Since Nothing is being acted on in Congress before Nov 7

[rumpel]

All available precautions need to be put in place by the citizens - I believe, we are now on our own.

As for Los Angeles County, what is different from the primaries is that McCormack announced the precinct level "ballot reader":
http://www.inkavote.com/precinct_ballot_counter.htm
sounds good on their site in terms of checks - but once my vote is "digitized", well it's now part of computer data. It does not tell me whether my choices are in fact recorded as such. Another added source of malfunctions, errors and wrong software configuration.

Poll watchers have to be vigilant - armed with a log file
Irregularities reporting have to be easily accessible and the existence of such have to be widely publicized leading up to Nov 7- in all languages

I am not sure whether this audit formula by uscountvotes has been posted here. Sorry if this is a dupe, but I also wanted to put it out here for possible discussion, as an available tool immediately after the elections, since we have no choice but to deal with these machines.


The Election Integrity Audit, published 9/25/06

This paper focuses on manual audits to detect
discrepancy between electronic and manual counts sufficient to alter election outcomes.

Excerpt:

ABSTRACT
This paper explains how to directly calculate a sufficient number of electronic vote counts (either precinct
vote counts or DRE machine counts) to audit – count manually - in order to detect an amount of electronic
miscount that could wrongfully alter an election outcome. Vote count audits of fixed 1 or 2% of electronic
counts provide no confidence that outcome-altering errors would be detected in close races.
This new audit calculation method builds on the National Election Data Archive’s (NEDA) prior work on
the mathematics of vote count audits and the Brennan Justice Center’s threat analysis which assumes a
maximum amount of miscount that could occur in each electronic vote count without raising immediate
suspicion and investigation.10 Thus:
· The minimum number of corrupt electronic vote counts that could wrongfully alter an election
outcome depends on the margin between candidates and the assumed maximum miscount rate in any
vote count, and
· If we solve the equation for calculating the probability of detecting one or more vote miscounts for the
audit sample size in terms of the desired probability, the total number of vote counts, and the number
of corrupt counts sufficient to wrongfully alter outcomes, then we can determine the audit sample size
that will ensure election outcome integrity.
The “election integrity” probability equation cannot be solved by hand, but it can be solved by computer. In
this paper, we provide a numerical (computer) solution to directly calculate vote count audit sample sizes to
detect electronic errors to any desired probability that could have altered election outcomes.11
Success in detecting evidence of outcome-altering electronic vote miscount via audits is influenced by the
following relationships:
· As the margin between candidates decreases, the amount of vote miscount that could wrongfully alter
an election decreases and the audit amount must increase.
· For a fixed rate of corruption, as the total number of vote counts to be audited decreases, the required
audit amount does not similarly decrease. i.e. As the number of total vote counts decreases, the audit
percentage required to keep the same probability of detecting errors increases. For example if one in
ten electronic vote counts is corrupt, then at least ten vote counts must be audited in order to have a
good chance of detecting one corrupt count, regardless of how many total counts there are.
· When auditing optical scan precinct counts (as opposed to DRE machine counts), if precinct sizes vary
significantly, it becomes possible for miscounts to be targeted to a fewer number of large precincts,
and larger audit sample sizes are needed.
· When auditing DRE machine counts variability of the number of ballots cast on DRE machines may
have a similar effect (to the effect of precinct-size variation on optical scan audits) on the probability
of detecting outcome-altering miscount. The conservative approach would be to use the number of
ballots cast on DRE machines to adjust audit sample sizes but research is needed.
· The total number of electronic vote counts in each race, rather than the total number of vote counts in
the entire jurisdiction, determines the sample size that must be used to audit that race outcome.

http://electionarchive.org/ucvAnalysis/US/paper-audits/ElectionIntegrityAudit.pdf
document = 21 pages, including appendix

[OnTheOtherHand]

1. well, VoteTrustUSA published basically the same idea in August

http://www.votetrustusa.org/index.php?option=com_content&task=view&id=1664&Itemid=26

In a newly released paper, "Random Auditing of E-Voting Systems: How Much is Enough?", Howard Stanislevic argues that "Auditing protocols proposed and implemented at the federal and state levels that rely on small-percentage random sampling without replacement are unlikely to detect miscounts sufficient to change the outcome of Congressional or smaller local races, even if such races initially appear to be decided by relatively wide margins."

As far as I can figure out, Dopp and Stanislevic were working on basically the same idea separately.

That aside, any precincts people get to audit after the election, it's gravy. I agree with you about poll-watching, etc.