I could see managers at NASA seeing their role as a union or labor leader, trying to keep their budget and their people employed at least as much a priority as safety. So they're willing to take more risks because they're in a position of having to balance an equation.

And it may be that the flaw is this uni-dimensional hierarchy that we seem to like, that seem efficient most of the time, but in reality not only miss edge cases but can cause them. The managers simply did not always 100% trust the engineers. Ultimately, the people with the power over go/no-go were more managerial than engineers, were more concerned with the politics, cost, perception of delay than they were at risk assessment.

All spending of money in any company or government is basically the same: someone at the top delegates the vast majority to someone else. And they choose that someone else based on trust. And the longer than chain is, the greater the likelihood someone is put in position where the chain of trust is just flawed. And you hope it gets exposed where people don't die, obviously. But that didn't happen here.

And that's why a bunch of people got angry at Feynman as well as Boisjoly. They exposed that the wrong people were making key decisions. Had they been the right people, then loss of life wouldn't have happened. Anyone that wrong is going to be angry, it's a personality flip of the coin whether they get angry with themselves for their mistake or with others for exposing it.

There was pressure to ignore warnings because President Reagan was scheduled to be present for the launch. In higher levels of management, things get more political. Looking good and getting exposure count for a lot. No one wants to be the guy that made nasa look bad by slipping the launch date. In tech, most managers tend to be rated by how well their projects meet schedules and how much operating expense they can minimize, since that's the only thing their managers have any sense of control over. Your boss doesn't want to take the heat for delaying project "glitter unicorn" by a week so you can fix 2 major security holes in the service, one which allows customers to login without a password. And this is why your company won't spend $500 for a data backup system or why it takes 4 months to get your broken keyboard replaced. It's turtles all the way up.

I can totally see this - I mean, in my job management generates major stress for everyone to get a trivial product update shipped on time. Imagine the pressure to get a space launch to 'launch' on time. Makes our system seem like General Mao's. But better of course.

> There was pressure to ignore warnings because President Reagan was scheduled to be present for the launch.

I never heard that before. Is there a source?

I was wrong--he wasn't going be attend, but there (allegedly) were plans to have a televised conversation between him and Christa McAuliffe during his State of the Union address that night.

However this may not be true. While trying to find an e-source, I read that there were some politically-motivated rumors that the white house ordered the shuttle to launch for this reason. Feynman investigated that rumor and didn't find any evidence that the white house ordered the launch.

It's possible the detail about an on-air tv conversation between the president and a schoolteacher astronaut could have been fabricated as well to make Reagan look bad, or it could have been based on elements of truth. For example she was expected to broadcast 2 lessons to students from space, so the capability definitely existed. It doesn't seem too far out that they would have liked for this conversation to happen during the State of the Union, though that doesn't mean the white house ordered the launch.

I read this in Edward Tufte's essay Visual and Statistical Thinking (http://www.edwardtufte.com/tufte/books_textb), which discusses how better information displays might have convinced NASA management to postpone the launch. It also appears as a chapter in his book Visual Explanations.

I didn't have any luck finding sources on the internet (lots of keyword overlap). Tufte is much better informed than pretty much all of us, though it's possible he printed what amounts to an unsubstantiated rumor. FWIW he also served on the Columbia accident investigation board.

I think there were other pressures to launch as well. The launch date had been postponed 3 times already for various reasons. Cynical danek: I wouldn't be too surprised if some manager's performance review was based on how many launches took place on the scheduled date.

> It's turtles all the way up.

That's just perfect

That's a good point. It is easy to just say that the others were incompetent without knowing the rate of false positives ie. how frequently were such memos written talking about catastrophes. The issue of rate of false positives vs true positives (i.e.,real problem) is something that most engineers and managers don't have a good grasp of.

this is true for this early warning from the manufacturing company, but the launch day reasoning was chilling. I'll be back with links to the worse parts.

https://news.ycombinator.com/item?id=9277408

> we look at this memo with perfect 20/20 hindsight

On a tangential note, I've always wondered about this sort of idiom. 20/20 vision is the equivalent of a 100 IQ -- it is equivalent to that of a "typical human". Perfect vision would be 20/ε (you can see at 20 feet what an ordinary human could only see from an infinitesimal distance), and mundane glasses will often correct vision past 20/20.

It seems like such a low bar for hindsight to meet.