working group commitment

[Dave Crocker]

Ted,

Thursday, March 27, 2003, 11:49:11 AM, you wrote:
TH> I'm wondering whether
TH> the fairly obvious differences in approach reflect a different
TH> understanding of what a volunteer organization is like.

     1) We need to stop referring to the IETF as a volunteer
     organization. We are using the term as an excuse for poor
     management of our work.

Most attendees are paid to attend, or at least have their travel costs
reimbursed.  The number of people who pay for IETF participation out of
their own funds and who derive no job-related benefit is minuscule.

Yes, the IETF does not directly "employ" participants, but we are a body
of professionals trying to produce professional work-product. Let's stop
talking about ourselves in any other terms.

(and, yes, I think this issue is key to this thread.)


>> TH> 3) How do you manage the working group as energy fades (typical
>> for many wgs)
>> Terminate the working group.


TH> This seems to me to imply two things.  First, it seems to implythat 
TH> there is
TH> no objective benefit to the work being completed.

It implies that an effort, lacking significant participation, is likely
to continue to fail to have significant participation. It implies that
significant participation in the working group is at least an indicator
of industry interest. No participation equals no interest.

There are many wonderful things to do in this world.  We cannot do them
all.

     2) IETF working groups cost money. Lots of money. We need to manage
     our workload with a serious concern for efficiency and utility.

Standards are about obtaining a critical mass of community support. If a
working group cannot muster it, why should anyone believe that the
industry will?

Some years ago, I did a back-of-the-envelope calculation for the
aggregate cost of the smallest, simplest IETF Proposed Standard. The
number came out at US$ 1 Million. I suspect the cost has not gotten
smaller.

We need to run the IETF with a constant concern about wasting millions
of dollars.  Instead we tend to treat IETF work as free, with no serious
concern for cost, either dollar cost or time cost of valuable resources,
namely folks like Area Directors...


TH> I've been involved in wgs where the work energy faded almost to
TH> nothing as the process ran to completion, but the output did get
TH> used and re-used.

Details about these cases would be helpful.

Let's look for a) extent of industry use, and b) likelihood of industry
use. That way we can make some assessment of cost vs. benefit.


TH>    No doubt some other solution
TH> would have been found had those working groups died, but it would have
TH> been slower and the results almost certainly would have been narrower
TH> in scope.

Given the current, typical timeline for IETF working group productivity,
we probably need to be a bit judicious in claiming that an activity
pursued elsewhere will be slower...


TH> Though that is my subjective assessment of the result, I suspect an
TH> objective assessment would be that we are often better off in some
TH> measure by having work complete than by having it killed off.

     3) My subjective assessment is that the drain on IETF resources.
     the tone of frustration, and the tone of management-without-resolve
     that such efforts engender make it far better to cut our losses.

Now, how do we reconcile/resolve our basic disparity of subjective
perceptions?


TH> Second, it seems to imply a management structure that has the
TH> community's blessing to apply a major stick.

Yes it does.

1.  It has been an integral part of IETF management since its inception.

2.  It is a critical component to the management of any organization
with limited resources and a concern for productivity

3. There absolutely must be a mechanism for the larger IETF community to
eliminate a working group that is unproductive, because that working
group's existence has assorted negative impacts on the IETF community,
as noted above. The ADs are delegated responsibility for representing
the interests of the IETF as a whole, for internal operations.

4.  RFC 2418:

2418> 4. Working Group Termination
...
2418>    If, at some point, it becomes evident that a working group is unable
2418>    to complete the work outlined in the charter, ...
2418>    the Area Director, in consultation with the working group can...
2418>
2418>    3. Disband.
2418>
2418>    If the working group disagrees with the Area Director's choice, it
2418>    may appeal to the IESG (see section 3.4).

Seems pretty clear to me.


TH> In a volunteer organization, there are often difficulties in setting
TH> deadlines and deliverables. Here, we get buy in to those
TH> deliverables and milestones both by the working group and the larger
TH> community in the charter process.

The buy-in is meaningless unless working groups are held accountable to
them, by that larger community.  The question is how to do the holding.

Historically, the IETF approach is simple: Let the working group figure
out out to achieve its milestones. If it fails to satisfy them, RFC
2418, Section 4 kicks in. (And we can skip past the fact that we have
not used it nearly as often as has been warranted.)


TH> If a group decided to evaluate itself by its charter and stop when it
TH> didn't
TH> meet the charter, I think we would agree

It appears that your are forgetting that the premise to my comments was:
"A working group does NOT have an active set of productive
participants". That is, the "working group" has ceased to have enough
participation for such self-evaluation to be valid. Hence it no longer
has any legitimate basis for declaring rough consensus or otherwise
laying claim to be representative of some larger community.

     4) IETF chartering is not a perpetual license of community
     validity. Validation of a working group is a continuing process.

     
TH> who is it, exactly, that we are holding to this deadline and
TH> deliverables? The doc authors and chairs are the well-known
TH> stuckees, but they may well be doing their work even when a working
TH> group is in a problematic state for review or general energy.
TH> Killing a group in that case runs the serious risk of losing the
TH> volunteer energy of folks doing work, as they have had their work
TH> stopped based on the lack of interest of others *without any way to
TH> identify who those others are*.

This does rather go to the heart of the typical case, doesn't it? A tiny
group of workers slugs through the mud of specification effort, in
isolation. The working group has noone else providing assistance and
validation.

     5) A working group without a critical mass of community involvement
     is really a private effort of a few individuals.  There is nothing
     wrong with such an effort, as long as we do not claim that it has
     community rough consensus.

In other words, when a working group becomes a private effort, make it a
private effort officially.

And while we are talking about "losing volunteer energy" let us be sure
to look at the frustrations that often are the cause of all those OTHER
"volunteers" dropping out of the working group, as well as the rather
serious damage that is done by promulgating specifications that purport
to have community support, but do not.

d/
--
 Dave Crocker <mailto:dcrocker at brandenburg.com>
 Brandenburg InternetWorking <http://www.brandenburg.com>
 Sunnyvale, CA  USA <tel:+1.408.246.8253>, <fax:+1.866.358.5301>