ekr at rtfm.com
Mon Jun 23 07:42:02 CEST 2003
Melinda Shore <mshore at cisco.com> writes:
> > > but it *doesn't* prevent a WG chair from declaring rough consensus.
> > It shouldnt. But it does.
> Then isn't that a problem with how our chairs operate? I
> think that to a very great extent working groups face a
> prisoner's dilemma situation and not all self-interest is
> enlightened. There are countless, I suspect, aspects to
> this problem, ranging from consensus process really being
> fairly ill-suited to situations in which there's not a
> shared commitment to the process itself to working group
> chairs not being sure how much authority we've got. I think
> it's extremely useful to talk about how our processes are
> failing to deal with participants who refuse to take "no"
> for an answer, but for the purposes of the working group and
> its follow-on (the solutions group) it's more useful to
> focus on how the problem is instantiated within the IETF and
> less on the fact that there are some people who are just
> plain difficult.
> That's a windy way of saying that difficult people aren't a
> problem if we've got mechanisms to get our work done anyway.
I think this is a useful formulation of the problem. Unfortunately,
it also leads to some pretty depressing conclusions.
The game theoretic aspects of parliamentary and voting systems have
been pretty extensively studied (see Couter's "The Strategic
Constitution" or Buchanan & Tulock's "The Calculus of Consent") and it
turns out that a number of the nastier aspects of politics are pretty
much the immediate consequence of people pursuing their own
interests.  In particular, systems that require unanimity and near
unanimity are highly susceptible to participants "holding out"
in hopes of getting a better deal, which is of course what we're
It's not known, even in principle, how to get rid of these problems,
which is why pretty much all existing political decision making
systems have them.
 Yes, I know that we talk about the "good of the Internet" as the
goal people are supposed to shoot for, but I don't think that's a
particularly useful model of what's happening here. First, people
often don't agree on what that is and so are pursuing divergent goals,
for whatever reasons. Second, I think it's pretty clear that in a
number of cases people are in fact pursuing their own interests.
[Eric Rescorla ekr at rtfm.com]
More information about the Problem-statement