[remarkEon]

I know this point is made a lot but I’m not sure it’s that meaningful.

WW2 boats were not the enormously complex network and sensor systems that today’s are. While, yeah, 10 years is extreme, but I don’t think 39 days is physically possible.

[bryanlarsen]

The 39 day boats built in WW2 were probably more similar in complexity to a modern boat than a WW2 reactor is to a modern reactor.

[yamtaddle]

It was Liberty ships—slow, high-capacity, not especially durable cargo ships built as fast as possible, assembly-line style. The median construction time only got that low after the assembly line really got going and they were able to work out some kinks. It was over 200 days early on.

Warships—none of the capitals ships, certainly—weren't built that fast. Not even the cheap-as-hell escort carriers we built tons and tons of, I don't think. Destroyers and little corvettes and other smalls screen and utility ships, maybe some of those weren't too far from 39-day construction times.

[eternityforest]

I would be surprised if some of the systems were actually harder to work on back then. Old battleships always seemed to have insane amounts of wiring and cabling, because they used the nightmarish one signal per wire protocol in the pre digital age.

[remarkEon]

> because they used the nightmarish one signal per wire protocol in the pre digital age.

Can you say more about this? I find that kind of thing fascinating.

[eternityforest]

https://softsolder.com/2017/11/05/battleship-wiring/

They presumably used to have to run an individual wire for every. single. signal, or at most, they'd have extremely simple multiplexing.

I have no idea what these cables do, but I'm glad I'm not the one repairing it. Not only is there physically a lot of wire, 90% of it looks to be going directly to something mechanical.

I don't know how modern ships do it, but with civilian digital logic , you'd probably run a few redundant ethernet(Or fiber, or CAN) links to some kind of fan out box right near the switches or motors.

Cars these days are very extreme about it, sometimes even running the entertainment system commands over the fairly critical CAN bus for some reason.

If one of those cables went bad, spanning tree protocol would save you till you fix it, and probably tell you which cable broke and maybe even where. Mix them up, no problem, firmware knows what to do.

Plus, everything goes through a computer, so you've got logging and all kinds of stuff to help spot issues, rather than the classic "Oh well, noise and intermittent connections, we live with it till it gets bad enough to be all the time".

The only rats nest is right from the mechanical switch to your PLCish thing, but that's short enough to trace by eye.

You might have zero concern about interference in cables, because they might be fiber.

You won't have complicated color code schemes or wiring diagrams, or cut off wires that nobody knows where they go to anymore, you can just test it, because you don't have 50 thousand things to sift through.

People argue about whether digital or analog is more reliable, but digital is definitely easier from a hardware perspective, the hard stuff is done in factories by robots.

I'm sure the technicians had special training and it was a lot better than average civilian industrial uses of the pile o wires system, but I would think it would still be time consuming and not paralellizable, only so many people can fit around one of those giant bundles at once.

I could also be vastly misinterpreting how modern ships are wired, I know some industrial setups today insist on home run wiring instead of condensing to digital.

[remarkEon]

That picture makes “cable management” take on a whole new meaning, my goodness.