There's a bunch of good stuff on YouTube from the folks who actually developed the Internet. Here's a couple from Leonard Kleinrock, whose MIT thesis laid out

There's a bunch of good stuff on YouTube from the folks who actually developed the Internet. Here's a couple from Leonard Kleinrock, whose MIT thesis laid out much of the math behind packet switching and whose lab sent the first Internet message: https://www.youtube.com/watch?v=vuiBTJZfeo8 https://www.youtube.com/watch?v=vuiBTJZfeo8 https://www.youtube.com/watch?v=rHHpwcZiEW4 https://www.youtube.com/watch?v=rHHpwcZiEW4 And from Bob Kahn, who designed the router of that early Internet (interviewed by Vint Cerf, who invented TCP/IP): https://www.youtube.com/watch?v=EKxNMTVnBzM https://www.youtube.com/watch?v=EKxNMTVnBzM https://www.youtube.com/watch?v=hKZ6tJcQpcI https://www.youtube.com/watch?v=hKZ6tJcQpcI A key innovation of the Internet was packet-switching. Previous networks like AT&T's telephone system were circuit-switched: the configuration of the network, and route between source and destination, is an inherent property of the network, and once a connection is established it can't be easily reconfigured. Packet-switching makes the source and destination a property of the message, and then the network is responsible for figuring out a route from source to destination. Notably, because all information needed to specify the destination is included in the message, it can be retried or take a totally different route. Most things have multiple causes, and the Internet is definitely one of them. The scalability and distribution properties were certainly one of them: a centralized system like the telephone network cannot scale to new uses and many new endpoints the way a distributed system like the Internet can. According to Kleinrock, the need for management to keep an eye on all the research they were funding was apparently another one of them. But given that it was funded by DARPA, the resilience of a packet-switched network to scenarios where individual circuits might go down was probably a major reason for the interest in this technology. It doesn't necessarily have to be a nuclear strike, but there were a number of scenarios of interest to RAND and DARPA that could involve a portion of the nation's communication network being disabled and still needing to get messages through. This is also a good lesson to designers of future networks and computing systems. The end-to-end principle remains as valuable to system designers today as it did in the 1960s.