by Darius Kazemi, Mar 6 2019
RFC-65 is titled “Comments on Host-Host Protocol Document No. 1 (S. Crocker – 8/3/70)”, dated August 29th, 1970, and is authored by Dave Walden, formerly of BBN, very recently of A/S Norsk Data-Elektonikk.
The technical content
This document is a list of suggestions for the Host-Host protocol proposal.
The first is to get rid of marking entirely and instead break every message sent into two messages: the first containing the leader (header) and metadata, and the second containing the actual content of the message. Because the content would be contained in its own sub-message, every computer no matter what its architecture would know that the content starts on the second message at position 0. “Thus, no more hunting for the beginning of the data is necessary.”
Next the author states there should be a control socket rather than a control link. No reason to hard-code a link for a special purpose when a more flexible socket could be used.
Next Walden says that “assigning sockets permanently to certain network resources should be encouraged and a directory of the socket/resource associations should be available somewhere in the network, perhaps in physical book form at each site.” This is essentially what we do today when we agree to things like “websites over HTTP are assumed to use port 80”.
There are a few more comments I won't go into detail on, followed by a comment that the error message system as laid out in the current proposal is perhaps over complicated and not as useful as people might think it is. An argument is made for a simpler error handling system.
Finally, Walden notes with some regret that he is now making suggestions that his former coworkers at BBN would need to do extra work to implement on the IMP. It's one thing to do the work yourself, it's another to ask that others do the work for you.
The most important bit of this RFC, to me, is the suggestion for messages to be broken into two. This multiple message model will eventually be evident in the design of TCP/IP.
There is an interesting comment about how links don't really matter to the Host-Host interface; they are really more of an internal IMP thing, useful to Hosts only a shorthand “so that socket numbers don't have to be included in all messages and to simplify table look-ups”. It seems to me that where the fundamental transmission identifier of an IMP is a link, the fundamental transmission identifier of a Host is a socket.
You can read about Walden's move to Norway in section 9 of his brief memoir. It was a short-lived stay of about a year, after which he went back to work at BBN.
Socket numbers did eventually get standardized and assigned to specific purposes. The first such listing is from 1972 in RFC-433.