[mitchty]

How so? If I build something that doesn't conform to specifications, what does it matter if it happened to be unit conversions? They could have skipped a decimal point and not noticed, because they didn't conform to the specification. More importantly it never got checked or tested or simulated further on up the chain to catch the issue.

Look at the Araine 5. Unit conversions had nothing to do with a computer saying "gimbal an engine 90 degrees".

I'm sorry but this canard of two units being the problem is missing the entire point.

Its an engineering issue and management failure. Just like the space shuttle O rings before them, space is hard. Distilling things down to "america should have used metric cause reasons" is trying to pass off complex problems with bite sized quotes. The only issue is it is wrong.

[mturmon]

This is a smart comment. It sucks to have to keep repeating this comment every time someone trots out this supposedly easy example.

Sure, the units issue was the root cause, but the resulting navigation errors should have been acted upon. They were noticed ("The discrepancy between calculated and measured position, resulting in the discrepancy between desired and actual orbit insertion altitude, had been noticed earlier by at least two navigators, whose concerns were dismissed."), but the issue wasn't acted upon. This is a management and budget problem.

Even if the units are given as SI, there are always interface issues -- which coordinate system to use, where the origin is, does z point up or down, degrees or radians. It can be hard to appreciate unless you have had experience with hard-to-find failures due to subtle numerical problems.

As you say, space is hard.