7 verified2 unconfirmed2 contested
Researchers have reconstructed and published a mathematical solution devised by Nobel Prize-winning physicist Richard Feynman that addresses the dilemma of whether to try a new restaurant or return to a familiar one. Feynman first posed the problem during a lunch conversation in the 1970s and jotted down equations in handwritten notes that remained undeciphered for decades. In a study published in the Proceedings of the National Academy of Sciences, the team reframed the problem as choosing a restaurant each night over a fixed number of days in a city. Feynman’s solution involves comparing the best restaurant found so far against a threshold that declines as the number of remaining nights decreases, making it less worthwhile to keep searching near the end of the trip. The researchers also tested how people actually decide in an online task with more than 2,500 participants, finding that participants used a simpler but effective shortcut instead of the optimal formula. The team noted that the strategy should adjust depending on the overall quality distribution of available restaurants.
What’s verified
The solution originated from a problem Richard Feynman considered during a lunch in the 1970s.
Feynman’s handwritten notes on the problem remained undeciphered until analyzed for the current study.
The study was published in the Proceedings of the National Academy of Sciences.
Feynman’s approach uses a threshold: if the best restaurant found so far exceeds the threshold, return to it; otherwise, try a new one.
The threshold decreases as the number of remaining nights shrinks.
The researchers conducted an online task with more than 2,500 participants to test real human decision-making.
Participants did not use the optimal solution but a simpler heuristic that produced similar overall scores.
Where accounts differ
Source 1 states that in Feynman’s equations the threshold declines “more and more rapidly” as days left reduce, while Source 2 describes the threshold as “starts high and falls” without specifying the rate. Source 1 further reports that participant behavior showed a linear decline, not the more rapid decline Feynman’s formula predicted.
Source 1 identifies co-author Brian Christian as affiliated with the University of Oxford, while Source 2 identifies his affiliation as the University of California, Berkeley.
Not yet confirmed
The name of Feynman’s lunch companion (Ralph Leighton) is provided by only one source.
The exact year of Feynman’s death (1988) is provided by only one source.
Key figures
Richard Feynman, physicist and Nobel laureate
Prof Tom Griffiths, Princeton University (co-author)
Brian Christian, computational cognitive scientist (co-author)
Sources: The Guardian, sciencenews.org