[e2e] It's all my fault
jnc at mercury.lcs.mit.edu
Fri May 18 08:16:26 PDT 2007
> From: John Day <day at std.com>
Please excuse this last post, but I wanted to get in a plug for an
un-deservedly forgotten piece of our history: CYCLADES (see below).
>> what was unique about Baran's work was that he came up with the idea
>> of breaking up user's messages into smaller pieces, and forwarding the
>> pieces independently - something nobody before him had thought of.
> I have read Baran's reports
I'm not sure, from your comment, if you're disagreeing with the above
> I can't tell if he is describing packet switching as in the ARPANet or
> packet switching as in the CYCLADES.
I'm not sure if he had thought it through to that level of detail at that
> I am tending to the conclusion that Baran and Davies independently
> invented packet switching.
This is a complicated topic.
The problem is that Baran had published a lengthy paper summarizing his work
in the IEEE Transactions on Networking, a fairly significant open journal, in
March 1964, and an abstract of the IEEE ToN paper had been published in IEEE
Spectrum, which was of course very widely distributed (circulation about
160,000 in those days) in August '64. That was over a year before Davies'
work (starting in very late 1965). The question is whether some of Baran's
ideas had percolated at some semi-subconcious level - perhaps via a chance
conversation with a colleague - into Davies' thinking. We'll just never know,
and I suspect Davies himself couldn't know.
I liken this to the question of the influence of Babbage on the first
computers (circa '46-'47). Some people (e.g. Wilkes, IIRC) say "Babbage had
been forgotten by then, I certainly wasn't influenced by his work". The
problem is that there were others who were active in the field who did know
of Babbage's work (definitely Aiken, who explicitly recognized Babbage in
contemporanous writings), and the early people did all know of each other's
work, so it's hard to know what the subconcious/indirect influences of
Babbage's on the field as a whole were.
> But the kind of connectionless networking and clean separation between
> Network and Transport seems to have come from Pouzin. CYCLADES had a
> very clean distinction between CIGALE and TS which the ARPANet did not
Absolutely. CYCLADES is a mostly-forgotten - and *very* undeservedly-so -
piece of networking history, and it's quite clear that it's the single most
important technical pregenitor of the Internet.
The decision to make the hosts reponsible for reliable delivery was one of
the key innovations in going from the ARPANet to the Internet. For one, it
made the switches so much simpler when they didn't have to take
responsibility for delivery of the data - which they couldn't really, anyway,
as we now understand, according to the end-end principle.
When I was active on Wikipedia, I always meant to upgrade their CYCLADES
article, but I never got to it, alas...
> Networking happened when minicomputers come along and are cheap enough
> that they can be dedicated to "non-productive" work
> I hate to think what would have happened if at anytime from 1970 to
> 1990 if some crusading journalist had figured out all of the non-DoD
> activities going on the ARPANet/Internet and done an expose! The things
> we were doing! The waste of tax payer dollars!
I think there was some publicity, actually, but for some reason it didn't
make a big splash.
Amusing story: One thing that did make a bit of a hit was when some DoD
domestic intelligence (on Viet Nam protestors) was moved over the Internet;
the resulting newspaper headline (cut out and pasted on one of the MIT IMPs
for many years) was: "Computer network accused of transmitting files"! :-)
More information about the end2end-interest