Dave,
Thanks for the thoughtful response. Comments inline...
Meta point:
BOFs serve multiple purposes and therefore are appropriate at
different times. So the model I'm proposing doesn't cite BOFs
explicitly as part of the sequence. That's not to deny their use but
rather means "do a BOF whenever it makes sense to do one"...
> Firstly, the prerequisites for working group formation under those
> rules are going to be based on - essentially - forming a stealth
> working
It is already reasonably easy to get a non-working group IETF mailing
list. I like that and suggest that that should be the usual entry point
for nascent wg efforts. So: not stealth and not outside the IETF.
> group, which somewhat implies gathering people who know how this
> works, and I'm not entirely sure that an external group would figure
> this out.
Required familiarity with IETF process is a basic and long-standing
issue, independent of my proposal. There's been a theory of having one
of the co-chairs be IETF-savvy, but that's probably later in the
sequence than we need.
So I'll trivially suggest that when a non-wg mailing list is created, it
should be assigned an IETF-savvy mentor. (This would carry no
implication that the mentor would become a co-chair, should the effort
get chartered.)
> Secondly, just as I-D is the new Proposed, I worry we're in danger
> of "left-shifting" our process again, to paraphrase Scott Bradner. I
> make
Certainly a fair concern, since I share your view on the overly-high bar
for Proposed. (And I'm thankful we were, at least, able to make the
criteria for full Standard be simpler and more pragmatic.)
So, yes, I'm suggesting a higher bar for chartering.
But note that I am /not/ suggesting any change in what it takes to get a
mailing list. I think that it should be pretty easy, requiring no more
than a basic Statement of Purpose and passing a basic 'smell' test with
the sponsoring AD. So the basic ability to conduct an Internet technical
discussion in the IETF and start doing work would not change. To me,
that's the important step.
In fact, I claim this /lowers/ the bar for getting work done. Here's my
logic: right now, a group has to focus first on getting chartered.
That tends to take months. Eventually they get to focus on the work.
But with my suggestion, they are expected to focus on the work immediately.
One of the current wg efforts /refused/ to pursue substantive technical
work prior to chartering. This led to months of problematic technical
proposals being circulated, with no effort to deal with the problems.
(The problems did get attention, many months later.) With my
suggestion, the expectation would be that substantive technical concerns
would be addressed immediately.
> Thirdly, it's not actually clear to me that there is a problem with
> taking on work - and even creating working groups - and then later
> dropping it if the supposed interest doesn't materialize.
We tend to ignore the economics of unsuccessful or problematic efforts,
essentially treating working groups as a cost-free entity. And we
frequently produce useless specifications.
The reality is that these are extremely expensive. They consume
expensive engineer time, scarce management resources, and scarce IETF
meeting time.
They often also impose an opportunity cost, while the world thinks that
something productive is being done about a relevant problem, when in
fact progress is not being made. Alternatives are not pursued, based on
the mistaken belief that the problem is being worked on.
Successful IETF work requires clarity about the topic being pursued,
meaningful market need for pursuing it, and people willing to work on it.
Currently, the IETF criteria for chartering a working group pretends to
approximate measuring these, but in reality does not. Rather it takes
only the vaguest expressions of interest as sufficient. Expressing
interest is not the same as doing work.
I am suggesting fixing this disconnect by requiring at least
substantive, sustained demonstrations of interest in the topic and an
ability to make progress with it.
d/
--
Dave Crocker
Brandenburg InternetWorking
bbiw.net