Rob, do you care to comment on the Oakland Mayor race, using IRV for
the first time? I know the fascination right now is with the Alameda
RoV not releasing interim results... which makes total sense to me in
that IRV is a model where the outcome of an individual ballot depends
on all the other ballots. I'm just wondering if there's more
interesting aspects to that race than this one for those on this list.
On Wed, Nov 10, 2010 at 12:39 PM, Rob Richie <rr@fairvote.org> wrote:
Bottomline: We should have post-election audits of an appropriate number of
ballots to detect potential problems, but absent indications of such
problems or other evidence of fraud or election administration breakdown,
only need statewide recounts in much closer margins than the Minnesota law
of 0.5%. We won't have to do it often.
Hear, hear! And with ballot-level risk-limiting audits a la
California's recently passed AB 2023 pilot program (where individual
ballots are taken as the sample basis instead of machines, precincts,
districts, etc.) we should achieve a "holy grail" of sorts: we'll be
able to confirm very close statewide races by examining a few thousand
ballots (in statistical terms, we'll be able to soundly reject the
hypothesis that a full hand count would result in a different election
outcome without having to count every ballot).
If anyone is interested in the details, let me know and I can point to
literature. best, Joe