Washblog

1. We Didn't Test the DRS with Our Mail Ballots

   Diebold demo'd the DRS unit to Cuyahoga County OH earlier this year. Those election officials were smart enough to test some of their mail ballots. Those mail ballots jammed. Wisely, Cuyahoga County passed on the DRS. Diebold claimed it was a calibration problem. We've been trying to reach Lou Irizarry from Cuyahoga County to confirm. No luck thus far. (This tip comes from John Gideon of Voters Unite.)

   Diebold demo'd the DRS unit to King County in April of this year. No one thought to ask for them to try some of our mail ballots.
   

2. Diebold Stays Quiet on Mail Ballots

   A while back, King Co sent out a request for information to the four voting equipment vendors used in Washington State: Sequoia, Hart/Intercivic, ES&S, and Diebold. The response from ES&S loudly and proudly proclaims that their M650 hardware is the only solution which handles mail ballots. No where does Diebold claim that their DRS unit can handle mail ballots. An odd omission.

   We contacted DRS to ask for confirmation that the PS900 iM2 can handle mail ballots and for a list of customers. They referred us to Diebold. We've not heard back from Diebold.

   Councilmember Larry Phillips then followed up with his own request to DRS. We don't know if Phillips got a more informative response.

3. Election Meltdown in Scotland

   The DRS PS900 iM2 has been used in UK elections. In May, the DRS hardware was involved in an election meltdown in Scotland. DRS has accepted responsibility and has agreed to pay compensation. (Apparently, unlike in King County, vendors in the UK are held accountable for their screw ups.) The news accounts do not specify the problems.

   Counting machine company to pay out for election chaos (May 17, 2007)

4. London and King County Are Different

   King County Elections faithfully parrots Diebold's claim that the DRS unit has been used successfully in London. Well, yes. But their elections are not comparable to our planned vote-by-mail presidential election Nov 2008. London 2004 had three races, 1 ballot per race. Roughly 1.9m voters (34% turnout) cast 5,403,429 ballots. One race was counted per day (Friday, Saturday, Sunday). They used 298 DRS units to count all the ballots.

   Here's the kicker: London 2004's mail (postal) ballots were only 5% of all ballots, meaning there were only about 27,000 mail (postal) ballots. King County Elections is projecting over 900,000 mail ballots cast for Nov 2008.

   Here's the source data: London Elects 2004



5. Paying Extra for Features We Wouldn't Use

   King County Elections has trumpeted the physical paper sorting features of the DRS PS900 iM2. Presumably, the sorting feature would be used to "outstack" ballots which needed to manually duplicated. And yet, King County Elections intends to "electronically adjudicate" ballots. Meaning the database of votes will be changed directly. Meaning the sorting functionality won't even be used. Admittedly, we're thrilled to be paying for a deluxe feature set that won't be used.

6. How Will Manual Recounts Be Done?
   We have asked many times how a manual recount would be done with a "electronic adjudication" system. Most recently, we specifically asked if King County Elections would buy the add-on inkjet "carriage printer" for the DRS PS900 iM2. We have the impression the answer is "no", which would be bad.

   When a ballot is duplicated (or "adjudicated"), state law requires that a control number is added to the original and the duplicate. A carriage printer (vendor's term) would be used to print the control number on the ballot.

   All of this is important in order to enable manual recounts, which happen every general election and many primaries.  

   (Acting superintendent of elections Bill Huennekens did suggest a workaround: starting over with the original ballots. Alas, that'd be a recanvas, not a recount. And defeat the alleged productivity gains of this new system.)

7. Breaking the Certification Rules

   To be used for our Nov 2008 election, the normal state certification rules will either be ignored, changed, or broken. Explaining the arcana of certification is difficult, so I'll just quote John Gideon:

   On May 12, 2007, at 12:05 PM, John Gideon wrote:

   
   Mr. Wagner

   In previous statements I made comment regarding the likelihood of the Diebold High Speed Optical Scan system being federally tested, EAC certified, and state certified in time to allow King County to install and test the system in December as they are planning. While my comments regarding the unlikely ability for the federal testing and certification to happen before September still stand it seems that the state has filed a change to WAC 434-335-030 that changes the dates that I quoted in my statement.

   Application for certification dates are now December 1 and ending June 30th the following year. So Diebold must make application for certification by close of business on June 30, 2007. The application must include, but is not limited to, certain information including the completed Independent Test Authority (ITA) report. In other words the federal testing must be completed before Diebold can apply for state certification testing. As I stated in my previous statements the new federal testing process requires the ITA file a test plan with the EAC for review, including public review. This initial step in the process has not happened yet and there are two systems, including one submitted to the ITA 7 weeks prior to Diebold submitting their system, that still have no written and approved test plan. The ITA test process has historically taken over 6 months.

   The state certification testing dates have also been changed. Certification examinations and hearings must now take place between January 1 and September 15. If the application for certification cannot be accepted by the state after June 30, there is no way testing will happen this year.
   In other words, unless the state dangerously cuts corners, changes their administrative code, accepts less than required documentation and certification, and removes public hearing requirements; the Diebold Assure voting system will probably not be available for use this year.

   --

   John Gideon
   Co-Director and Information Manager
   VotersUnite.Org
   www.votersunite.org
   

   Interestingly, our state used to require that at least one other state was using a vote counting system before it could be certified here.
   

Conclusion

At this time, we very much doubt the DRS PS900 iM2 is suitable for counting our mail ballots. (It's really weird that this is even an open issue.)

• Minimally, we insist that King County require a full blown demonstration before moving forward.
• Ideally, a county with elections as large and complex as King County would not be the first customer in the nation to use an untested, uncertified, and unproven new tabulation system. This much seems like a no-brainer.

If the DRS PS900 iM2 does have a problem with mail ballots, we suspect it'd be due that model's unique design. The feed wheels are only along just one edge. This permits handleing any width of paper stock as well as exposing the paper path. It's very clever.

Alas, this DRS may not be suitable for King County's vote counting needs.

Further, there remain a few unanswered questions about how the DRS unit could be lawfully used in our election Nov 2008.

[Update Friday June 20th, 2007 @ 12:18pm]

Open Rights Group's Report

Announcement: ORG Election Report highlights problems with voting technology used

Actual report: May 2007 Election Report: Findings of the Open Rights Group Election Observation Mission in Scotland and England (June 20, 2007)

In brief, the report's conclusion is that there continued to be lots of problems with the UK's "e-counting" (electronic vote counting) systems. Despite reassurances from the vendors that all is well. There's a lesson there somewhere.

This bit about DRS having problems with fold marks is from page 44.

Also observed during adjudication in Edinburgh was a ballot paper where the system appeared to be confused by a fold mark. The result was that software voting options were presented only on the right above the fold mark and only on the left below the fold; the options should have been presented on both sides above and below the mark. DRS staff could not resolve the issue using the RO’s login so the ballot was counted using a special manual entry. It is unclear if the DRS system experienced the same error with other votes, but counted the votes instead of sending them for adjudication.

It's not clear if "fold mark" is referring to a printer's mark (showing were the fold should happen) or the actual folding of the ballot leaving a mark when scanned.

This comes from page 56:

Electronic Counting

Despite repeated assurances from suppliers in Scotland and England that their e-counting systems were proven and reliable, this was shown not to be the case. Systems were not robust in the face of minor issues such as imperfect perforations, folds, changes of ink used for printing, variable ballot paper sizes and mixing of ballots from different contests. Given that other countries have already experienced such issues, it is disappointing that they were not foreseen by suppliers or the Government. That these issues did cause detectable problems leads ORG to be concerned that other errors could have gone unnoticed. The resistance of ROs to conducting manual recounts and the complete failure of Government to mandate any form of manual counts, even of sample batches, indicates overconfidence in the systems purchased. Manual counts of at least some votes would have served to increase confidence in the results declared whilst providing important information on the level of confidence e-counting systems should be accorded. The one manual recount that did occur, in Breckland, found a significantly larger number of votes than e-counting reported.

It is also notable that very significant delays were experienced due to the number of ballots sent for adjudication. It is ORG’s view that this could and should have been anticipated, as the Electoral Commission had noted, after e-counting pilots in 2006, adjudication was the key bottleneck in e-counting procedures.

Because e-counting does not eliminate the paper ballot, ORG believes ROs and Government were overconfident and not sufficiently careful in their preparations for e-counting votes. Given e-counting’s long history, particularly in the United States, there were ample sources of information from which common problems and potential solutions should have been identified.