[ih] Why TCP?

[Noel Chiappa]

    > From: Jack Haverty

    > TCP/IP technology is fine, but there were numerous other ways to define
    > the technical mechanisms that could have evolved into The Internet. The
    > people driving all those technologies had to make a similar decision.
    > They all made the wrong choice, and their technologies have all but
    > disappeared.

My perception is that the single biggest reason that TCP/IP became the
protocols used on 'the Internet' (and I hate that some people don't get why
there's a capital on it - I look forward to the new US President being
installed in the white house) is not so much its internal technical goodness,
or even the openness, but other factors - the biggest of which, I think, was
the size of the inter-connected user base.

The point of a communication network is to communicate with people, and so
naturally people, when deciding which network to hook up to, will tend to
pick the one with the most, since it gives them the most benefit. (I think
Metcalfe has a Law about that?) And that tends to drive the smaller networks
out of existence, and the whole ecosystem toward one single Network To Rule
Them All. (The same force acted on the telephone network back when, I expect.)

The TCP/IP Internet, starting as it did with large chunks of the US academia,
and tech businesses (plus government and military, although they were
secondary), just had a lead nothing else could ever catch. And as time went
on, it got worse (feedback)...

(I expect the X.25 networks probably started with a large user base, but that
may be one place where technical capability _did_ play a role: TCP/IP worked
well over LANs, which were going crazy at that point.)

An ancillary factor was the wide variety of systems for which TCP/IP
implementations were available - and the two fed back between each other.

	Noel