[ERR5RS] Wiki voting mechanisms

[Harold Ancell]

I agree with Victor that we need to take a step
back and discuss the process we're going to use
before we get to details like how to do votes in it.

comp.lang.scheme has recently hosted a very
interesting discussion on the history of Scheme
standard decision making.  We need to think about
this, and the sooner we achieve some common
ground, the better, in terms of getting people to
buy into the various efforts we are doing.  You do
NOT want to decide on a process *after* people have
put a lot of effort into something.

Note that we (the greater community on this wiki)
are thinking of at least two "products", a limited
technical effort "Standard 1" and a "Standard 2"
of significant ambition to follow.

The first is intended to be of limited scope (not
a R7RS or alternate to but of the same level of
ambition as the R6RS), and is intended as much to
establish this community as it is to produce
something useful, since everything including it
depends on the former.

With Will's challenge to us, ERR5RS will be
Standard 1, and of course other efforts can
proceed in parallel on the wiki.

I'll note there also might be a separate effort
for defining or more likely blessing libraries to
declare them as "standards", and providing
packaged reference implementations.  I say
separate because I and others think they don't
need quite the same process, there is of course
the SRFI effort to mesh with, Marc Feeley has
invited us to use Snow for packaging reference
implementations (something we've already thought
of; see c.l.s), we need to establish our "Scheme
cred" to have anyone listen to us, etc. etc. etc.

As to process, I'll start with a few observations:

The stark object lesson that has brought pretty
much all of us here:

  Date: Sat, 08 Sep 2007 14:53:11 -0700
  From: William D Clinger <cesura17 at yahoo.com>
  Newsgroups: comp.lang.scheme
  Subject: Re: R6RS Ratified

  Andre wrote:

  > Who is to blame for not trying to find out?
  > Several people who voted no argued, even
  > before the vote, against the arbitrary cutoff
  > date that has left the R6RS in an unfactored
  > state of lower quality than prior reports.

  Look on the bright side: We need no longer
  speculate about what would happen to the technical
  quality of Scheme reports if the required
  supermajority were reduced from 100% to 60%.

  Will

 From what little I've observed, the old wisdom in
regards to the then newish US Constitution is
absolutely true: while a process cannot assume all
men are angels, no process can overcome imperfect
humans.  People with the wrong character for the
job using the best process in the world will not
deliver a good result.

The "traditional" "100% consensus" R2RS-R5RS
decision making process was considered post-R4RS
to be flawed, at least for what Scheme needs now,
if for no other reason than it has not produced
something that people are universally happy with
after the R4RS, or anything at all after the R5RS.

Democracy, a dictator, an oligarchy, a wiki
anarchy; we have many models from history to
choose from.  I'm not sure what is best, except
that "radical democracy" (everyone who shows up
gets an equal say in a simple majority vote) is
NOT the answer.  Not for something technical
like ERR5RS.

                                - Harold