Can one ever hear too many stories about Richard Feynman? Cosmic Variance brought my attention to a great piece from Physics Today by Daniel Hillis (who worked with Feynman in the 80's to build a pioneering computer) that shows us how a great scientist continued to tackle interesting problems well into his 60's. How did Feynman do it - how did he keep finding interesting problems to work on? This piece describes some interesting work Feynman did on scientific computing, something he had experience with in Los Alamos, but it was not his main area of focus when he was making his name in quantum electrodynamics. Hillis, who convinced Feynman to come out and spend a summer working in his new computer company, tells us how Feynman loved to pick up problems he deemed crazy. Picking up crazy problems is in line with the advice I've heard from various established scientists: you first need to make your name (that is, get tenure) doing something interesting but mainstream enough so that your work gets recognized. Once you've established yourself, you're free to work on any crazy problem you want (as long as you can get someone to fund it!). Feynman followed this path: he won his Nobel Prize making progress on one of the major problems facing the physics community at the time, but from there he picked up whatever problem interested him, no matter how far outside the mainstream. (The problems weren't always outside the mainstream - Feynman just went after whatever interested him.) By doing this, he made major contributions to many different fields, from computing to superconductivity. The other thing we learn about Feynman (which of course Feynman fans already know) was that he could often explain the work of Hillis' computer company better than anyone else. He could cut through the details and put forward compelling explanation for why this company was currently working on the most exciting problem in the world. As Hillis said,

Even when Richard didn't understand, he always seemed to understand better than the rest of us.

You can see this phenomenon all over science, when you compare grad students their faculty advisors: grad students, who know the details of their work better than their lab advisor, often can't make a compelling case for their research the way their professors can. Good scientists never forget the overarching justification for their work, and they can frame it in a way so that the big arguments become clear to an audience. Feynman was a master at this. This is a great article for Feynman junkies, and anyone who likes to read about how great scientists do their stuff. And one more great quote:

As it turned out, building a big computer is a good excuse to talk to people who are working on some of the most exciting problems in science. We started working with physicists, astronomers, geologists, biologists, chemists --- everyone of them trying to solve some problem that it had never been possible to solve before.