Spencer Dawkins sdawkins at
Wed Feb 19 10:18:21 CET 2003

Hi, Brian,

TCP is actually a good example - the basic packet format hasn't
changed much since RFC 793, which is a full standard - but 
that's the RFC that DOESN'T describe current congestion control
mechanisms - arguably the most important component of TCP. They
are described in RFC 2581, a Proposed Standard, which obsoleted
RFC 2001, which was a Proposed Standard, and are updated in RFC 3390,
which is a Proposed Standard ...

Transport Is Different, but we seem to be using Experimental for
something pretty close to "a proposed standard" - that's what we did
with RFC 3465, anyway. The idea was to let Appropriate Byte Counting
"age" before it was widely deployed, and Experimental was intended
to discourage wide deployment...


> -----Original Message-----
> From: Brian E Carpenter [mailto:brian at]
> Sent: Wednesday, February 19, 2003 10:07 AM
> To: problem-statement at
> Subject: Re: Standards
> Another cut at this.
> We often hear it said that "The Internet runs on Proposed Standards."
> Objectively, it's not 100% true (since IPv4 and TCP are Standards) but
> to what extent is it true?
> The reason this is interesting is that if, in practice, the Internet
> largely runs on PS's, then a legitimate problem statement might be
>   We waste a lot of time and energy on promoting PS to DS and 
> occasionally
>   to S, but this has no practical impact on products and 
> services, which
>   happily run on PS's.
> (If this is a problem, the solution is not hard to find.)
>    Brian

More information about the Problem-statement mailing list