Eastlake's Observation

An absolutely brilliant and succinct description of large social processes is below. For a more modern and functional take, and some practical solutions, see this lovely PDF by CAE (Critical Art Ensemble), Join the Community.

Subject: sociology
Date: Thu, 4 Mar 1999 13:50:54 -0800 (PST)
From: Randy Bush
To: Tom Jennings

From: "Donald E. Eastlake 3rd" 
Subject: Re: Draft new charter
Date: Thu, 04 Mar 1999 00:19:20 -0500
Message-Id: <199903040519.AAA03648@torque.pothole.com>

In connection with Dave's request, I would like to point out that
social systems do not scale linearly.  In particular, in the matter of
dissent, some studies have been done.  It is generally the case that a
single dissenter will give up quickly while two mutually reinforcing
dissenters will be much more stubborn.  This continues to increase
although beyond 3 or 4 there is little change.  This means that as a
group gets larger the probability approaches one that for any
majority/consensus position, there will be a self reinforcing
minority/splinter that will complain and fight the consensus more or
less forever.  Of course, the IETF may have attracted more that it's
share of individuals who will stubbornly persist even in the absense
of anyone supporting them but even there, you have somewhat the same
effect.  While strong clearly thought out technical dissent has its
place and may force the consensus to sharpen its arugments and
document more of its assumptions, beyond some limit it is counter
productive.  And even a single lunatic can produce a lot of noise and
it takes just two of them, whether they agree or disagree, to converse
at infinite length on a mailing list.  While one luantic can be fairly
easily filtered or ignored, by the time you have four or five of them
chatting they generally clog things up and create a high percentage of
the traffice and threaten to down out most participants.
-- my emphasis (tomj)
I'm not sure what the answer is but in the real world, the more people
you have, generally the stricter the controls and the greater the
contingency planning.  Most people don't worry much about a party for
20 or maybe even 200 people.  If you are running an event and having
20,000 or even 2,000 people at it and don't have a paid police detail
or security personel and/or fire marshal and/or emergency medial
personel on duty, you are at risk for negligence.


PS: Another fascinating result of social research is that essentially
all organizations of more than about 15 people are percieved as having
a small group of insiders that control things.  It turns out that the
extent of this perception and the feeling of the average member about
it are relatively independent of how open or closed the organization
is or how much power whatever leaders there are actually have.  As a
result, almost all organizations have periodic attempts by "rebels" to
take over from what they perceive as an entrenched old guard.  Only
something like 5% of such attempts are successful on the average, and
when they are, generally within 3 months or so, the "rebels" are
perceived as the new insiders.