[ih] NCP (was: Separation of TCP and IP)

[Noel Chiappa]

    > From: Steve Crocker

    > I coined the term Network Control Program, and hence the acronym NCP,
    > to refer to the software that needed to be added to the operating
    > system of an Arpanet host. ... as you've just seen, I used the term
    > Host-Host protocol to refer to the protocol.
    > Over time, the term Host-Host protocol was too bland

Or too long! :-)

    > the need to explicitly refer to the implementing software waned. People
    > repurposed NCP to refer to the protocol. ... I think it will help
    > readers to understand the two different but related uses of "NCP"
    > during that period.

A very good point.


I wasn't there, obviously, but I now have what I think is a very plausible
theory about what happened. The ARPANET 'stack' is composed of four main
layers:

- 1822/VDH at the bottom, to transfer bits/'messages' between the host and
	the local IMP 
- Host-to-IMP Protocol, to transfer messages between the local host and a
	distant host
- Initial Connection Protocol/Host-to-Host Protocol, to provide host to host
	connections
- Applications (Telnet, FTP, etc)

Each of these layers/units had a well-defined name. However, there was no
official name for _the stack as a whole_. Which wasn't such a problem,
initially; but once IP/TCP appeared (and note the date of that first RFC that
I found which uses 'Network Control Protocol'), there was a need/use for _a_
term for the 'NCP' stack. Organically, 'NCP' seems to have been adopted to
fill that role.

Of course, the original formal expansion of 'NCP' (as 'Network Control
Program') made no sense in this new use, so a new 'backronym' (as the term
goes) of 'Network Control Protocol' _later_ appeared. (So I guess Ms. Hafner
is innocent! :-)


I will fix the NCP page on the Computer History wiki to explain the two
meanings of the term 'NCP'. If I get energetic, I'll do Wikipedia too -
but only if nobody else deals with it! :-)

	Noel