Update: More lessons from the Wallingford Recount <read>
This was a town council race with 17 candidates and 9 elected. 11,000 votes 59 poll workers counted in 6 hours.
If we assume a $0.05 per vote (9) that would $0.45 per ballot. If so it works out to paying an average of $14.00 per poll worker per hour. I don’t know what they were paid, but its in range of best practice claims.
Also we see that there is need for public education about what to expect for accuracy. Counting each vote per voter’s intent vs by the machine’s reading ability should yield slightly different results — the very reason why close races should be recounted manually. The machine will never be able to judge voter’s intent perfectly.
We have repeatadly warned that claiming the the machines performed “flawlessly” or “perfectly” sets up false expectations.
_____________________________________
The East Haven re-recount ended yesterday on a reassuring note. The problem apparently being a human error in the recount process itself. East Haven Politics has the story and videos <read>
What we might learn from this:
- It would be a good idea to publish detailed recount procedures and develop best practices. We do not have to reinvent the wheel.
- One of the advantages of selecting Dieblod/LHS optical scanners, often cited by the Secretary of the State, is that they have been in use in other New England states…so we can learn from how they perform recounts and audits efficiently.
- PEW is funding research into best practices for manual counting of ballots, hopefully, that study will be complete and available in time for consideration in creating procedures in Connecticut for the November 2008 election.
- The legislature can also fund research into best practices as requested in the petition on this site: “Request that you fund research into mechanisms for counting votes, looking carefully at practices employed elsewhere that indicate an average cost of $0.04 per ballot per race counted are attainable.”
- A lot of negative publicity and drama can be avoided by semantics: When counting shows huge discrepancies, count again until at the minimum the same result is produced at least twice. You do not have to announce a huge discrepancy as the result of a recount until you verify it. What caused all the excitement here was announcing the result before verifying it.
- Its a good idea not to jump to conclusions, especially those that blame election officials and the voters. As suggested on this site, it was highly unlikely that more than 100 voters took two ballots by mistake or that several ballot clerks in one municipality each lost track of ballots or that checkers made a significant number of errors. As we said here “We don’t know what happened. It could be human error, but more likely one or a sequence of two or three errors rather than 130 errors by multiple people in multiple places.”
Finally, East Haven Politics quotes a figure of $15,000 for the initial inaccurate recount. I don’t know where that figure came from, however, it seems wildly out of line. Best practices in New Hampshire indicate that $0.04 per ballot per race counted is obtainable. That may take some doing, however, $15,000/8,000 votes yields about $1.87 per vote.
Update: My answer to a question posed on my cross post on MyLeftNutmeg
greenpeas: I don’t see any way out of saying there’s a discrepancy
The press is waiting for a result and the registrar could not announce one the day of the first recount, nor did she want to announce a recount without making sure she had to do one. I think we should encourage, not discourage, coming forward to provide open and transparent access to information about what is going on so people have confidence in the process. finding mistakes and correcting them will be part of getting the process right, and se should both try to vastly improve our overall processes as well as inform the public what it will look like to witness this process and what is normal/not normal about sorting out a manual count.
I think the problem was not the discrepancy announcement; it was the lack of SoTS office involvement that would have allowed the registrar to either to confirm the vote achieved or to confirm the next step to be taken.
It would also be helpful for the SOTS to explain what voters should expect on recounts — perfect recounts? discrepancies? What margin of discrepancy is acceptable to the SOTS office?
We may have to look to other states and cumulative studies to figure this out, since as we know, in CT everything is coming up roses.
Luther Weeks (aka BlastFromGlast) I still think that the announcement could be different
Such as “We have all had a long day, the initial count we have is way off. What we need to do is secure everything, get a good night’s sleep and comeback again and carefully count again, until we get the same results twice. Perhaps we counted some votes more than once or added incorrectly of transposed some digits, or perhaps it is more serious. Right now we need to relax and get some sleep and continue deliberately”. As I have said there is a need for more detailed procedures. That should be done by the SOTS and/or ROVAC (Registrars Of Voters Assn Connecticut). Yet, counting again and simply counting the number of ballots to balance with the result, should only require common sense.
We have heard over and over from registrars in the last few months that going all paper would be bad, for one reason because of the difficulty of counting the paper and getting it right. Here we have an example of that but suddenly normally rational people find it surprising that we don’t count the paper correctly the 1st time.
