The IETF's problems
Iljitsch van Beijnum
iljitsch at muada.com
Fri Jul 18 12:10:48 CEST 2003
As a relative newcomer (2 years active on a mailinglist for a wg,
Atlanta was my first IETF meeting and this is my second) this is what I
think are the IETF's main problems.
To me, it's not very clear whether the IETF wants to be a top-down
standards organization or a bottom-up place where people with good
ideas can work together to make those ideas even better. When both the
IETF leadership and members of the community feel some work is valuable
there isn't much of a problem, but when people bring work to the IETF
that the wg chair / area director / IAB infrastructure doesn't
recognize as important, this often leads to unhappiness. Now obviously
this is somewhat unavoidable because of the open structure of the IETF:
"anyone with time on their hands and a keyboard can write a draft". But
when even large vendors are unable to get protocols that they feel is
important (and have implemented or are implementing) "through" IETF,
there is a problem.
The other major problem for the IETF is that it doesn't manage its
resources very well. The major resource the IETF has is the combined
intelligence of a large number of very capable individuals, many of
which are under no obligation to spend time on a given subject, but do
so because they want to contribute to the community. The only thing the
IETF has to offer in return is a little (usually very little)
recognition. The part that I don't understand is that there is no real
structure for technical review in place. The IETF almost exclusively
relies on unsollicited comments from the membership at large, and on
the efforts of the IESG.
Meetings aren't always effectively chaired. People line up at the
microphones and address any subject they want, rather than that all
issues are dealt with one at a time. Many other aspects of meetings
aren't thought out very well either. The overlap in sessions is very
bad, whith no apparent rhyme or reason. It also seems strange to me
that there should be fixed breaks for meals. Eg: now wg A and wg B
overlap, and then we all eat. If wg A and wg B are scheduled after each
other, people who only want to attend one can still eat, but people who
really want to attend both can do so if they feel this is important
enough to skip a meal. (Due to timezone and cultural differences not
everyone wants to eat at the same time anyway.)
A non-problem: everyone seems to assume that everyone who shows up for
a session is an active member of the wg. That isn't the case. Many
people just wander in to see what's happening in a group that seems
interesting as they're there anyway.
Then there is the fact that the IETF has no collective memory, but only
a rewind button. There are posts to mailinglists, comments in meetings
that may or may not be recorded for future generations and drafts that
are guarantueed to disappear in 6 months, and RFCs. This means any and
all results either entirely disappear or are relegated to the obscurity
of long forgotten mailinglist archives, or they must be published as
RFCs. There is no middle ground.
About the RFCs: there are simply too many. Maybe at some point issuing
a new RFC number for each update of a document made sense, but
constantly having to search through the index to find the new number
for an update (titles could be better too, searching can be hard at
times) doesn't really increase productivity.
More information about the Problem-statement
mailing list