McCormack Hit Job Video on VVPAT
So, it appears that the Registrar-Recorder of LA County, Conny McCormack, sweet-talked her husband, Austin McCormack of GEO Corporation, into doing a hit-job video on the Voter-Verified Paper Audit Trail (which I've started to call the Voter Verifiable Paper Audit Trail).
Here is a link to the video hosted by VerifiedVoting.org: "First Roll-Out of VVPAT Equipped Electronic Voting in a Major Election".
In the rest of this post, I'll post my transcription of the video:
[More:]
Opening screen: "first roll out of VVPAT Equipped Electronic Voting in a Major Election"
E-Day, Nov. 2, 2004. Poll workers take their oaths knowing that a president will be elected today.
This is Clark County, Nevada; the site of the first roll out in a major election of electronic voting equipment retrofitted with a Voter Verified Paper Audit Trail printer or VVPAT. Under the most intense media, poll watcher and legal scrutiny, Clark County held an efficient and accurate election using a combination of VVPAT and non-VVPAT-equipped electronic voting equipment. Two different models: the Sequoia Advantage and the newer Sequoia Edge.
[In sans-serif type on full screen] Clark County has successfully conducted election with electronic voting equipment without VVPAT for nine years.
However, this year [the rest is displayed full screen] Nevada's Secretary of State ordered counties to attach a VVPAT printer to all Sequoia Edge models [not on screen] under the assumption that [rest on screen] voters would want to use it to verify their electronic selections.
As numerous observers of Nevada's 2004 elections have reported, and this video will show, [rest on screen] the vast majority of voters ignored the printed record of their votes.
While highly promoted and heavily lobbied, [rest on screen] VVPAT is an expensive addition of mechanical equipment to electronic voting that is not required under the federal Help America Vote Act or HAVA. HAVA requires "a paper record... available for an official recount." All existing, federally qualified electronic voting equipment already meets this standard without VVPAT.
Indeed, the principal authors of HAVA [shows signatures of Representatives Ney (R) and Hoyer (D) along with Senators McConnell (R) and Dodd (D)], in a March 2004 letter [March 3, 2004:
http://www.house.gov/cha/dearcolleaguemarch3-04.htmto their congressional colleagues, [rest on screen] expressed concerns regarding pending "voter verified paper record" federal legislation stating that VVPAT "... would undermine essential HAVA provisions such as the disability and language minority access requirements, and could result ini more, rather than less, voter disenfranchisement and error."Preparing for VVPAT implementation required extensive poll worker training. Let's take a look at a poll worker training class in Clark County.
Each VVPAT cartridge is a separate attachment containing [some of this is onscreen] rolled printer paper, paper advance motor, printer and glass viewing area.
The cartridge mounts on the side of the Sequoia Edge and must be plugged into the back using two cables. The Edge must be rebooted every time a paper cartridge is changed. For this election the ballot consisted of 21 races which produced a 12 inch paper printout with a capacity of about 200 ballots per cartridge. [screen displays: Each cartridge could print about 200 ballots]
Poll worker were shown how to remove and replace the cartridge. During early voting each cartridge was replaced every night requiring multiple cartridges per Edge. The VVPAT comes into play after the voter completes his or her selection. The voter is presented with a review screen, then prompted to review a paper printout of the ballot before the vote can be cast. All of these steps were demonstrated during a poll worker training class.
[poll worker trainer voice] as you get all your selections made, it will automatically take you to a review. You automatically go to review. Now, if there are any of those races you want to change, you go touch on it and it will take you back there. OK? If you don't want to make any changes, then you touch next again. And it will say 'touch here to print and review a paper record of your ballot'. You have to go through this screen in order to cast your ballot. So you touch here and it's going to print it up. All the selections you've made will print up right here. There's going to be a magnifier hanging here so that in case the voters can't read the fine print, they can always use this. It'll be hanging right on this thing here. [More... I'm not transcribing this]. Hopefully, that [making the wrong selection] won't happen a lot because that uses up a lot of paper.
We go now to an early voting site at the Meadows Mall in Clark County. Over several hours of observing and recording the actions of several hundred voters, the average voting time is 5 min 13 sec to make up to 21 ballot choices. But when the VVPAT began printing the vast majority of voters spent no time at all actually reviewing it.
Voters frequently glanced in the direction of the paper receipt once they heard the whirring of the printer's motor. A few reached out to touch the printer cartridge while even fewer leaned up closer to take a look at the VVPAT. This activity took place during the 15 seconds required for the paper to scroll up into place under the glass window not after it finished scrolling. Also, no one was observed using the magnifying glass to assist in viewing the VVPAT by increasing the size of the ten point type.
Exit surveys were completed by many voters during early voting and on election day. When asked if the paper component was important to them only 4% said that it was. Most indicated that it was not necessary or they didn't care. Some did ask why they couldn't take their paper "receipts" with them?
By watching voters' movement from the time they make their voting selections to the time needed to print the paper ballot through the time the cast ballot button is pushed, we can determine how much time, if any, is spent reviewing or verifying the paper ballot.
First, we'll show a demonstrative voter. We'll wait for the voter card to be ejected and then rewind back 7 seconds, which is the time it takes the ballot to be recorded electronically after the voter presses "cast ballot". Then rewind 15 seconds, which is the time it takes the paper ballot to scroll into place... and continue in reverse to find the point when the voter pressed "print paper ballot" and the paper ballot started printing. You can watch the voter from the time between these two points we can see how much time, if any, the voter spent actually reviewing the paper ballot. This voter did spend ten seconds reviewing his ballot after the VVPAT scrolled into place and before pressing "cast ballot".
[changes to another voter] During the time the VVPAT was scrolling into view, this voter glanced briefly at the screen. When the VVPAT finished printing, he took 7 seconds before pressing "cast ballot" and this time was spent putting things in his pocket not looking at the paper printout at all.
[changes to another voter] This voter pressed "cast ballot" six seconds after the VVPAT finished scrolling and during that time, took a look around. She did not review the VVPAT.
[changes to another voter] This voter spent 8 seconds.
[changes to another voter] This voter 6 seconds.
[changes to another voter] And this voter 7 seconds.
Hundreds of voters were observed during early voting and on election day in Clark County. The typical voter did not take time to review or verify their selections on the paper printout after it scrolled into view. The electronic review screen was sufficient for virtually all voters.
After the election, the VVPAT is retained for audits or recounts. However, recounts using the VVPAT cartridges proved cumbersome and time consuming. According to Clark County Registrar of Voters Larry Lomax who conducted an audit of the paper ballots inside 20 cartridges it required a board of 5 employees to count 75 ballots in 4 hours. To put this in perspective, to count a single precinct with 900 voted ballots it would require 6 boards of 30 employees 8 hours. This is the equivalent of 1/4 man hour per ballot or 3 times the average time it took the average voter to vote in this election with 21 possible selections.
Recounting VVPAT ballots cast during early voting on DREs in conjunction with the pilot program ran for the November 2002 election in Sacremento County, California proved even more labor intensive. Sacramento County Registrar of Voters Jill LaVine, in Congressional testimony on July 7, 2004 reported **the recount of 114 VVPAT ballots took 127 hours, approximately 1 hour per ballot due to the complexity of the long ballot for that election. VVPAT recounts in both Clark County and Sacramento County showed the electronic vote totals exactly matched the paper totals.
What was learned from observing Clark County's election introducing VVPATs? First, the addition of a VVPAT printer cartridge on the Edge equipment costing approximately $500 each adding 15% to the cost of a DRE purchase. Additionally, there is the ongoing cost of paper and storage. Clark County built a new secure vault at a cost of $200,000 for this purpose as cartridges and voted paper rolls must be retained for 22 months.
Second, most surveyed voters didn't believe that VVPAT was necessary and didn't spend time reviewing it making it a misnomer to call it a Voter Verified Paper Audit Trail.
Third, VVPATs are cumbersome and time-consuming when attempting to use them for recount or audit purposes. In their letter, the HAVA authors expressed concern that a VVPAT requirement would "... take the most advanced generations of eleciton technologies and systems available and reduce them to little more than ballot printers. ... if adopted, likely give rise to numerous adverse unintended consequences." Some of those unintended consequences occured in Sacramento in 2002 and in Nevada in 2004 including:
1) Occasional printer jams leaving the paper ballot choices exposed to repair technicians. For example, here's a poll worker in full view of voters in Washoe (sp? could be Washing) County, Nevada opening the printer cartridge attempting to unjam the paper roll at the September, 2004 primary election.
2) Additional time needed to complete the voting process while awaiting the VVPAT to print.
3) Confused voters requiring continual explanations of why they can't take their paper ballot "receipt" with them like they do with an ATM.
The question remains: Are perceived VVPAT benefits worth the cost and very real risks associated with implementation?
[On screen but not spoken] Clark County Registration: 684,313 (70% of all registered voters in Nevada)
Clark County turnout November 2, 2004: 546,858 (79.91% of eligible voters).
Early Voting: 271,465 (50%)
Election Day: 222,036 (40%)
Absentee: 53,357 (10%)
Total: 546,858
Sequoia Edge machines used: 70
cartridges required: 1090
(1 for each machine plus 1 extra for each polling location)
4,000 voted paper rolls require storage for 22 months
Produced for Los Angeles County Registrar-Recorder/County Clerk
by Austin McCormack, GOE Corporation
for additional information call (562) 462-2716
[shows seal on screen... copy of seal here:
http://dream.sims.berkeley.edu/~jhall/nqb2/media/la_rrcc_seal.jpg]
