How passionate is the onboard shuttle software group about understanding its own mistakes? Consider Discrepancy Report #107743. It tells the story of a single error in a single character on a single computer display inside the space shuttle.
The error was found October 7, 1994, and, unlike most things involving the space shuttles computer code, its very easy to understand. Because of a software slip, an on-screen indicator of the position of the shuttles starboard manipulator arm shows up one character to the right of where it should — in column 18 instead of column 17. The error didn't involve any problems with the position of the arm itself, just its reported position on the computer screen the astronauts consult. The display coding for the port manipulator arm is correct.
A half-dozen pages of forms detail what version of the software the starboard-arm display error was introduced on, track its history through subsequent versions of the software, explain how it was discovered ("informal inspection"), and discuss why the error wasn't caught sooner. Mostly, the error resulted from confusion of coding for the left-side (port) manipulator arm versus coding for the ride-side (starboard) manipulator arm. The investigation suggests some minor changes so that in the future, separate code for right and left side equipment will be checked separately.
Patti Thornton, a senior technical engineer in the shuttle group, explains the most remarkable thing about the error and its paper trail. "There is no starboard manipulator arm," she says with a smile. "The code is set up to accommodate one, but there isn't one now, and there's never likely to be one."
The investigative forms take note of this, too, in language so dry its almost wry: "The probability for this problem is very low as the starboard arm is unused and there are no plans for its use." Even a phantom starboard manipulator arm had a small lesson to offer.
A version of this article appeared in the Dec 1996/Jan 1997 issue of Fast Company magazine.