s restaurant problem is just a ‘best-choice secretary problem’, a problem that was solved before the 1970s” (see here or here, for example). I am highlighting this problem now only because I just noticed a minor connection between Feynman’s restaurant problem and me: the restaurant that instigated Feynman’s investigation was located on Verdugo Blvd. in Glendale, California, just a few blocks from my childhood home!
Bonus link: My 2016 paper “Probabilistic Interpretation” models the problem of legal interpretation (i.e. how best to interpret a statute or constitutional provision) as a “best-choice secretary problem” in which a problem-solving judge strives to select the best interpretation of a given rule from a finite set of n possible interpretations of such rule. That paper, in turn, was was inspired by two of my heroes, Martin Gardner and Richard A. Posner.


