re: Bang! Old Email, Usenet and the End of the Cold War VIEW POST

FULL DISCUSSION
 

In 1981, sending an email could take days.

Or so says an unverified wikipedia source.

Primary source, here: sending email could take weeks. The longest I specifically remember was an email that took just shy of three weeks to arrive. This would have been 1989/1990 timeframe. The particular email exchange in question was between me (@PSUVM.PSU.EDU) and a friend going to school at Berkeley.

When you wanted to send an email, you needed to know every hop skip and jump that you were sending through.

Not quite accurate. You simply needed to know the path to the biggest exchangers. Fortunately for me, @PSUVM@ was immediately adjacent to @PSUVAX@ (acutually a Sun 3/80, by that point) and my friend's domain was only two hops off the main Berkley exchanger (@UCBVAX@?).

That said, if you did specify a full path, you might get your email to transit faster than just leaving it up to the main exchangers. Plus, sometimes, you just wanted to see how many nodes you could bounce an email through "just because". Kind of the computer equivalent of filling up your passport with stamps.

If I wanted to send my mail from Stanford to MIT, I might need to specify that it's got to go through 🚀 NASA AMES, Moffett Field, UC Berkeley, Area 51, and University of Illinois.

If you were lucky enough to be on/near a main exchanger, your paths could be short. If, as was more common, you were on some BBS six+ hops away from a main exchanger, your bang-path could be multiple lines long.

Still, UUCP was better than trying to route via FidoNet or its peers. And Being on UUCP or FidoNet was better than trying to route between the two.

Ironically, the best (most timely) way to communicate, at the time was via BITNET RELAY or, later, IRC. Only real problem was those mediums had no "memory". Your communication partner had to be online to get your message. And, at the time (and BITNET was particularly subject to it), network partitioning was common - so, you'd lose your chat-link multiple times in a given chat-session. Similar for MUDing.

 

Primary source, here: sending email could take weeks. The longest I specifically remember was an email that took just shy of three weeks to arrive. This would have been 1989/1990 timeframe. The particular email exchange in question was between me (@PSUVM.PSU.EDU) and a friend going to school at Berkeley.

I heard that, wanted to put it in the article, but assumed people were exaggerating so I left it out. Did it really take that long? What was the slow-down exactly, just slow processing power and a lot of requests? Or were things not connected? This stuff is traveling at the speed of light effectively right?

Plus, sometimes, you just wanted to see how many nodes you could bounce an email through "just because". Kind of the computer equivalent of filling up your passport with stamps.

That is ridiculously interesting, I never thought about it that way.

 

What was the slow-down exactly, just slow processing power and a lot of requests? Or were things not connected?

UUCP was specifically designed to transit emails to/from through systems that weren't connected 24/7. It was a "store and forward" technology (even early Sendmail was designed to be able to operate this way). This was a point in time where connecting was fairly expensive and where links were slow (the Internet's "backbone" in the late 80s was sub-T1 speeds). Many nodes were dial-up based, only connected for a few hours each day, and, depending on how much traffic they needed to exchange at those slow speeds, my not be able to transit it all within the period they were online. Factor in that systems also weren't super reliable - being wholly offline for day or more at a time or might have an outage during their scheduled exchange windows - and it made for a really slow system.

Participating in USENET was a similar experience. It might take weeks for a message to fully propagate.

code of conduct - report abuse