The Providence Journal has the story of a very close primary in our neighboring state. There are some interesting lessons here. <read>
First there is a reason CTVotersCount wants manual hand-to-eye counts in close races, rather than reading the votes through a machine. Primarily (no pun intended) any machine errors will likely just be repeated. Also, CT like RI is a voters intent state – we count votes as the voter intended, not as the machine requires – so counting by hand in close races is the only way to really determine voters intent in races where the margin is very small. Lets not be like RI and depend on courts and candidates to do the right thing:
The amended court order mirrors one issued Wednesday in the case of a candidate for state Senate in Warwick, David Bennett, who lost his bid for a place on the November ballot to Erin Lynch in the Sept. 9 primary.
Both orders require the recounting of all provisional, mail and regular ballots cast. In addition, the court ordered the state Board of Elections to examine any ballots rejected by the optical scanning machines to determine which candidate, if any, the voters intended to choose but failed to mark the ballot correctly.
Actually the RI court should have gone further and ordered a full hand count — just because the machines read the ballots, it does not mean they read the voters intent accurately. Having observed 16 post-election audits in Connecticut, from first hand observation, I can attest that sometimes an optical scan ballot is such that the circle that is filled in is crossed out, but the machine cannot determine that. There is also the unlikely possibility that an optical scanner might just not work well. OK so its not so unlikely anymore, given recent tests in a real election in Washington, D. C.
Back to RI, it seems there was also a bit of voter fraud, with a good evidence trail. The only question remaining is, will it be prosecuted? Remember the Attorney General scandal? It was all about prosecuting non-existent voter fraud.
During a hearing involving the Alves protest, West Warwick’s deputy town clerk reported to the state Board of Elections that 13 Republicans had voted in the Democratic primary. Taveras reviewed each of the nearly 2,000 ballot applications, finding the signatures of two more Republicans who apparently voted. Taveras also told the board that there were three more Democratic ballots cast without corresponding ballot applications, creating a total of 18 questionable ballots.
In the Bennett hearing, Taveras said that his comparison of numbers from the Board of Elections and from the Warwick Board of Canvassers showed that 31 ballots were cast without signed application forms. He said that as many as 19 Republicans voted in that Democratic primary.
“This isn’t an isolated mistake,†Lynch, the chairman of the state Democratic Party, said. “There are [many] separate instances where Republicans signed for Democratic blue ballots and voted in the Democratic primary. That’s no mistake. 
However, Lynch said the law isn’t as lenient. Rhode Island law considers it a felony to vote in any election that an individual “knows or should know†that he or she is not eligible to vote.
“I think if you’re a registered Republican, you know enough not to vote in the Democratic primary,” he said. “I know enough not to vote in the Republican primary.”
******
Update 06/06/2009 Washington Post: Firm to Give D.C. Information About Its Voting Devices <read>
Sequoia agrees to hand over source code and documentation under rasonable terms. This should be the normal response to every voting discrepancy traced to a voting machine. It should not require going to court.
Sequoia Voting Systems agreed yesterday to turn over sensitive information to the D.C. Council about how the District’s voting machines work and tabulate results, setting the stage for one of the most comprehensive probes on the reliability of electronic voting equipment.
The agreement is a response to the election night chaos in the September primaries, when Sequoia machines tabulated more ballots than there were voters, resulting in thousands of phantom votes.
Electoral change advocates said the agreement, finalized yesterday in D.C. Superior Court after the city threatened a lawsuit, is one of the first times a manufacturer of electronic voting machines has been forced to endure a public vetting of how its equipment tabulates returns.
“It is certainly going to serve as a precedent not just for further investigations in the District of Columbia, but around the country,” said John Bonifaz, legal director for Voter Action, a national voting rights organization.
close vote recounts… Connecticut uses a combined hand and machine recount. Each ballot is reviewed by at least two officials with a Moderator to settle any disputes, all under public observation. Only those ballots agreed to be ‘clear in intent’ are put through the tabulator. All others are hand counted. And of course, all ballots used in the election are recounted.
There will always be the potential for differences in machine and hand counts. If a stray mark hits a circle, and an overvote is created, the tabulator will pick this up and reject the ballot, allowing the voter to correct the problem, but only if the voter marked another circle in that office. But if a voter uses x’s or circles which may or may not hit the mark, then some voter intent will be lost, since the tabulator has read and recorded valid votes on the ballot (undervotes are not a reason for rejection). If a voter uses a marker than the one supplied in the booth, then the reader might not be able to read all the markings.
The case of non-party voters crossing into the other party’s primary is a result of the use of ‘open primaries’ where a voter declares his party choice at the polls on election day. This creates opportunities for much shenanigans and many mistakes by once a year polling place officials, rather than by trained voter registration officials working in offices with greater resources.
This is the main problem I have with open primaries (one of many problems…), and with election day registration at the polls. While Connecticut alolows party enrollment up to the day before the primary, there should be a reduction to 30 days for switching parties from the current 90 days. I agree with the delay in principle, which was created to minimize the abillity of a party to ‘stuff the ballot box’ of another party’s primary.
Now as to Audits……
The problem with machine recounts are:
1) Any error or fraud in one memory card or the software in the scanner is likely to be in all, so any counting errors in the machine will occur again on the recount.
2) We could have a problem like they have in D.C. where they so far have run the same set of ballots thru the machine three times and have gotten three different results.
As far as I know there are as yet no recount procedures such as you imply.
The latest procedures are:
http://www.sots.ct.gov/sots/lib/sots/electionservices/misc/2007recanvassmanualpdf.pdf
See Section A. 1. (h) it still says recount all the ballots by hand, yet we know that the Secretary of the State has said all recounts are now hand and machine, but as yet has no procedures.
If I am wrong, please correct me.
In a sense we are both correct.
The new regulations do call for a machine recount – and worse than I thought it requires they be counted through the very same tabulator as originally used, so if that tabulator is poorly adjusted or has compromised software it will presumably make the same mistakes it did originally.
However, the Recanvass Manual on the Secretary of the State’s web site calls for completely hand recounts.