From: Russ Allbery <r...@stanford.edu>
Subject: Big Eight hierarchy management transition
Newsgroups: news.announce.newgroups,news.admin.announce,news.groups
Followup-To: news.groups
Archive-Name: other.articles/big8-board-transition
Approved: newgroup...@isc.org
Date: Sat, 30 Sep 2006 17:04:56 -0700
Lines: 597
Message-ID: <1159661096.29508@isc.org>
Path: g2news2.google.com!news3.google.com!newsfeed.stanford.edu!bounce-back
X-Auth: PGPMoose V1.1 PGP news.announce.newgroups
iD8DBQFFHwYoXMotZRinPKkRAkytAKDCF+qD0mKkdiPyh2OrGjNurUfCFgCgqRfx
QK7sQ1ZmWIx5zTXf3llOnQU=
=VtGq
Xref: g2news2.google.com news.announce.newgroups:1311 news.admin.announce:3 news.groups:36373
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Big Eight Hierarchy Management Transition
September 30, 2006
Introduction
As most of the reading audience is probably aware, as of tomorrow,
Todd and I are stepping down as moderators of news.announce.newgroups
and ending our involvement in Big Eight newsgroup creation. Since May
of 2006, newsgroup creation in the Big Eight hierarchies has been done
under the aegis of a new Big-8 Management Board as a preliminary trial
period that Todd and I would evaluate before deciding how to handle
our resignations. This is my final report on that evaluation.
This will be a very long message. I don't want to leave anything
unsaid that may help someone understand my personal reasoning or that
might help those involved with Big eight newsgroup creation going
forward. It will be longer than many people will want to read; for
those who don't want to wade through the whole thing, please see the
next section for a summary.
Due to the increased worries that this sort of message might be
forged, in addition to the PGPMoose signature that all posts to
news.announce.newgroups receive, I have also signed it with my
personal GnuPG key. That key is in the Debian archive keyring and
part of the well-connected portion of the PGP web of trust and is
available from any major PGP keyserver as well as from my personal web
page.
Some additional personal thoughts about the time I've spent involved
in news.groups will be posted separately.
Summary of Decision
Based on the work in the past five months and the discussions
preceding that work, I believe that the Big-8 Management Board has
demonstrated their ability to handle new proposals in a prompt and
reasonable fashion and make defensible and reasoned decisions
concerning management of the Big Eight newsgroup list. I have some
concerns about their ability to maintain the Board, encourage useful
input, recruit new volunteers, and prevent burnout, but I believe that
the system they have designed is at least clearly superior in that
regard to the system that preceded it and has a reasonable chance of
success.
I am therefore handing over management of the news.announce.newgroups
control message signing key to the Big Eight board, namely Brian
Edmonds, Marty Moleski, Tim Skirvin, Joe Bernstein, Thomas Lee, Dave
Sill, James Farrar, and Jonathan Kamens, following the procedures
described at <http://www.big-8.org/>.
The control key for news.announce.newgroups used to issue control
messages in the Big Eight hierarchies (comp.*, humanities.*, misc.*,
news.*, rec.*, sci.*, soc.*, and talk.*) will not change as part of
this transition. The tradition has been for the key to be entrusted
to the best judgement of each news.announce.newgroups moderator or
moderator team to pass on to their successors, and my best judgement
is that this team of people will act in the best interests of the
users of these hierarchies and the sites carrying them.
I encourage any news administrator or Usenet user who is concerned
with the operation of these hierarchies to review the rest of this
message and the web site referenced above and to contact the Board
with any concerns that they have.
Analysis by Criteria
Output of New System
My primary criteria in evaluating the actions of this new management
structure is to look at the work they've done in maintaining the list
of newsgroups over the past five months. This is, in the end, what
matters. The purpose of this system is to make good decisions about
what newsgroups we recommend Usenet sites carry.
This criteria requires some clarification since the quantity of
proposals has declined sharply over the past few years, as has the
success rate of new newsgroups. My interest is neither in creation of
a large number of new groups nor in success of every newsgroup
created. Rather, what I wanted to see was prompt and reasonable
handling of new newsgroup proposals, a working system that was clearly
taking new proposals as input and making affirmative decisions on
them. Secondarily, I wanted to see a system capable of handling the
proposals and changes that were structurally difficult under the
previous voting system, namely group removals and handling of inactive
moderated groups and absent moderators.
Finally, I wanted to see that the new system was capable of handling
controversial groups and hard decisions as well as simple creations.
One part of that evaluation is incomplete since no group reorgs were
proposed during this five month period. However, I think enough
information is nonetheless available to arrive at a conclusion.
First, I think it is clearly demonstrated that the system is handling
new proposals and reaching conclusions on those proposals. The new
Board has taken over handling of incoming messages to newgroups,
group-advice, and news.announce.newgroups and is responding in a
timely fashion to proponents. Proposals are following a clear
sequence and decisions are posted publicly. The new process is also
already handling significantly more simultaneous proposals than the
process it replaced and resolving them more efficiently.
Over the last five months, the Board has created the following groups:
soc.religion.asatru
comp.soft-sys.octave
soc.support.vision-impaired
soc.men.moderated
(talk.current-events was also created a few days ago, but is too
recent to be part of this analysis.)
Of these groups, soc.religion.asatru has been a clear success, with
sustained on-topic traffic (222 messages in a recent 21-day period).
Furthermore, this was a controversial proposal with a lot of noise in
the discussion, and to date the concerns raised during the discussion
have not manifested on the group. This is exactly the sort of
successful decision that the old system would have had more difficulty
reaching.
comp.soft-sys.octave has been a clear failure so far, with no messages
in the 21-day sample period. Opinions on whether it was worth trying
will vary; I tend to lean towards not being too concerned if groups
are created and turn out not to be used if the creation itself won't
cause other problems.
The other two groups are more recent. soc.men.moderated is a special
case with several possible success criteria. It is a moderated
companion group to a long-standing high-flame group and may prove
useful even if it only provides an occasional outlet for the other
group (as has happened in several other similar cases in the past).
However, it is dormant at this point, apparently, to the selected
moderators no longer moderating. It's probably too early to say for
sure whether this is a success or failure; driven by the deadline for
this post, I would call it a possibly recoverable failure.
Finally, soc.support.vision-impaired has been a moderate success to
date with 63 messages over a recent 21 day period, although it's still
too early to tell whether this trend will continue.
From the creations done so far, then, the new system appears to have a
50% success rate, which as good as the last few years of the prior
system. More importantly, the Board has demonstrated an ability to
deal with two very controversial moderated group proposals, with mixed
success but with a process that was able to terminate. This is a
substantial improvement.
However, the bulk of the work done so far by the new Board is in other
areas. Newsgroup removal has been stymied for years by the previous
infrastructure and the Board has dealt with a backlog of over 45 group
removals. Most of these were long-dead INET groups promoted
previously so that we could issue checkgroups control messages. These
removals had very little controversy, and in the few cases where there
was some controversy, the Board acted with care and in some cases
helped revive the group.
Included in these removals is the removal of comp.binaries.apple2, the
only unmoderated binaries group in the Big Eight and a chronic thorn
in the side of Big Eight news admins due to its excessive share of the
total bandwidth required for a Big Eight feed. The Board was able to
deal with an ongoing problem that the previous system had been unable
to do anything about for years.
The Board has also dealt with several other cases of inactive
moderators, changing moderators of sci.physics.plasma, unmoderating
soc.culture.galiza, and robo-moderating soc.religion.hindu. In these
cases, the groups have often been inactive for so long that it will be
months, if not years, before we see if they find a new audience. More
importantly from my perspective, the Board is taking reasonable action
with these groups and has a procedure in place to deal with such
cases.
In summary, while there will not be unanimous agreement on all the
decisions taken, I believe that they are all reasonable and that most
other observers looking at the corpus of decisions will arrive at the
same conclusion. The results are, in my opinion, clearly superior to
the results that were being produced by the previous system,
particularly in the ability of the Board to deal with proposals like
removals and dead moderated groups that have no obvious voting base.
And, most tellingly, the Board has been able to deal with backlog of
known work that the old system had been accumulating for some time,
accomplishing more concrete improvement in the group list than we've
seen in years.
Management Structure
The second major criteria I had for success of the new system was a
sustainable structure. I think there are more significant risks in
this area, as I detail below. However, the Board has spread the work
across considerably more people and established a replacement
procedure and a sufficiently large active group that members have a
hope of being able to step down before they have burned out.
There is a structure in place that can absorb additional volunteers
down the road and let them make progress on their own concerns, not
just as members of the Board but as outside contributors to problems
such as inactive groups. One significant problem the previous system
had was accepting systemic contribution from people not directly
involved in the management of the hierarchy. The Board's handling of
the long-pending inactive moderator and inactive group removal
problems shows that they are doing a significantly better job at this.
Sustainability of the system and acceptance of input from new
volunteers is the hardest problem by far for ongoing management of the
hierarchy. The new system does not fully resolve all of the problems
I see in this area, but I don't believe any system could. It has,
more importantly, demonstrated far more flexibility than the old
system could muster, which gives me hope that it can continue to
adjust to this challenge going forward.
Documentation of System
Finally, part of setting up a new newsgroup creation system is to
document the new procedure. I believe that the Board has clearly met
this evaluation criteria and gone beyond it by providing a clearer and
more comprehensive information resource for Big Eight newsgroup
creation than we have ever previously had. <http://www.big-8.org/>
has not only the new policies and procedures but an easily readable
archive of decisions and more information about the format of a
proposal and about the overall process than we had under the previous
system.
Rejected Criteria
The above criteria are the three that I consider the most significant
in evaluating the new system. Many other criteria are possible, and
for the most part I won't comment on other possible choices. However,
there are several significant criteria that I did not apply, and which
I feel deserve some explanation.
Voting System
The original mandate for the new Board called for the creation of a
voting system to elect board members. This has not happened, which
has some possible negative consequences as detailed below. However,
after consideration of the arguments put forward by Board members
and the discussion of this point in news.groups earlier this year, I
decided not to require this in my evaluation.
Election theory says that any voting system requires a defined
electorate as well as several other security guarantees to provide an
election that can be considered fair. I have been convinced that,
while not impossible, establishing those conditions in the Big Eight
is at least exceedingly difficult. Elections are being used in some
other hierarchies, most notably uk.*, so it is clearly possible to
manage a hierarchy this way. However, even with newsgroup creation
polls (for which the stakes are lower than a Board election), the old
system was having significant problems running reasonable votes and
found clear evidence of people successfully manipulating the CFV
system and achieving results that were probably not representative of
the intended voting base. In the uk.*, with a small hierarchy and a
fairly limited set of participants, it's possible to apply more web of
trust metrics to evaluating votes that one can use in a general Big
Eight election.
If we tried hard to come up with vote vetting processes to work around
this problem, we would run into another problem, namely frustration
with strange rules and hoop-jumping necessary to vote. We were
already seeing this with the CFV process; it was one of the largest
problems with the previous group voting system. The resulting
exasperation doesn't contribute to one of the primary goals of a
voting system, namely a perception of fairness, and seems likely to
create more energy-wasting arguments.
Additionally, even if a fair vote could be held (meaning in this case
a vote in which each Usenet participant had one and only one vote),
I'm dubious that we could get a representative vote. In this respect,
votes for Board members face a problem similar to small city
elections, with the same likelihood that the results will be mostly
dictated by a small set of people directly involved in the process and
will otherwise face general indifference. If we had the sort of
healthy, broadly-representative, and extensive participation in
news.groups that we had back in, say, 1999, this wouldn't necessarily
be a problem. As matters stand right now, I think the expected voting
base is so small that the results would be dominated by specific
concerns unrelated to the general health of the Big Eight or, even
more likely, would be essentially random from election to election.
Finally, it's extremely important, given how thinly volunteers are
spread, that all of the members of the Board be willing to work
constructively with each other, back each other up on various internal
responsibilities, and work together to keep the system working. It's
possible to maintain this with generally elected members, but it's
certainly more difficult and would introduce a significant risk.
I would like to see a successful voting system created because it
provides a natural way to cycle new blood into the process and because
when done well it creates a strong perception of fairness that is
extremely difficult to achieve via any other process. However, after
long consideration, I believe the challenges are too difficult to have
this be a fair evaluation criteria.
Popularity
There are two aspects to this possible criteria. The first is the
number of news administrators who honor control messages from the new
Board. One possible criteria by which to judge the new system is by
whether it results in an increase in the number of sites honoring the
Big Eight newsgroup list.
On the surface, this seems like a criteria that drives straight to the
heart of the credibility of this process. However, efforts for the
past decade in getting news administrators more involved in the
process have mostly been a failure. I can say from personal
experience that most news administrators simply don't want to get
involved, either because they don't care or because they're too busy
or because they'd rather have independent management. While I would
be thrilled if it happened, I don't expect to see any significant
movement in the number of sites honoring control messages. Some will
stop; any major change of any sort will lose some people at least in
the short term. If we're lucky, some will start in the longer term.
Expecting any more than that is, in my opinion, unrealistic.
The second type of popularity that one could judge the Board on is
popularity in news.groups, either in the form of general approval of
the Board's actions or in the form of building consensus and
attracting new volunteers. Again, and in this case more sadly, I
think this is unrealistic.
There has been a steady erosion in the usefulness of news.groups for
holding meaningful discussion for several years, predating the Board
or any effort to create a new system. With the combination of two
highly controversial proposals and the arguments surrounding the
creation of the Board, that trend has drastically accelerated, but
not, I believe, fundamentally changed.
The Board is clearly unpopular with many news.groups posters. Anyone
evaluating this trial period should be aware of that; as spelled out
below, this creates clear problems. That unpopularity seems mostly
based on three areas of disagreement: the lack of voting and
accompanying loss of a concrete way to change newsgroup creation
results, the Board's willingness to create groups without proven
interest and see if they succeed, and the choice of moderators in
controversial moderated groups. However, the objections underlying
that unpopularity are mostly not expressed in a way that the Board can
respond to constructively, making it difficult to determine whether
they contain ideas that could lead to a better system.
Furthermore, in most cases I personally don't agree with the direction
expressed by those objections. I addressed voting above, and I
believe a more liberal newsgroup creation policy with a working group
removal system is reasonable way to proceed (and is well-supported by
many previous discussions in news.groups). As for the choice of
moderators, this has always been difficult and controversial and any
newsgroup creation system will have difficulties in this area. The
Board has, in my estimation, done at least as well as the previous
system did with controversial moderator selection. The success of
soc.religion.asatru to date, with none of the anticipated problems, is
significant evidence of that. Since I disagree with the primary
justifications of the unpopularity, I don't find the unpopularity
itself convincing.
Another possible evaluation criteria would be the ability of the Board
to foster as positive of an atmosphere as possible for discussion of
group proposals. This, however, is exceedingly difficult to measure
and largely not under the Board's control. Much of the debate has
been heated and personal, and while it's always possible to improve
how one handled such a situation, my evaluation is that the Board has
handled the situation better than I could and I'm dubious whether it's
possible to handle it significantly better. Given that, I think
successful output of the process over time is a superior evaluation
criteria and expecting the Board to simultaneously be popular in the
current atmosphere is too high of a bar to set.
Stability
Finally, one possible way of choosing a successor would be to look for
someone who would run the Big Eight largely the same way that Todd and
I have. This is roughly the criteria that has been applied in the
past.
I explicitly rejected this criteria at the beginning of this process.
I believed, and still do believe, that the prior system was
irrecoverably broken and that it was time for a much-deferred complete
overhaul. I was interested in seeing the system transition into the
hands of people who would not run things the same way that I have
since I believe the path I was on was heading for general failure of
the system. I wanted a group of people who would try more, risk more,
and experiment more. I believe what's needed at this point is the
opposite of this criteria.
Risks
news.groups
The biggest risk facing the Big Eight newsgroup creation system going
forward is the lack of a congenial and constructive place for
discussion of changes to the group list. This applies to any possible
system, including the previous system which also suffered greatly due
to this lack. However, the discussion of the creation of the Board
and subsequent reaction to Board discussions has clearly exacerbated
the problem.
At this point, most news.groups threads quickly acquire flamewars and
rehashing of previous disagreements that have to be ignored by the
thread participants. It is difficult for a proponent to discuss a
proposal in this atmosphere, and it's also difficult to extract
objections to and constructive criticism of proposals. Participation
in this emotionally charged of an environment frequently leads to
burnout, thus raising the risk that the available number of volunteers
will drop below what's needed to keep this system running.
Additionally, news.groups has traditionally served as the training
ground for new volunteers, but an angry and confrontational atmosphere
is more likely to drive potential volunteers away, making it difficult
to find new volunteers when the current ones inevitably move on.
This atmosphere also has a more subtle negative effect. It selects
for people who can work in an environment of frequent public attacks
and further cultivates the necessary attitude. This leads to a
concentration of participants who expect harsh discussions, frequent
flames, and personal attacks and who therefore have aggressive
personal filters, an instinctive defensive emotional response, and a
willingness to quickly stop listening to people who are perceived as
abusive. Not only does this create a self-perpetuating emotional
intensification of the posts (one natural response to this sort of
atmosphere is to try to be even harsher and even more dramatic in
order to be heard over the background noise), it makes it difficult to
de-escalate discussions and find legitimate disagreement under the
emotional presentation.
This effect hits everyone to some degree, no matter how experienced
with Usenet, and affects those who feel obliged to participate more
than others. It poses a direct risk to the Board's continued ability
to evaluate proposals, both through difficulty in obtaining
high-quality input to that decision-making process and through
difficulty in completely separating decision-making from negative
emotion and reaction to the discussion atmosphere.
I don't know what can be done about this risk. I am deeply concerned
that unless it can be corrected for somehow, no newsgroup creation
system that uses public input will survive. I don't believe that the
Board can single-handedly fix it, but they will have to address it
somehow going forward. Unfortunately, most of the possible solutions
that have been discussed over the years either decentralize the
conversation (with a resulting loss of ability to recruit general
volunteers and a lack of an overall view of the Big Eight) or are
directly confrontational in trying to exclude posts that contribute to
a toxic atomsphere, with all the resulting problems of impartiality,
personal animosity, and continued necessity of confrontation.
No Voting System
This new system contains no inherent public voting system, either for
groups or for Board members.
The lack of a voting system for groups poses challenges for the type
of proposal that the Board has not yet handled, namely a group reorg.
For creations, the negative effects of a newsgroup creation on other
groups are generally negligible or at the least possible to overcome.
For removals and inactive moderators, the correct choice of action is
normally obvious and one can afford to be conservative. Reorgs,
however, are one of the few places where a yes/no vote has clear
advantages and measures input that is quite valuable and useful.
Furthermore, it's hard to justify group renamings or removal of groups
that are currently used without a clear public mandate to point to.
Right now, the Board does not have a system in place to take such
votes, which may pose problems should such a proposal be presented.
The lack of public voting for Board members creates other problems.
First, without a public election, the Board lacks a clear public
mandate. It may drift away from the goals of the general user
population of the hierarchies due to the lack of clear and unignorable
public feedback. Votes provide a valuable and unambiguous evaluation
point that is difficult to arrive at any other way. A working voting
system often produces outcomes that are quite surprising to someone
who had reviewed only the public discussion.
Second, since the original introduction of votes on Usenet proposals,
votes have had the valuable effect of clearly concluding an argument.
Most people have an inherent respect for the popular vote and will
accept that they're in the minority if they lose a vote. This effect
had been undermined by the successful manipulation and gaming of the
voting system, but it was still present to a degree. Without a voting
system, the Board loses the aid of a valuable system for terminating
debate and getting people to move on to other questions.
Finally, public elections would cycle new volunteers into the Board.
This has both positive and negative effects and can cause serious
issues if new volunteers aren't willing to compromise and form
consensus with existing members, but without some system to do this,
it is very difficult to replace volunteers faster than the burnout
rate or to bring enough volunteers up to speed to create a
self-sustaining organization.
Little News Administrator Involvement
Lack of direct feedback from news administrators has been a problem
for the Big Eight newsgroup creation system for many years and
continues to be an issue under this new system. News administrators
are in some respects the primary consumers of the output of the
newsgroup creation system. If they do not act on control messages or
group changes, there's little point in making them. However, as
mentioned above, most news administrators appear to simply not be
interested in participating. As a result, any group creation system
has a significant risk of going off in directions that news
administrators do not actually approve of, thereby hurting the
usefulness of the system for its primary audience.
Next Steps
This is not the opening of an argument; rather, it is the conclusion
of one that began about a year ago. It is my position paper on all
that has been discussed since then. It is not, at this point,
something that I intend to discuss further, beyond any necessarily
clarifications in areas that are significantly unclear (if any).
Implementation of this decision is effective tomorrow. I am
completely leaving a decision-making role in Big Eight newsgroup
creation as of then, and will be unsubscribing from news.groups
shortly. I do plan on continuing to provide purely technical
assistance to the Board, both as part of the ongoing transition of
technical capabilities and as the ongoing maintainer of the
ftp.isc.org archive and backup maintainer of the moderation forwarding
database. However, whatever involvement I have in newsgroup creation
going forward will be limited strictly to my professional role as a
news administrator for one university site, and even that I plan on
limiting sharply for the foreeable future.
How to Object
If, after reviewing the current procedure, you have concerns or
objections, I strongly encourage you to talk to the Board about them.
I believe that every person on the Board is a reasonable, approachable
person who will discuss concerns in a productive fashion. I have
known many of the people on the Board for years and have had the
chance to observe their interactions in many different environments,
and if I didn't hold this opinion, I wouldn't be handing the system
over to them. I believe that if you extend to them the presumption of
good will and recognize that the system they came up with came from
months of difficult discussion and is supported by reasons they
believe in (and therefore is unlikely at this point to change
quickly), they will return that presumption and will try as best they
are able to find workable compromises.
As with all such discussions on Usenet, firm facts are thin on the
ground. If you can provide concrete information, measurements, data,
or the means for acquiring them, your concerns and objections will be
much easier to respond to and far more persuasive.
If for whatever reason the above is untenable or you cannot reach an
agreement you can live with, the last resort is to start publishing a
separate newsgroup list and issue separate control messages. I don't
consider this sort of further fragmentation of the Big Eight newsgroup
list to be a good option, but if I'm completely wrong and the Board
acts in some way seriously detrimental to Usenet, it's the recourse of
last resort. In such a situation, I do believe this last resort could
be exercised effectively. In some ways, it would be easier now than
it was in the past, given that the Usenet readership is increasingly
concentrated at a few large sites.
I don't recommend that anyone take this approach, but since it exists
to some extent as a check on our evaluation abilities, I believe it's
appropriate to make it possible. Therefore, if you want to start your
own system or your own ftp.isc.org-style archive, my long-term
intention is to make the software that I have used publicly available
on my web site. Until such time as I have a chance to do this, feel
free to send me e-mail directly and ask for it. I cannot provide help
with customizing it for your purposes, and it will require
customization, but I can provide a starting point.
How to Help
If you want to make this new newsgroup creation system a success,
again, I encourage you to contact the Board and volunteer. There is
always more work than there are people, and there is work for a wide
variety of different skill sets. If you are a news administrator, I
am quite certain that any input you can provide on what sort of
newsgroup creation system is the most helpful to you would be greatly
appreciated.
Finally, everyone can help ameliorate the greatest risk for any Big
Eight newsgroup creation system by being patient and constructive in
news.groups. Try to extend a presumption of good will. Try to make
any reply less of a flame than the message to which you were
responding. Try to understand the other person's perspective, or
failing that, at least accept it. If it doesn't feel right to support
someone in public (sometimes it escalates matters), send private
e-mail to people who say things well, or who do a good job at the
above, and let them know that it was noticed, at least to those people
who welcome private mail.
It's difficult but not impossible. And if enough people are working
at this, it creates a positive reinforcement cycle and starts to build
a community. The reconstruction of such a community would be a
wonderful step for Usenet as a whole.
-- Russ Allbery
September 30, 2006
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (GNU/Linux)
iD8DBQFFHv+i+YXjQAr8dHYRAkfOAKDah6FRZZgDz/N+ujtvuxD5u0fxpACfbJ/E
1fgKNX0hYSl6IqNz74FaQ+8=
=cQ3q
-----END PGP SIGNATURE-----