Just because TCP/IP was made to be robust enough to survive holes in the network topology caused by nuclear bomb devastation, it doesn't really mean that TCP/IP networks would magically keep on working after a large-scale nuclear war.
I mean, during a nuclear war, the networks would only need to stay working until a counterattack could be coordinated and launched.
After that point, the devastation would be so large planetwide that I think any potential TCP/IP users roaming the irradiated wastelands looking for food while coughing up blood would have other things on their mind than network access in any case.
For smaller scale events, sure, but while the networking design might be robustness-focused, there is still a huge amount of auxiliary stuff required, these need power and cooling, maintenance and administration, repairs, and so on.
Some parts of the internet system might be robust and resilient, but the whole system certainly isn't.
I think we can just assume that in a nuclear war or major EMP event the Internet will be one of the first things to go.
But this is one of the best uses of a Raspberry Pi. A Pi with an external USB drive and a small display will run off solar battery power, but still have enough storage for a complete Manual for Rebuilding Everything.
Add some EMP protection [1] and update the Manual from GitHub once a month or so, and you're set.
Indeed, a Raspberry Pi that could presumably work fine for the short term.
However, I don't think it'd work for the long term given a suitably large catastrophy. This may sound like a No True Scotsman argument so let's just for the sake of argument define "suitably large" as "permanently disrupts the current industry base for things like chip fabrication facilities".
Once the tiny computer and/or its peripherals fail for whatever reason, such as simple wear and tear, that's it -- no new parts will be available for the more advanced components. Spare parts could be scavenged, but even that is a limited resource, and will just push forward the moment when the reference library goes away. At that point it had better be either useless or fully memorized.
So I think books are better for the longer term. Of course, books are brittle too in different ways, but they don't need to be paper books... A book could be printed on Tyvek like some hiking maps are; Tyvek doesn't mind water and can be folded and doesn't rip so easily.
Please note I might be biased since I really love books :)
I'd argue that the digitized versions would likely last several years. After a catastrophe stuck, one of the instructions after securing immediate survival would be to transcribe the contents to whatever media is available for longevity. Likely entire libraries of books. Otherwise you'd need to store and maintain those entire libraries everywhere prior to the event, which would be harder to accomplish and keep safe.
it would be interesting to find out just how much would survive.
ok, maybe not "find out" but... speculate?
Cellular networks have been resilient because they're possibly distributed mesh networks. I'm pretty sure the carriers have some gurus who officially or unofficially have figured out what happens.
DSL/cable type networks might be different.
I'm pretty sure dns and similar systems are widely distributed.
I wonder if a proper analogy might be the roadways. Even though one road or freeway might be blocked, people will be able to drive around and (maybe) get through. Who knows, maybe there are some "blocked mountain crossing" equivalents with the internet.
On a personal level, a laptop, a cloned git repository and a solar panel. :)
I mean, during a nuclear war, the networks would only need to stay working until a counterattack could be coordinated and launched.
After that point, the devastation would be so large planetwide that I think any potential TCP/IP users roaming the irradiated wastelands looking for food while coughing up blood would have other things on their mind than network access in any case.
For smaller scale events, sure, but while the networking design might be robustness-focused, there is still a huge amount of auxiliary stuff required, these need power and cooling, maintenance and administration, repairs, and so on.
Some parts of the internet system might be robust and resilient, but the whole system certainly isn't.