On Thu, 2009-02-19 at 19:00 -0500, William D Clinger wrote:
> If that precedent were continued, then an R7RS process
> needn't show much respect for the R6RS, and would need
> only a 60% supermajority in a single vote to ratify
> R7RS documents that make arbitrarily large changes from
> the R6RS (and all other documents and SRFIs etc).
>
> In my opinion, that would not be a good process for
> the long term. On the other hand, it doesn't seem
> fair to treat the R6RS the same as documents that
> have passed a higher bar. What to do?
It's an engineering specification. What to do is to judge
it on its merits. If R6 made mistakes then we can expect
slightly less of R6's new material to stand the test. But
what comes out of R7 needs to be decided by the quality
standards of the R7 process.
> That would also work, but 60% is such a low bar that
> I don't think it would work for the long term. Maybe
> changes introduced by the R6RS could be undone by a
> vote of only 60%, because that was the threshold the
> R6RS as a whole barely met, but for the long term we
> need the stability provided by a higher threshold.
There is a model used by many standards processes
where all the editors rank the proposals they think
are consistent and have technical merit. Each
proposal is usually ranked by 80% to 98% of the
committee, and "runoff voting" applies an editor's
vote only to those proposals that an editor has
ranked. It's a pretty good model, for a couple of
reasons. First is that technical problems seen by
only a fairly small minority can keep a proposal
from becoming canon. Second is that no matter how
passionate *some* proponents of some proposal are, a
different proposal can win if fewer people see a
problem with it.
When you say "a majority" you're implying a process
in which there is a simple "for-or-against" type
binary vote. It's been my experience that in a good
standards process there are usually at least a half-
dozen proposals to choose from, not just two, and
in practice having even ten percent of the editors
come out against something should assure that something
with more merit is adopted instead.
Bear
_______________________________________________
r6rs-discuss mailing list
[email protected]
http://lists.r6rs.org/cgi-bin/mailman/listinfo/r6rs-discuss
