[e2e] Nature mag, DARPA, and the Internet

Dave Crocker dhc2 at dcrocker.net
Sat Aug 9 07:39:50 PDT 2003

JC> oh and the termal-host thing - well maybe an IMP was such a thing (as were early DEC
JC> networks as well as SNA ) but pretty soon, real peer-to-peer (in the network or LU sense)
JC> emerged 

Actually, the Arpanet -- including IMPs -- were notably *not* terminal
oriented. Neither was DECNet, except for the LAT application protocol.
They were pure "classic", "generic" packet-switching, roughly comparable
to IP, in terms of the style of service they offered client

This was in marked contrast, for example, to Tymnet, which had an
infrastructure entirely tailored to bringing character-based terminal
traffic to a central site. The techniques they used to optimize things
for this function were clever and efficient. There system highlights the
difference between tuning the infrastructure, versus making things

It's true that Telnet was the most popular mechanism, for many years.
But FTP and email (using the FTP commands for it) were also popular and
the Arpanet infrastructure did not make any distinction in the traffic,
or perform better or worse for any particular application.  Well, ok.
More data aggregation performed better.

The fact that the application protocols were designed with the intention
of permitting them to be typed from a terminal can be misleading. From
the standpoint of protocol architecture, distributed computing, and
other abstractions, there was nothing "terminal-to-host" about them.

Oh, and from the list of things being claimed to have not come from the
Arpa-funded efforts, we have to correct the reference to shared
printing.  The first public demonstration of the Arpanet included at
least one printer attached to the TIP (IMP with terminal h/w and s/w)
used in a shared fashion.  The feature was added by... The SRI-ARC team.
And Engelbart was at the event; it's when I met him.

Oh, and by the way, when Day says he was there, he means that he and his
crew at U. Illinois built the first Unix-based Arpanet implementation.
As I recall, that was around 1974, some years the Arpanet protocols had
been defined and implemented.

 Dave Crocker <mailto:dcrocker at brandenburg.com>
 Brandenburg InternetWorking <http://www.brandenburg.com>
 Sunnyvale, CA  USA <tel:+1.408.246.8253>, <fax:+1.866.358.5301>

More information about the end2end-interest mailing list