Vote machine operation counter to instructions [FL-13]
Mon Jan 01, 2007 at 09:58:22 PM PDT
In the disputed race between Christine Jennings and Vern Buchanan —the two are separated by 369 votes in their claim to a seat in the US House of Representatives— the "iVotronic" touchscreen machines at the precincts did not alert voters to undervotes, and did not ask if they wished to input a vote choice in a race that was not yet recorded.
Yet, that operation at odds with specific printed instructions on the website of election supervisor Kathy Dent (R) that showed voters the sequence to expect in casting a vote on Sarasota County's iVotronic, sold by Election Systems & Software (ES&S). [Screenshot here - scroll; the link is at the jump.]
18,000 voter records registered no vote for a congressional seat, 13% of voters [1 of 8] who walked into the booth, yet only 1.2% of Sarasota absentee (paper) ballots cast no vote in that race. >> More people had a vote recorded for "Hospital Board southern district" than for the Jennings-Buchanan race to replace Katherine Harris.
The link is at the website of Sarasota Dept. of Elections run by supervisor of elections, Kathy Dent. You can produce the instruction shown above at the site if you hover your mouse over the "vote button" circled as feature #5, then click on it. Or, you can just scroll to the bottom of the website page and get to the information.
Note the language, "If the undervote question is answered "NO," then the voter will have the opportunity to return to the ballot and make a choice."
The credit for finding the supervisor of elections instruction goes to bradblog.com. [Visit Brad's amazing site coverage of voting machine controversy; he has a similar screenshot from the Sarasota County elections website.]
According to a petition filed in 2006 to the US Supreme Court, touchscreens programmed to comply with Florida law notify the voter of an undervoted race: "With respect to
undervotes, touchscreen machines only permit an undervote
after notifying the voter at least once that he/she has not
selected a candidate for a particular race as well as giving
the voter an opportunity to review the selections before
casting his/her ballot."
Florida statute covering the "electronic voter interface" (video display ballot for a typical voter, audio ballot for a sight-impaired voter) says much the same. [link, appendix definitions and pp. 20-22]
The system must communicate to the voter the fact that the voter has failed to vote in a race (under vote) or has failed to vote the number of allowable candidates in any race (under vote) and require the voter to confirm his intent to undervote before casting the ballot.
Despite these supposed required procedures, why do ES&S machines spit out such massive undervotes? Other nearby counties experienced problems in other races, but less attention is paid because the races weren't close.
Here is a letter from a Port Charlotte voter (Charlotte County, Fla. - also using iVotronic touchscreens by ES&S) to the Sarasota Herald Tribune [registration required], December 5.
The Nov. 7 general election in Charlotte County was not as smooth as some would like the public to believe. My vote for Walter "Skip" Campbell in the important race for attorney general was not recorded on my ballot. I reviewed the ballot and repeated the vote. Was it really counted? I'll never know. The same thing happened to my friend at another precinct, and to many more voters. ....
Campbell's race had ... 4,173 undervotes on the early-election touch-screen machines, compared with 595 and 904, respectively, for the chief financial officer and commissioner of agriculture races. That indicates a serious problem.
Likewise, on Election Day, the attorney general race had 7,196 undervotes, compared with 1,101 for CFO and 1,712 for commissioner of agriculture, again indicating a serious problem with the machines.
We deserve better in Charlotte County. Don't let Christine Jennings' stolen votes in Sarasota County overshadow what happened here.
Joan Fischer
Port Charlotte
The number of missing votes in the AG race are not enough to overturn an election, so it won't pull in much notice.
But the problem of dropped votes is the same.
Errors in recording and counting votes with ES&S touchscreens are legion. When Wake County, NC tested 6 iVotronic voting machines at 2 poll locations in 2002, the machine counts indicated votes were dropped. ES&S had to input new "firmware" to remedy the problem, but the elections director Cherie Poucher decided to keep optical scan voting for future elections instead of touchscreen. Her letter to ES&S requested reimbursement for costs to contact voters and give them a new chance to recast their lost vote.
ES&S reimbursed Wake County. Her letter to ES&S, Jan. 28, 2003, explained the situation:
Re: Test of i-Votronic
Dear Ms. Emerson:
The Wake County Board of Elections tested the i-Votronic direct record equipment at two additional one-stop voting sites for the November 5, 2002 General Election.
On October 21, 2002, the one-stop supervisor in the Cary site advised me that the total votes cast on the six units did not add up to the total number of one-stop voters. We learned that the situation was the same at the Northwest Raleigh one-stop site. At that time, we pulled the equipment from the sites since there appeared to be a problem with the vote totals.
We immediately called ES&S, who then notified Jim Johnson the technician. Mr. Johnson called our office and requested that Wake County’s technician, Sherwood Brantley, verify the firmware version that was installed in the units. It was agreed that ES&S would retrieve the audit data, to allow us to identify which voters’ votes were not recorded. This step was crucial and needed to be done immediately to maintain the integrity of the election process. Unfortunately, there were
many problems from that point on and data could not be retrieved until October 24.
On Thursday, October 24, Al Marcheski had informed me that ES&S knew there was a problem with the firmware. I later learned that Jackson County, North Carolina had the same problem with their equipment several days prior to our problem being discovered. The firmware in their equipment was changed and their problems resolved. Had
ES&S been aware of the firmware in use in Wake County and acted immediately when the problem in Jackson County was corrected, Wake County would not have experienced the problem of votes not being recorded.
It was imperative that the Wake County Board of Elections immediately notify the voters affected by the votes not being recorded and offering them the opportunity to vote again. The priority of staff at that time was to contact by phone as many of the voters as possible. Also, a letter was mailed to all affected voters listing his/her options.
The Wake County Board of Elections staff hand delivered ballots, Federal Expressed ballots, and faxed ballots to voters to maintain the integrity of the elections in Wake County.
Listed below is the cost to the Wake County Board of Elections to rectify the situation created by the failure of the firmware to record all the votes:
Staff time to determine list of voters affected; to obtain telephone
numbers; to make calls to the affected voters; to process
letters, faxes, etc. to the affected voters $5,590
Letterhead, envelopes, postage, Fed-X charges, mileage for
ballots hand delivered $ 408
TOTAL $5,998
We respectfully request reimbursement for these expenses. The check should be made payable to
the Wake County Board of Elections and mailed to our office at PO Box 695, Raleigh, NC 27602.
We expect payment within the next 30 days.
Thank you.
Cherie R. Poucher, Director
Wake County Board of Elections
Cc: State Board of Elections
Other vote count mishaps, if you would like to see a sample, are shown on this map graphic [pdf] from 2004, from the votersunite.org website. It has errors found for ES&S, Sequoia, Diebold and Unilect.
Really, if you're going to get to the bottom of any count discrepancy, you need paper to count it.
Permalink | 44 comments