Louis Pouzin may be my new computer networking hero. His writing (in English, not his native language) just jumps off the page, and he always includes political analysis with his technical analysis. He understands that you can't do one without the other.
Here's an OCR translation of the attached photograph of text: https://gist.github.com/dariusk/c6ed0c0b8f6a1f2e0c8d2686a447b94b ("Virtual Circuits vs Datagrams", 1976)
Paper on Pouzin as a lost figure in networking history: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.692.1974&rep=rep1&type=pdf
It's really wild to me that there was a debate between whether we should have a datagram-based system (independent packets moving around a network that are routed to their destination as best as possible) or a virtual circuit system (where the network "clears out" and reserves a specific physical pathway for a connection between two computers). I mean obviously these decisions needed to be made, and early ARPANET IMPs used virtual circuits extensively, so. Yeah.
Oh here's a link to what I believe is the full Pouzin paper: http://delivery.acm.org/10.1145/1500000/1499870/p483-pouzin.pdf?ip=22.214.171.124&id=1499870&acc=ACTIVE%20SERVICE&key=70F2FDC0A279768C%2E3D342327617A783A%2E4D4702B0C3E38B35%2E4D4702B0C3E38B35&__acm__=1560278305_9bc361ec9192c0a56aea1cce2913cb5f (though I'm at a university right now so maybe only I can see it from behind this network)
@darius I remain unconvinced that one was decided in the right direction. :-)
@darius: have you read this?
@slightlyoff No but I just bought it!
@darius: I assign it as required reading to the standards folks here; tells a lot of this story, along with some colour on how/why IBM played it the way they did (as a standards pro, not how I would ever want to operate, but respect the *hell* out of their discipline).