A classic software development story

Back in the early 70s, NASA engineer tests on a part indicated that a joint with 2 O-rings was too wide and could expose the o-ring. Northrop Grumman and NASA's project manager said it was fine, 2 o-rings meant one was redundent right? and the design made it into the solid rocket booster.

Then in 1977, a different test indicated 1 oring was letting gas during certain levels of mechanical stress. The engineers proposed a solution, which was ignored.

Then in 1980, they asked to test what would happen if 1 oring weren't there and what would happen if the oring was cold. This was denied.

Then in 1981, a return booster was inspected and they found soot between the orings and one eroded, and the problem was added to the critical issues list. And ignored.

This happened again in 1984.

In 1985, they realized when the oring was cold at launch, the problem got way worse. Northrop Grumman finally changed the design to fix it.

But they had a bunch of the old, unsafe part laying around, and NASA didn't want to miss deadlines, so in January of 1986, they launched a shuttle with the part that they knew was unsafe in cold conditions, coldest morning they'd ever launched and a middle-school class watched a live stream of their teacher exploding 10 miles in the air.

I believe they are referring to the Challenger lunar launch that exploded. What they believe it came down to was a tolerance issue in the O-Rings they were using which if I remember right was a concern already placed by the engineers.