[ Jocelyn Ireson-Paine's Home Page | Publications | Dobbs Code Talk Index | Dobbs Blog Version ]

On Handling the Data

Against a student's faulty program, I might scrawl "infinite loop" or "uninitialised variable". Against a broken SF story, a Turkey City Lexicon reader might scrawl "Cozy Catastrophe", "Signal from Fred", or "Squid in the Mouth". Or "AM/FM". The classic AM/FM story must be Heinlein's The Day After Tomorrow: six men in an underground citadel, sole remnant of the US Army after invasion by the PanAsian Empire, take up a dead colleague's discovery of the electrogravitic spectrum, the magnetogravitic spectrum, and the electromagnetogravitic spectrum. Maxwell via Hertz to tractor beams and transmutation in a month: ultimate humiliation of the PanAsians, and I wish my hardware behaved like F*ck*ng Magic rather than the backlash and hysteresis of Actual Machinery. So, and with last week's Murphy's Law comments in mind, I want to point you at a very non-AM/FM short story that I was pleased to rediscover online: M.I. Mayfield's On Handling the Data.

That link is courtesy of Project Gutenberg; if it doesn't work, you might try the copy at manybooks.net. Perhaps I feel an affection for this story because of my time doing Chemistry. I once heard an anecdote about an inorganic chemistry practical given to new first-year students. This was a long and tedious titration of some barium compound, intended to verify the student's ability to manipulate delicate glassware without dropping it or emptying something toxic into their shoe.

At the end of the practical, the demonstrator collected the titration results; and as customary, plotted them on a histogram which he pinned up at the end of the lab. All students had the same solution to titrate, so in an ideal world, the histogram would have been a single vertical line. In our non-ideal world, one usually gets a Gaussian. Though because students don't trust their results, so "fiddle" them to agree with their neighbours' results, it tends to be an unnaturally narrow Gaussian. But in this histogram there were two peaks, separated by a wide trough.

For several days, the demonstrator couldn't work out why. Then he realised it was due to the geometry of the lab. There were two long files of benches separated by a wide aisle. The students up one side of the aisle had been mutually fiddling their data. And the students up the other side of the aisle had been mutually fiddling their data. But nobody had bothered to compare results across the aisle.