I remember enjoying the very early ARPAnet (about 10-15 hosts) when a frosh in 1972 at Harvard (I had talked my way into taking grad courses as a freshman since all the undergrad CS courses were pretty babyish, so had access to the grad computing center--nirvana). Back then, the ARPAnet was completely wide open, had only a few hundred users across those hosts in total, and was an absolute blast.
For example, every system had a well-publicized guest account you could log in with and poke around. Many of the systems were TOPS-10 and TENEX systems (both based on DEC-10 hardware), but some were IBM 360-based, and some were custom hardware-based. The most fun were things like SRI's (Doug Englebart's) NLS system, which was almost useable over the network with a CRT (which were pretty new-fangled themselves back then) and UCSB's symbolic math system (forget its name now). One could learn a lot by just poking around those early systems.
We used to telnet from host to host around the world, to see how many we could hop through before the connection crashed. (Plus, it was danged near impossible to "escape" your way back to any particular host to close the connection, anyway, since you'd have to remember how many escape/quote telnet sequences to send before the "real" escape you wanted to send.)
I was also on the first mailing list on the ARPAnet, hosted at BBN, which was, appropriately meta-ly, about mail software and mailing lists. That was a lot of fun, bikeshedding about header formats, email address canonicalizations, etc. (Back, then addresses were just user@host, no domains; I think mine was 67,377@harv-10. (TOPS-10 systems used octal programmer,project pair codes for login, each 18 bits).
And, back then, mail systems delivered mail by (pre-TCP) ftp'ing into your system and appending to your (world-write) mailbox file in your home directory. One could have lots of fun with that.
I remember sitting by the HARV-IMP (a Honeywell minicomputer with custom hardware, talking to HARV-10 and HARV-1 (a DEC PDP-1 that the first spacewar was written on), and the phone would ring in the middle of the night, asking whoever was around to reboot the IMP because it had crashed and they couldn't remotely reboot it. That IMP had a couple of high-speed 56Kb links to several other nodes, one down Mass Ave at MIT, and one out to the Air Force base out in Lincoln (?) MA. The super-high-speed cross-country links were 256Kb.
Perhaps the most surprising thing about most of us who grew up with the ARPAnet then Internet was just how much we took it for granted, never thinking a thing about it, especially not its commercial potential. It was just the air one breathed with mail, ftp, etc. (this is all pre-web), part of the fabric of life.
(I think Gates admitted to the same problem in his famous "we've gotta catch up with the Internet" memo.)
I remember doing some Lisp Machine consulting at HP Labs in the early 80's (I had become an idiot savant about MIT Chaosnet integration with DEC-20's while working for the MIT EECS dept), bumping into Len Bosak (then of Stanford) and hearing about his plans to start up a router company (Cisco), and being honestly baffled as to who (outside of a few military/industrial complex companies like HP) would buy a router? (I think it surprised Bosak as well.)
Not really, since anyone working at the school in 1969 has long retired or moved on to other things, mostly forgotten by now. A lot changes in forty years. I wrote a thing about why UCSB even ended up as the third node: http://jeweledplatypus.org/cgi-bin/blosxom.cgi/text/ucsbnet.... - they pretty much just happened to have the budget for it and a person working on interesting related stuff.
I remember enjoying the very early ARPAnet (about 10-15 hosts) when a frosh in 1972 at Harvard (I had talked my way into taking grad courses as a freshman since all the undergrad CS courses were pretty babyish, so had access to the grad computing center--nirvana). Back then, the ARPAnet was completely wide open, had only a few hundred users across those hosts in total, and was an absolute blast.
For example, every system had a well-publicized guest account you could log in with and poke around. Many of the systems were TOPS-10 and TENEX systems (both based on DEC-10 hardware), but some were IBM 360-based, and some were custom hardware-based. The most fun were things like SRI's (Doug Englebart's) NLS system, which was almost useable over the network with a CRT (which were pretty new-fangled themselves back then) and UCSB's symbolic math system (forget its name now). One could learn a lot by just poking around those early systems.
We used to telnet from host to host around the world, to see how many we could hop through before the connection crashed. (Plus, it was danged near impossible to "escape" your way back to any particular host to close the connection, anyway, since you'd have to remember how many escape/quote telnet sequences to send before the "real" escape you wanted to send.)
I was also on the first mailing list on the ARPAnet, hosted at BBN, which was, appropriately meta-ly, about mail software and mailing lists. That was a lot of fun, bikeshedding about header formats, email address canonicalizations, etc. (Back, then addresses were just user@host, no domains; I think mine was 67,377@harv-10. (TOPS-10 systems used octal programmer,project pair codes for login, each 18 bits).
And, back then, mail systems delivered mail by (pre-TCP) ftp'ing into your system and appending to your (world-write) mailbox file in your home directory. One could have lots of fun with that.
I remember sitting by the HARV-IMP (a Honeywell minicomputer with custom hardware, talking to HARV-10 and HARV-1 (a DEC PDP-1 that the first spacewar was written on), and the phone would ring in the middle of the night, asking whoever was around to reboot the IMP because it had crashed and they couldn't remotely reboot it. That IMP had a couple of high-speed 56Kb links to several other nodes, one down Mass Ave at MIT, and one out to the Air Force base out in Lincoln (?) MA. The super-high-speed cross-country links were 256Kb.
Ah well, sic transit gloria mundi.