Monday, January 26, 2015

What would Feynman do?

What would Feynman do?  This is an old blog post by Eric Lippert, but I saw it for the first time today.  It speculates on how Richard Feynman might have answered one of the crazy “lateral thinking” interview questions that used to be a fad at many high tech companies, including (most notoriously) Microsoft, Oracle, and Google.  Some of it is quite funny, especially if you're a Feynman fan.  Here's a sample:
Interviewer: Forget about measuring the voltage already! Suppose you can't reach the fixture to measure its voltage.

Feynman: Again, I must point out that it seems very odd to ask a question about diagnosis of an electrical system while not allowing the diagnostician to use common electrical tools. But anyway, you said that I was on the right track, so let's go with that. We know that modern dimmers do not put a variable resistance across the AC signal; rather, they selectively "cut out" a variable-sized portion of the wave and leave the rest of the cycle in its normal size and shape. We could build a device that works analogously to a dimmer, but much slower. The device could have a couple of rotating cams that flip a switch on and off once a second. Now we need not disassemble any of the switches, or cut the power at the panel. We attach the device to the first switch, flip the second switch off, and the third switch on. Since we have already established that the switches are single-location switches that have been wired correctly according to the NEC, we know that the switch in the "up" position is energizing its lamp and the one in the down position is off.  Now we go into the other room. The lamp that is off is controlled by the third switch, the lamp that is on is controlled by the second, and the one that is flipping on and off every second is controlled by the first. This system will work no matter what kind of lamps are in the fixtures, provided of course that they are good lamps, not burned out.
This reminds me of a story I read about Feynman, though I've forgotten where I read it, and why the story was being told.  In the story, someone who knew Feynman wondered out loud why the altimeter in an airplane always seemed to be wrong.  Feynman launched into a learned discussion of how the altimeters worked (bellows-type pressure gauge), the loose correlation between air pressure and altitude (due to humidity, temperature, and wind), and the numerous engineering problems associated with bellows-type pressure gauges (including non-linearity of the spring force in the bellows, thermal expansion and stiffness variation, and the “stiction” effects caused by metal crystals rubbing on each other).  Feynman's companion asked how Feynman came to know all this about how altimeters worked – and it turned out that the only thing Feynman knew was that they used bellows-type pressure gauges.  Everything else he deduced or inferred – classic Feynman...

No comments:

Post a Comment