[Tfug] Any SQL gurus out there?

[Jim March]

Ronald,

Some of the best ideas in improving voting machines involves improving
the oversight on the final data.  Done right, you don't need to be
quite so cautious about the code.

For example: do a paper ballot of the "fill in the bubble" sort, and
then scan it.  The trick is, the scanner station doesn't have OCR and
hence doesn't know what the votes really are.  It just saves graphic
snapshots (hashed to a fare-thee-well) of ballots, and saves them to
CD-R, DVD-R or similar write-once media.

(This is an example of "artificial stupidity" - the scan station
doesn't know enough about the ballots to be able to cheat, or be
commanded to cheat.)

A separate system owned by the county does tabulation (interprets the
results).  But extra copies of the scans can be handed out to anybody.
 Imagine on election night, the various news outlets bring along their
most potent gamer-class rigs running open source software to tabulate
the data based on the scans - fast so as to chew through it before the
other media outlets.  Hacktivists with the same data would be clunking
along behind on likely slower rigs but just as accurate and with
possibly different code.

So at the end of the night, we get to see if the county's tallies, the
media's tallies, the political party's tallies and the hacktivist
tallies all come reasonably close to matching.  If they don't, OK,
let's figure it out.

We're no longer slaved to one "monoculture" screwed up vote tally
application by somebody like Diebold.

BUT, we're not going to get to a point like that unless we tear down
the existing mess - and that means proving they've been used for fraud
and/or proving that vendors such as Diebold have played seriously
fast'n'loose with the existing rules.

Jim