This is also the reason massive software rewrites often fail; you rebuild the general gist quite fast, nice code, lovely interfaces etc but it will have a trillion bugs which come from decades of adding an exception here, adding one there etc. So now you have a beautiful albeit worthless product. And often these get scrapped: I know of some tax system rewrites from mainframe to modern code that costed 10s of millions and were scrapped, multiple times for this reason.

"Know-how rot" is a thing. Actual people involved are literally dying off, and knowledge is being lost. It could have been a peak, the Apollo program.-

Even if the drawings are there, they are often not as good as you may hope. It’s not uncommon to have missing specs, part numbers etc. Modern engineers have to fill in the gaps and sometimes build one-off small parts when the OEM is no longer in business (or no longer interested in building small quantities of some part)

We need more mass to orbit, more time on orbit, larger spacecraft volumes, less expensive mass-to-orbit, and less toxic materials in order to do the new missions that we'd like to do.

> After extensive tests and analyses, Boeing engineers concluded the helium leaks were the result of slightly degraded seals exposed to toxic propellants over an extended period....

> The thruster problem, testing indicated, was caused by high temperatures that, in turn, caused internal Teflon seals to deform in poppet valves, restricting the flow of fuel.

> The high temperatures, the engineers concluded, were largely the result of manual flight control tests that caused the jets to fire hundreds of times in rapid-fire fashion while the craft was oriented so those same jets were in direct sunlight for an extended period.

> The most obvious reason why productivity remained high after World War II, despite the end of the military emergency is that technological change does not regress. People do not forget. Once progress is made, no matter under what circumstances, it is permanent.

Robert J. Gordon, The Rise and Fall of American Growth, Chapter 16, p. 550.

Emphasis in original.

The best counterexample is the apocryphal quote, since adopted by many programmers: "When I wrote that passage, there were two who understood it.—God and myself. Now, alas, God alone understands it!"

<https://quoteinvestigator.com/2013/09/24/god-knows/>

I've certainly been in that circumstance with my own code multiple times, as well as at organisations which no longer knew why, or how, certain things were done.

And people, cultures, do forget. The whole notion of "lost wisdom" or "forgotten knowdge" is legion (though unfortunately neither seems to have a Wikipedia article yet which I can conveniently cite here).

Psychology has the notion however of a "forgetting curve", based on the work of Hermann Ebbinghaus, and I'm pretty sure that there's a similar notion which applies at a cultural and/or social level:

<https://quoteinvestigator.com/2013/09/24/god-knows/>

A related concept, and one that's given me pause for thought for some time, has been what the minimum sufficient level of continuation of cultural knowledge is, especially given the distinction between explicit and tacit knowledge. The former is easy to codify and transmit via text or speech, the latter is hard to express, and requires direct experience, personal training, or otherwise greater-than-verbal or greater-than-textual instruction. "You can't learn to swim from a book" is one expression of this.[1] In a technological society which becomes ever more specialised, at what point is there no longer sufficient transmission of a concept for it to be considered still "remembered"? How many crafts today have only a single practitioner? Perhaps retired?

(I've posted previously about the art of scientific glass-blowing, which seems headed down this forgetting path.)

________________________________

Notes:

1. Though of course there are exceptions. Theodor Kaluza taught himself to swim aged thirty from a book: <https://en.wikipedia.org/wiki/Theodor_Kaluza#Personal_life>.

> Did we have a higher risk tolerance back then?

NASA lost enough people during their "higher risk tolerance" epoch that they don't want to go through it again. Challenger and Columbia were such huge traumas each time...

How is this old style? At least in software development / IT-consulting in Europe cost plus is done all the time now, since there were too many legal battles about fixed price not being finished according to spec.

It seems rather reminiscent of their other problems of doing a hack on the 737 Max which caused the crashes and ignoring safety inspections which caused the panel to blow out.

It's kind of interesting if you see Musk talk about the rockets design it's evident he pretty much understands it all and controls the money and so is able to make sensible decisions to do this, don't do that - see this 2021 vid for example https://youtu.be/t705r8ICkRw . With Boeing you get the impression you don't really have that and you have an accountant type saying your budget for this is $x without undertanding they engineering they are paying for.

https://qz.com/elon-musks-spacex-and-tesla-get-far-more-gove....

Is that not “cost-plus?”

I wonder if the fear suppressed the natural behavior of Jim McNerney and others, keeping them from sowing division within the company because of the perceived danger from outside?

Here are some other examples, where today these might sound like an excuse for aggressive business behavior, instead of a unifying mantra --

Andy Grove, "Only the paranoid survive"

Dick Cheney, "Principle is okay up to a certain point, but principle doesn't do any good if you lose."

(Yes, yes, he's outing himself as unprincipled; I'm quoting him intentionally since people were still being people all throughout; does this show the larger populace wide zeitgeist?)

John F. Kennedy, "We choose to go to the Moon in this decade and do the other things, not because they are easy, but because they are hard."

I think it's interesting how the same thought can sound different when my mind is set in different contexts.

Software from 50 or 30 years ago couldn't solve some of the problems current software is solving because the hardware wasn't powerful enough. Example: we had neural networks theory and a little practice but without the current hardware we could not work on them and advance the theory. Could we be confident that LLM could be as good as they are? We could not.

However let's go back 20 years or maybe 30 and more mundane tasks like word processing and spreadsheet were already a solved. We needed only to port them to newer versions of the OS. A Word from around 2000 would be OK for most tasks today.

Basically an IQ test, if you're lucky to never have been subjected to one, is in fact multiple different tests (for me it was either 4 or five, I don't remember, I know I scored under the mean in the temporal-spatial test, barely above in the verbal, and way above in at least 2 others). It's a 'proxy' for intelligence but, but an incomplete one, and different tests exists. According to Flynn (2012), the gains were mostly with performance-based mesures, and the gains in verbal tests (which highlight education, and imho, culture) were low or negative. The reverse-Flynn could very well be caused by a decline in crystallized intelligence (general knowledge subtest, vocabulary/verbal subtests), while performance-based tests (logic, adaptibility, spatial vision) plateau or stopped increasing. In any case, we should avoid making broad claims about IQ: tests aren't _really_ standardized, vary between countries (and psychologists tbh) and are a very complex subject.