Software runs the space shuttle. These guys write the software.

The average error rate for most software, over 11 revisions, and of comparable size and complexity, would be 5000.  The shuttle software, over the last 11 revisions, has had 17.

Little bit of a difference in numbers there (no, there are no typos, it really is 5000 VS 17).

How did they do it?  Step one was take out all the glamour.  The people who work there are business men and women; these aren’t the Jolt Cola and pizza fueled scooter riders we envision when we think of “the best” software companies.  Everything done has a plan.  Everything.

At one point, they wanted to have the shuttle track its location from GPS satellites;  this necessitated the changing of about 3, 500 lines of code.  The documentation for the change spanned 2500 PAGES.

Interesting article, makes some comments as far as where programming as a whole is heading (besides talking about some really cool stuff done by really stuffy and cold people)

(I was going for the joke by rearranging that sentence.  I’m sure they’re all wonderful people)

(Except Ted in accounting, he’s a dick)

(That’s right Ted, you’re a dick)