Correction of my last two comments "the net"

[David Farber <dave () farber net>]

Begin forwarded message:

From: Steve Crocker <steve () shinkuro com>
Date: September 12, 2009 7:08:34 PM EDT
To: "David P. Reed" <dpreed () reed com>
Cc: Steve Crocker <steve () shinkuro com>, Richard Bennett <richard () bennett com 
>, Dan Lynch <dan () lynch com>, Dave Crocker <dcrocker () bbiw net>, Dave  
Farber <dave () farber net>, ip <ip () v2 listbox com>, Gordon Peterson <gep2 () terabites com 
>, John Shoch <shoch () alloyventures com>, Harold Burstyn <burstynh () iname com 
>, Lauren Weinstein <lauren () vortex com>, Paul Robichaux <paul () robichaux net 
>
Subject: Re: Correction of my last two comments [IP] "the net"

As the person coined both "Host-Host Protocol" and "Network Control  
Program (NCP)" I'll chime in briefly.

As you point out, the original designation of the connection oriented  
protocol between hosts on the Arpanet was "Host-Host Protocol."  In  
describing the protocol, we focused attention the need to implement  
new software within the operating system of each host.  That software  
would talk to the IMP and provide access to the net to (usually)  
multiple users on that host.  I called that required piece of software  
the Network Control Program, abbreviated NCP.

The term "Host-Host Protocol" was too generic and didn't provide  
enough of a cognitive distinction from other protocols, and the  
designation "NCP" got adopted as a replacement for "Host-Host  
Protocol", and, of course, it was naturally to redefine the acronym to  
mean Network Control Protocol.  Though I did find it bemusing to see  
the shift in terminology, I'm not inclined to denigrate that process  
by calling it slang.  At the end of the day, it's a question of what  
works, i.e. which terms take root and which ones don't.

Now, on the other hand, I have always thought TCP was indeed the  
replacement for the Host-Host Protocol.  They serve the same purpose,  
or at least were intended to, viz reliable, end-to-end stream on which  
various application protocols would rest.  It obviously wasn't a drop  
in replacement.  Rather, it was a competitive, incompatible, new  
protocol, but it occupied the same position in the protocol stack.   
Hmm... Not unlike IPv6 re IPv4.  TCP also rested on top of the new IP  
layer, which had no counterpart in the original Arpanet.

Steve




On Sep 12, 2009, at 3:36 PM, David P. Reed wrote:

> Oops.  Vint Cerf pointed out to me that in fact NCP is the wrong  
> term for what I'm referring to.  The historically correct term was  
> "RFC1822 host-imp protocol."  [I should be more careful - my excuse,  
> which is not a reason, is that I was NOT involved in developing  
> ARPANET protocol stacks (I used them) - the Multics one was done  
> while I was working on my S.M thesis on another aspect of Multics,  
> by my friends Ken Pogran, Doug Wells, and Deborah Thomas, as I  
> recall.]
>
> But we are observing an odd evolution of terminology here: NCP was  
> not a term used to refer to a protocol.  NCP was an endpoint program  
> that operated in each host.  It stood for Network Control Program  
> (NOT Network Control Protocol).  [Citation: ARPANET PROTOCOL  
> HANDBOOK, "Host-to-Host Protocol" NIC 8246, October 1977].  The  
> "Host to Host Protocol" was an out-of-band protocol used to set up  
> RFC1822 connections between hosts, by assigning and managing links.
>
> TCP ran over dedicated RFC1822 "links".  It did not, however, use  
> the facilities of the Host-to-Host protocol.
>
> Slang in the ARPANET community later started to use the term NCP  
> (incorrectly) to refer to the RFC1822 protocols when they were under  
> the management of the ARPANET Host-to-HOST protocol.
>
> This unfortunately is reflected in the Wikipedia entry for NCP!    
> Wikipedia is defining a "slang" term as if it were the "real  
> thing".  Historically (as of 1978) this is inaccurate.
>
> I apologize to the historians and to those of my contemporaries who  
> are probably embarrassed for me.  I'm embarrassed to not have  
> checked this out.
>
> However, my primary point absolutely stands:  TCP was not a  
> replacement for NCP.  Nor was it a replacement for the Host-to-Host  
> Protocol.  The Host-to-Host protocol became obsolete.  That is not  
> because TCP was substituted for it, however - TCP had been alive for  
> years, grown more and more capable, and finally any reason for  
> continuing to use the Host-to-host protocol within the relatively  
> tiny ARPANET itself evaporated.  The world had already passed it by.





-------------------------------------------
Archives: https://www.listbox.com/member/archive/247/=now
RSS Feed: https://www.listbox.com/member/archive/rss/247/
Powered by Listbox: http://www.listbox.com