This brings to mind an incident during a rocket launch in 1981. The ground guidance computer hiccuped, just once, while running the mission trajectory a couple of weeks before the launch. It did not repeat, and the computer scientists told the rocket scientists that this sort of thing would occur once every two hundred hours of operation. Then days later it hiccuped again. Well this was great, it appeared. Two such errors had occurred in the space of less than a week, so under the original explanation we had at least 400 hours to go before it happened again. The computer failed big time on launch night, spewing data all over the screens as it did a core memory dump, after an entirely successful test run.

The original assessment of “it’s just one of those two hundred hour hiccups” did not apply because the machine was broke. And two such random hiccups in a row does not mean that you are extraordinarily lucky and can expect a long period of flawless operation – it means that the machine is broke, busted, Tango Uniform, or at least soon will be.