"Aerospace is plumbing with the volume turned up" -- John Carmack

I think somewhere in this sci.space archive of rocketry usenet wonders is a quote along the lines of "Rocket science is just extreme plumbing". Cryogenics with powerful oxidizers is hard!

https://yarchive.net/space/index.html

"All engineering is plumbing - some fields just manage to avoid working on broken sinks & toilets" -anonymous

Even the arts aren't necessarily safe. A story told by the composer Philip Glass in 2011:

> While working, I suddenly heard a noise and looked up to find Robert Hughes, the art critic of Time magazine, staring at me in disbelief. ‘But you’re Philip Glass! What are you doing here?’ It was obvious that I was installing his dishwasher and I told him I would soon be finished. ‘But you are an artist,’ he protested. I explained that I was an artist but that I was sometimes a plumber as well and that he should go away and let me finish.

My day job is plumbing. I really very badly want to be able to make a decent living through the skills I've been practicing outside of my job, and this quote is motivating to me. Thank you! :)

There's some serious chemistry, thermodynamics and fluid dynamics involved as well.

I suppose someone is thinking that fluid dynamics is plumbing, but it would be a stretch to make that claim over issues like the size and shape of the combustion chamber, its throat and the bell, which are dependent on the characteristics of transonic and supersonic flow.

I guess one glaring difference might be that they got this right when it mattered. Not a single Saturn V ever failed during launch. The aerospace engineering culture is vastly different than the "engineering" that goes on in ML/AI, and software development more generally. I have a lot of respect for it.

Saturn V suffered several engine failures. It had persistent pogo oscillations that put Apollo 6 into the wrong orbit and forced much of its test plan to be canceled, and nearly destroyed Apollo 13 during launch. Vibrations severely damaged Skylab during its launch. I would not say it never failed. It never went kaboom in the sky, but there are other ways to fail.

It was an exaggeration, but the comment's underlying point is spot on: the "engineering" that we do doesn't hold up in terms of rigor or cleverness that AE and other related engineering disciplines exhibit.

Counterpoint: if we tried to build all software using the same rigorous and "clever" processes that aerospace engineers do, almost nothing would ever get built. It's a different process because it's a different field, not because aerospace engineers are fundamentally more clever.

Signed, Not an aerospace engineer ;)

Have an aerospace engineer cow-orker; one favorite quote for why he works in enterprise software is, "Because this flies and that (pointing at an Ares-1 mockup) doesn't."

On the other hand, there's nothing "engineering" about software engineering.

Hahaha, harsh. I like your coworker.

And I would tend to agree with your second point. We might occasionally borrow some formalism and organizational structure from our engineering cousins, but it's different in enough ways that it probably shouldn't bear the same moniker.

While I generally agree with that underlying point, if our computers were as unreliable as the Saturn V, we’d never get anything done with them.

The cost of failure is higher in aerospace. But I must say that in my job, when you manage to make your modified code pass all the thousands of unit tests and see the benchmark results increase and the 80 cores of the test machine firing up, feeding GB/s of data in their pipeline at the optimal fillrate, there is something of of a "ignition aaaaaand liftoff!" moment that is deeply satisfying.

I am just allowed to try pressing the red button several times a day instead of once every two years.

80 modern processors have around four orders of magnitude more "moving" parts than a Saturn V, according to what I can Google. And the Saturn V, it is said, had a particularly large number of parts because it was hand-made to an extent that modern machinery is not.

Comparing the number of parts underestimates the complexity of a computer system, of course, because the software is a machine with a vast number of parts, on top of the hardware.

In other words, I'm on the side of those who say software is hard to get right for good reason, and not because the practitioners are less competent.