Comment by cratermoon | Hacker News Reader

Comment by cratermoon | Hacker News Reader

Parent Full thread

Counterpoint: all of those incidents, except Apollo 1 are proof that the engineering was great, because nobody died.

For example, you mention the computer on the Apollo 11 lunar module crashing. In fact, it was recovering and working properly. The astronauts had left the rendezvous radar on during descent, in case it was needed for abort. That was not a nominal configuration, and the radar kept stealing cycles and causing the guidance computer to be overloaded with tasks. Remember, it was a hard real time system. What did the computer do? Reset and prioritize the key task: landing.

Apollo 12: Got hit (twice) by lightning. The electrical system wasn't fried, it survived it, in a protective mode. Importantly, the computers in the Instrument Unit, placed on the third stage, were completely unaffected.

Apollo 15: One lost parachute, still landed safely (if a bit hard) because of redundancy.

I could go on, but you get the point. It was a well-engineered system backed by a team of engineers.

Maybe. But it's hard to tell whether nobody died because the system was robust vs. nobody died because we got lucky.

For example, there were several cases of burn-through on the O-rings before Challenger. The engineers thought there was enough margin to not worry about it, so they didn't

Similarly, when Columbia was hit by foam-ice on ascent no one worried because it had happened before and nobody had died.

Correction -- at least for Challenger, engineers did not think there was margin, and argued against the launch.

At the technical level, both tragedies were caused by design flaws. Organizationally and culturally, multiple factors contributed, but an attitude of "nothing has happened yet, so this is fine" (normalizing risk) was a major one.

Normalization of deviance is how my professor described it. At least it’s taught in school now.

In my mind I have stored this phrase that "production pressures move the Overton window of acceptable shortcuts closer to disaster."

I think it captures several important nuances, like how it's a gradual process, how it ecpnomises/improves things at first, that there is a destination, that it covers even things such as discussions about shortcuts and not just their usage.

We don't disagree about the engineering being excellent. I was commenting on safety culture. A few days ago I saw Tory Bruno explain with visible frustration how they canceled the launch due to a valve that had to be cycled before it behaved. In that environment, the Apollo risks would not have been tolerated, even though they turned out to have been good bets.

Reply on news.ycombinator.com