Review of: Characterization of Proposed Standards

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 




Review of:    Characterization of Proposed Standards
I-D:          draft-kolkman-proposed-standards-clarified-04

Reviewed by:  D. Crocker
Date:         24 October 2013


Summary:

 This draft offers revised normative text for the criteria used to
qualify IETF drafts as entry-level Proposed Standards status. The
draft's background discussion makes a number of assertions of fact
without substantiation, at least two of which are probably not correct.
The revised normative text includes two criteria that also are likely
inaccurate.

   The draft is not ready for approval.



Detail:


Abstract

RFC 2026 describes the review performed by the IESG on IETF Proposed
Standard RFCs and states the maturity level of those documents.
This document clarifies those descriptions and updates RFC 2026 by
providing a new characterization of Proposed Standards.

"those" descriptions?  As the Introduction confirms, the draft covers
only one of them.


2. IETF Review of Proposed Standards


The entry-level maturity for the standards track is "Proposed
Standard".  A specific action by the IESG is required to move a
specification onto the standards track at the "Proposed Standard"
level. Initially it was assumed that most IETF technical
specifications would progress through a series of maturity stages
starting with Proposed Standard, then progressing to Draft Standard
then, finally, to Internet Standard (see RFC 2026 section 6). Over
time, for a number of reasons, this progression became less common.

It was /always/ uncommon.  There was never a time when progression
through the full sequence typically happened.


In response, the IETF strengthened its review of Proposed Standards,
basically

It was not 'in response'.  There was never a process of discussing the
issue and changing a policy formally and with clear IETF rough consensus.

The higher bar for Proposed happened organically and mostly through many
independent decisions of those running the standards process over the
years.

Even better, it was always controversial, when discussed in public.  Or
rather, the question of how high the bar should be was always
controversial and was never subject to an explicit rough consensus process.


operating as if the Proposed Standard was the last chance for the
IETF to ensure the quality of the technology and the clarity of the
Standard Track document.  The result was that IETF Proposed
Standards approved over the last decade or more have had extensive
reviews.

Many have.  Many have not.  It has depended upon the diligence chosen by
everyone working on the draft or doing reviews.

I'm pressing this point because the language being used for this draft
is leaning towards mixing formal process measures to raise the /effort/
needed to get approved, with actual quality assurance /results/.  In
fact we often publish drafts that have actually have very few eyes doing
deep and thoughtful review and we often publish drafts that don't work
very well.  That is, the much higher formal bar still allows quite a bit
of cruft to jump over...


Because of this change in review assumptions, IETF Proposed
Standards should be considered to be at least as mature as final
standards from other standards development organizations.

An assertion like this requires some discussion detail that compares the
IETF process to specific, other SDO processes, both formally and
effectively.  I'm not disagreeing with the conclusion, but with its
being made rather casually and firmly, but without substantiation.
(Part of the real challenge in doing the comparison is looking for
alternative forms of quality assurance that other SDOs might do, other
than the explicit reviews the IETF does.)

By the way, such a comparison discussion will probably serve to greatly
strengthen the ability to explain/argue for the comparative maturity of
Proposed...


The IETF review is possibly more extensive than that done in most
other SDOs owing to the cross-area technical review performed by the
IETF, exemplified by technical review by the full IESG at the last
stage of specification development.  That position is further
strengthened by the common presence of interoperable running code and
implementation before publication as a Proposed Standard.

There's actually a reasonable possibility that we do /less/ pre-Proposed
interoperable implementation now than we did 20 years ago.

Consequently an assertion that implementation is 'common' needs some
objective substantiation in comparison to the time we defined the
standards labels.



3. Characterization of Specification


Section 3.1 of this document replaces RFC 2026 Section 4.1.1.
Section 3.2 is a verbatim copy of the characterization of Internet
Standards from RFC 2026 Section 4.1.3 and is provided for convenient
reference.

3.1. Characterization of IETF Proposed Standard Specifications


The entry-level maturity for the standards track is "Proposed
Standard".  A specific action by the IESG is required to move a
specification onto the standards track at the "Proposed Standard"
level.

A Proposed Standard specification is stable, has resolved known
design choices, is well-understood, has received significant
community review, and appears to enjoy enough community interest to
be considered valuable.

The text has two major flaws that serve to assert a much higher
practical maturity for the Proposed Standard documents we publish than
is very often observably true:

Well understood:

   For a specification with any interesting level of innovation or
complexity, it is almost never 'well understood' at the time it is
approved by the IETF.  That requires significant implementation,
deployment and use experience.

   The fact that many eyes might have done a static review or even that
a couple of folk have written some code or even that those folk
interoperated does not come close to qualifying the specification as
"well understood".

Enough community interest:

   We often publish specifications that actually have very little
community support.  This is evidenced by a) very low working group
participation levels, and b) failure to successfully deploy and get used.




Usually, neither implementation nor operational experience is
required for the designation of a specification as a Proposed
Standard.  However, such experience is highly desirable, and will
usually represent a strong argument in favor of a Proposed Standard
designation.

The IESG may require implementation and/or operational experience
prior to granting Proposed Standard status to a specification that
materially affects the core Internet protocols or that specifies
behavior that may have significant operational impact on the
Internet.

A Proposed Standard will have no known technical omissions with
respect to the requirements placed upon it.  Proposed Standards are
of such quality that implementations can be deployed in the
Internet.

And yet we know that this often is not true.  However, language like
this is going to cause the IESG to naturally raise the bar even higher,
for fear that a spec still is not good enough.

The real goal of the language change, here, is to give assurances that
the specification is "sufficiently" mature, in terms of concern for
market deployment.

The part that the document can't say, but has as an implicit point since
it is motivating the draft, is that the maturity is at least as good as
that produced by other SDOs.

Perhaps:

    Proposed Standards are of such quality that they are ready for the
usual market-based product development and deployment efforts into the
Internet.


However, as with all technical specifications, Proposed Standards
may

Remove "However".


be revised if problems are found or better solutions are identified,
when experiences with deploying implementations of such technologies
at scale is gathered.



4. Further Considerations


While less mature specifications will usually be published as
Informational or Experimental RFCs, the IETF may, on occasion,

On occasion?  In fact it's quite common and possibly the norm.  Let's
not pretend otherwise.


publish a specification that still contains areas for improvement or
certain uncertainties about whether the best engineering choices are
made.  In those cases that fact will be clearly and prominently
communicated in the document e.g.  in the abstract, the
introduction, or a separate section or statement.


d/


--
Dave Crocker
Brandenburg InternetWorking
bbiw.net




[Index of Archives]     [IETF Annoucements]     [IETF]     [IP Storage]     [Yosemite News]     [Linux SCTP]     [Linux Newbies]     [Fedora Users]