[ih] Who starts with standards? (was Re: IETF relevance)

[Craig Partridge]

On Wed, Aug 30, 2023 at 11:35 AM Andrew Sullivan via Internet-history <
internet-history at elists.isoc.org> wrote:

> [ObDislcaimer: speaking only for myself and not the host of this list.]
>
> On Wed, Aug 30, 2023 at 12:47:35PM -0400, Miles Fidelman via
> Internet-history wrote:
> >Traditionally, protocols have never "originated" with the IETF
>
> Without wishing to be cheeky, why would anyone start with a standard and
> try to work outwards from there?  If you don't at least have a prototype or
> a scratch proposal, you will spend the first 8 years arguing about the size
> of the problem and the next 8 regretting that you didn't ship any products.
>

Speaking from the perspective of IETF c. 1986, what happened then (for
SNMP, for MIME, for CIDR, for TCP large windows, etc.) was:

   - A gap in the protocol suite was identified.  E.g., we did not have a
   standard network management protocol -- and those protocols that did exist
   clearly did come close to scaling/meeting project Internet needs.  For
   CIDR, we were using the address space inefficiently and would run out in a
   few years (aaggh!). For TCP LW, TCP performance was capped at a data rate
   that no longer made sense.
   - Time was short -- we had anywhere from 6 months to 2.5 years to solve
   the issue or some community was going to be chasing its tail all day, every
   day, patching the problem using people rather than protocols.
   - So we picked the best group of folks who could (based on expertise,
   availability, and, in some cases, how hard we could twist their arm), and
   sent them off with directions to come back with a standard ASAP or else.
   Now sometimes this provoked dissension and splinter groups with competing
   standards (see IP over ATM standards, or network management), but IETF
   quickly developed processes to deal with such situations.

In those days, standards decisions were critical technology enablers -- aka
the Internet in some cases might fail if we didn't have a standard in time
and a prototype didn't exist.  From that perspective, what I continue to
find amazing is how many of the things we did in that era -- without time
to prototype fully -- turned out right.

Craig

-- 
*****
Craig Partridge's email account for professional society activities and
mailing lists.