[wg-d] Forwarded note re. thoughts on wg-d discussions

[Theresa Swinehart <Theresa.Swinehart@wcom.com>]

WG-D members -

Please find attached a note I've received from John Klensin which I am
forwarding to this list. It is very insightful with regards to the
discussions in this group, and reaching resolution on identified issues.


Theresa
Theresa,

I have just reviewed the content of the working group D archives
and related GA materials, and am quite concerned about what
seems to be going on there. The shorter note I posted to the GA
list yesterday is, to some degree, part of this thinking and I
hope consistent with it. Since I am not on the list, I am
forwarding this to you in the hope that you will post it if you
think it would be useful.  It is also pretty long; feel free to
edit it or break it into pieces if that would help.

There appears to be a good deal of wandering around through
utopian theories and the unfortunate belief that things could be
made true just by wishing them so or repeating them enough.
This approach doesn't work; it might work better if we had
infinite time but I don't perceive that as being the case.
Perhaps worse, the list has gone off into extensive discussions
of things that really don't make a lot of difference while
virtually ignoring major strategy themes.  I'm going to try to
identify some of the latter below in the hope of refocusing the
discussion a bit.

This note is written as an individual and from the perspective
of my original training in the behavior and dynamics of
political systems, especially communications, influence, and
decision patterns.  I am not "representing" anyone here,
especially not MCI WorldCom or the IETF.


_Constituencies within constituencies_

The whole way in which the constituency model, and its extension
to WGs, seems to be shaping up, strikes me as unfortunate.  From
what I've seen of late, the constituencies are being defined as,
or transformed into, core advocacy groups for different,
potentially polarized, positions.  This is akin to a process for
resolving differences of opinion in which "you" round up
everyone you can get to support your side, "he" rounds up
everyone he can get to support his, each side develops a
position and locks it in, issues weapons to its supporters, and
goes at it.  From the viewpoint of a third party, this is the
old "let's you and him fight" model.

There are major problems with that approach --or any other
approach in which people believe their mission is to act as
advocates for particular positions on the assumption that there
will be qualified and effective advocates for the others--
which is that these things require a mechanism for determining
who won, and most of those mechanisms don't lead to
non-polarizing or win/win outcomes.  For example:

 - One can adopt the "battle until one side has no one left
   standing" model, in which case the surviving side, and its
   positions, "win".

 - One can appoint a judge.  One might even give the judge the
   authority to pick a middle position and impose it.   But the
   judge better be supported by sufficient agreement in the
   community that her decisions are accepted with relatively
   good grace by all.

 - One can try to vote, but see below.

Community consensus --except as determined by one of the above
mechanisms or something similar-- doesn't work here.  If one
had community consensus, one wouldn't need advocacy
constituencies.  Indeed, advocacy constituencies often work
_against_ consensus-development.  

Contrast the constituency model as it seems to be evolving in
the DSNO with the historical Jon Postel/IANA problem-solving
mechanism.  There, the starting assumption was that both parties
to a disagreement ultimately wanted what was best for the
Internet but differed on what that meant or how to get there
(that wasn't always true, even in the hypothetical "good old
days", but it is a much more useful assumption if two parties
are going to negotiate than having each assume that the other is
an instrument of the devil).  The role of the IANA then became,
not deciding, but creating an environment in which the parties
had maximum opportunity and incentive to jointly figure out a
mutually-acceptable solution.  The IANA also sometimes acted to
widen the debate by introducing possibilities for discussion
that hadn't occurred to either party.

It didn't always work, but it did work fairly often.  And it is
worth noting that, at least IMO, one of the things that
contributed to its working was that the "sides" were not forced
into taking clear, well-defined, and very public positions
before starting discussions.  That made it easier for them to
discuss a range of options, rather than just advocating their
starting positions, and, if compromises emerged, easier to
accept them without loss of position or face.

I don't know if we can move the DNSO processes back in that
direction, or even if doing so is desirable (although I do have
prejudices on that subject, which are probably clear by now).
But I am quite sure, as a matter of logic and history, that, if
we go down the current advocacy-constituency path, we had best
be spending more time on figuring out who the judges are going
to be, what authority they are going to have, and how their
edicts are going to be enforced.  


_Voting_

The problem with trying to work out voting rules is that one
must first figure out who gets to vote and, more specifically,
must be able to count: "for", "against", "abstaining", and "not
voting for some reason, but ought to be able to do so".  If the
last group outnumbers the other two, then conflicts about
"silent majorities" and people claiming to represent that group
(whether its members are aware of it or not) are almost
inevitable, and that puts all of the ambiguity back into
whatever it was that voting was designed to solve.  A number of
the recent discussions appear to me to be setups for "votes" of,
say, 5 to 3 to 7 to 3 or 4 billion.  While that might be
reasonable if the 3 or 4 billion had the clear opportunity to
participate and declined, I suggest that no one will find those
ratios satisfactory if there is any hint of mechanisms at work
to exclude people (need to attend meetings, admission
requirements to working groups, lists with too much noise to
follow over expensive links or with little time,...).  And
people would would like to appoint themselves the
representatives of the 3 billion are likely to object especially
strenuously.

The IETF's stress on "rough consensus", rather than voting is,
IMO, less a high moral principle than a pragmatic one: if people
can look around, conclude that everyone interested has had
adequate opportunity to understand the various issues, and that
a significant majority of the interested portion of the
community is clearly leaning one way or the other, then that is
enough.  This not only avoids the charade of votes in a
community that permits completely open participation and that,
strictly speaking, has "participants" and not "members", but
avoids having to get dragged into complex rules, e.g., about
whether two people "from" the "same organization" can vote and
what is, or is not, the "same organization" and what sorts of
relationships constitute "from". 

And there are judges: IETF selects, through a fairly convoluted
procedure, members of the IESG and IAB and then pretty much lets
them --with WG Chairs appointed by the Area Directors who make
up the IESG on the front lines-- determine when consensus has
occurred (and when there has been adequate consideration and
exposure of the ideas for that consensus to be meaningful).  As
judges, they operate in the sight of an often-suspicious
community most of whose members aren't shy about expressing
their opinions: if there is community consensus that the IESG
and IAB have gotten out of touch or are becoming arbitrary, the
community rebels and applies methods only slightly more gentle
(and selective) than "load all of the nobility into carts, drag
them off to the guillotine, and start over". That doesn't happen
very often (for which those of us who might be among the first
into the cart next time are quite grateful) although some
members of the community seem to regularly want to stir up such
a revolution, some perhaps in the interest of an enjoyable
spectacle and others because they are deeply offended over one
incident or another and see bloody revolution as the only
remedy.  

Do I think that model would work without modification for
ICANN?  No more so than Karl does, but for rather different
reasons, but a discussion of those reasons would be just
another diversion here.  But voting isn't a solution either
unless there is real consensus about who gets to vote and it is
a closed set, with no one claiming to represent anyone else
without explicit authorization and mandate from that person and
with general agreement that the rules for who does and doesn't
get included aren't unacceptably discriminatory.  Of course,
governments get to operate without such authorization, but we
are trying to avoid creating conditions that require
governments to run ICANN and the Internet, aren't we?


_Rules of Procedure_

Robert's Rules of Order is a fine document.  It may be a little
too grounded in the legislative traditions of the UK and USA
for an international forum (a point I'm surprised that
apparently no one has made loudly), and there are too many
different (and often contradictory) versions around for
"Robert's Rules of Order" to be a good unambiguous reference.
But those sets of rules are rooted in two assumptions:

   * that all participants in the assembly are acting in good
     faith and for what they perceive as the best interests of
	 the group (and that anyone who doesn't can be ejected by
	 the group and will stay ejected as long as the group
	 thinks that appropriate).

   * that all discussions are synchronous, i.e., I hear what
     you are saying as you say it and have little opportunity
	 to initiate a discussion without the possibility of
	 knowledge of prior discussions (I might do it anyway, but
	 would be likely to be ruled out of order, or at least as
	 being in poor taste).

   * if there are disputes within the group about the proper
     procedures to be followed in a particular case, they can
	 always be settled by voting.

The first is observably false in the ICANN case, much as I
would wish it otherwise.  The second doesn't characterize
discussions that are held, in large measure, by email.  Of
course, what I think of voting in membership-open and
international bodies should be clear from the above.  So, while
I think the topic is interesting, it, like getting into the
details of how IETF works, are just diversions here.


_WGs as Design/Writing/Strawman-formulation groups vs WGs and
decision groups_

Following some of the themes above, there are a few different
things one can do with a WG.  Most are variations on one of two
themes:

  * A WG's real job is to formulate tentative positions for
    review and modifications by a larger group.  This model
	starts from the recognition that large committees are
	traditionally lousy design or drafting groups, and that
	bodies larger than committees are usually worse.  With it
	(which IETF calls a "design team") there is no particular
	reason for the WG to be either open or democratic: its
	critical role is only to produce positions to which a
	different group can react.  To a considerable extent, the
	more it can be broadly-based and still function, the more
	likely it is that its proposals will be acceptable, but
	"function" is the important thing.  "Democratic paralysis"
	is a total failure.

 * A WG is where the real work gets done.  It is expected that
    the vast majority of the recommendations it produces will
	be generally acceptable.  To some extent, if the broader
	groups have to reject or alter the output of a WG, then
	there is something wrong with the system.  As an almost-
	inevitable consequence of this, the WG must have
	representation from, or a way to tap the ideas and
	positions of, all materially-concerned parties.  But, by
	that very fact, it may be harder for it to formulate
	initial positions, rather than that work out balance points
	among them, modify them, or approve or reject them.

It seems to me that much of the confusion in WG-D is that
people want WGs to serve both of these roles at the same time.
I can't tell you it is impossible, but I've never seen it work.
Either answer is plausible, but arguments about how to make the
first type of group open to everyone who might be interested
and representative of all possible opinions appear to me
--especially after reading the comments on the list-- to be a
waste of time.

Interestingly, in many organizational settings, either type of
WG going off and not being able to reach a conclusion on a
given topic is considered an answer and useful information.  If
a WG, especially of the second type, is charged with figuring
out how to do some X, and cannot agree on how to do X or even
if X is desirable, the organization has to either conclude from
that result that X should be deferred or has to have other
mechanisms for getting to a result.


_Decision Locus_

Regardless of the role into which you cast the WGs, you are
going to need some mechanism for reviewing and approving their
work (or rejecting it, or sending it back for more effort).  If
WGs operate on the first model above, that mechanism will also
need to be able to draw together differing opinions and either
to write documents or commission other WGs to do so.  In the
existing structure, either the GA or the NC can, in principle,
serve this role.  The GA has the difficulty that it will, I
think, be forever difficult to figure out who the GA actually is
and how consensus is determined.  Its strength is in an "open to
anyone interested" membership, but that leaves it, as a decision
body, extremely subject to capture by people claiming to have a
mandate from people not directly present.  On the other hand,
the NC, constructed as it is from adversarial constituency
groups, is, as we have seen already, not without problems as a
consensus-determiner. 


_Conclusion_

WG-D has a relatively small number of critical decisions in
front of it, but those decisions need to made --by the WG or
elsewhere in the DNSO-- if there is going to be any progress.
Other things are just diversions, and diversions should, IMO, be
deferred until the DNSO is running smoothlyl

(i) Are you going to try to make decisions on an adversarial-
constituency basis or on a cooperative, compromise-seeking one?
Probably can't do both at the same time.

(ii) Are you going to try to determine agreement or the lack
thereof, or to reach agreement, by rough consensus or by voting?
If the former, then what is the mechanism for determining that
consensus exists and for reviewing such a decision if some
people believe it has been made erroneously?  If voting is the
choice, then you have a significant task ahead of you in
figuring out who can vote, a task that recognizes the issues
associated with a "vote" in which the votes not cast outnumber
the ones that are by several orders of magnitude and especially
the near-certainty that those in the minority will claim that
the supporters of their position were unfairly excluded from the
voting process.

(iii) If working groups are going to be a permanent part of the
landscape, you need to decide whether their principal role is in
designing and producing initial proposals or whether they will
produce finished proposals for action that are normally expected
to be adopted as such.  And you need a theory about how to
interpret a WG's inability to reach a conclusion on some topic
assigned to it.

(iv) Finally, you need to figure out how a decision is taken to
do something (e.g., forward a proposal to the ICANN Board, or
respond to a question from it, in the name of the DNSO).  If
that responsibility is to rest in the GA, then you have to work
out voting or consensus-determining procedures for what, as I
read the bylaws, is a pretty amorphous organization.  If it is
to rest in the NC, then a number of concerns about the
constituencies that make up the NC are going to have to be
worked out to the satisfaction of a very large fraction of the
community.  

    --john