Bucks County Voting Machines

Results of Work Group Assessment

I.       Background

An interdepartmental, interdisciplinary work group was formed by the Board of Elections for the purpose of examining in depth the proposals from three voting machine companies.

Over a period of the past several weeks, the work group undertook the following activities:

v      Interviewed three vendors

v      Analyzed cost proposals

v      Compared cost proposals with the CoStars-10 (state) purchasing contract

v      Secured and reviewed the contracts from other jurisdictions for voting machine acquisition

v      Projected the 10-year operating costs for each of the three systems

v      Assessed cost, reliability, and compliance with the Help America Vote Act (HAVA)

v      Discussed deliverables and schedules with each vendor

v      Prepared a comparison of the three systems, noting pros and cons

v      Considered the input from the public at the Voters’ Forums held in January, 2006

v      Formulated observations and conclusions to pass on to County administration

       Why didn't they examine research material provided by the Coalition for Voting Integrity?

The work group contained representatives from the following departments:

v      County solicitor

v      Information Services

v      Purchasing

v      Finance

v      Voter Registration/Board of Elections

v      Voting Machines

v      Community Services Division/Planning Commission

Who exactly were the persons who comprised this group?

Why aren't names provided?  Why aren't we told how often they met?

The three vendors under consideration are:

1.   Advanced Voting Solutions, Inc.

2.   Election Systems and Software (ES+S)

3.   Danaher Industrial Controls

These three vendors demonstrated their machines at the public Voter Forums sponsored by the County Commissioners on January 18 and 19, 2006.

II.     Factors Considered

The factors considered by the group were the following:

These factors are covered in the comparison matrix. Information on how assessments were made in discussed in the text.

Information Systems Assessment

Two members of the IS staff were assigned to the work group and assessed the software/hardware/IS needs of all three system. Advanced uses a Windows-based software, which was considered to have pros and cons. The other two vendors use proprietary firmware. Advanced uses Smart Cards which were confusing to poll workers.

Did they have actual poll workers trying these systems?  If not, where did they get this claim about Smart Cards being confusing for poll workers?  This report either relies solely on the “work group” members, or it also uses outside sources.  If that is the case, we can introduce reams of outside sources to contradict nearly everything in this report.  Also, any IS professional is more aware than the average citizen how computers can be error-prone, can crash, be hacked, etc. They would know how important having an independent backup is essential to preserve data.

HAVA Compliance

All three systems meet the system requirements of HAVA and have been certified by the federal and state governments.

Voter-Verified Paper Audit Trail (VVPAT)

ES+S uses a paper ballot which is scanned. The other two systems can be adapted to include a paper ballot, although these add-on components systems have not been approved in Pennsylvania. The Advanced VVPAT system is not available.

All VVPATs are not equal in reliability, security, and ease of use.  As stated above, the optical scan is the only system that inherently provides a voter-verified paper trail. The paper ballot is filled out by the voter by filling in circles with a pencil or pen next to the candidates, fed into and read by scanner, then retained in the scanner for use in recounts or audits. In one very easy to do operation, the ballot is voter-verified, filled out, cast and counted: it’s very simple, easy, reliable. 

The so-called VVPAT’s in touchscreen systems are provided by hooking up a separate printer (more about that later) to each machine, which allegedly prints out the voter’s choices, and requires the extra step of the voter rechecking to verify it is correct.  Problems with this system include the documented extra time this takes and the difficulty of reading and verifying the ballot print-out, since the print is small, hard to read, and not in the same format as the original computer screen ballot. There have been studies done that confirm that nearly 80% of voters don’t take the extra step to make sure the print-out verifies their actual choices, due to confusion with computers, inability to see print-outs, and nervousness about taking too much time to vote.  When we pointed out this difficulty to the commissioners, their response was that it would be the voter’s fault if they don’t choose to make the extra effort to verify their choices!  Wouldn’t the best solution be to make the voting process as easy and secure as possible for the voters instead?

Ease of Use and Familiarity

Consideration was given to the ease of use of the machines. The survey results from the voter forums are included. The stability and care of the machines was discussed.

County Staff Responsibilities

County staff will have responsibilities in any system, in terms of setting up and printing ballots; transporting machines; storing machines; answering questions; working with the software provided; collecting and tabulating election night results (unofficial count); placing election night results on the website. The county will also have to dedicate computer hardware to the election reporting system. The systems for communication with the courthouse and means by which election results are collected and tabulated were discussed.

     In the certification report for Danaher, the State Examiner, Dr Shamos, questions the use of a dedicated phone line to transmit election results to the County courthouse.  Dr Shamos says that it is a large expense and that the results transmitted electronically cannot be used as official results because electronic transmission of the official results are not authorized by the Election Code.

Training

All vendor proposals include training for county staff, election technicians, and poll workers. Public training is an optional add-on and can include demonstrations, videos, and public service announcements. This is an important component for each system.

There is much documented evidence that expecting non-technical and possibly computer-phobic poll workers and voters to work with the more-complex touchscreen systems leads to more error and security lapses than should be acceptable in an election system. The commissioners pointed out that most of the problems they read about concerning the Danaher system were all “human errors.”  Aside from the fact that there are hundreds of documented Danaher machine failures in Philadelphia alone, one should also keep in mind that a system that results in many human errors is not designed well and that should be a major concern.

Warehouse, Transport, and Storage and Needs

Each voting system would require storage in the warehouse, including electrical drops. None of the systems require climate control.

Danaher machines need to be plugged in year-round.

Deliverability

None of the system vendors can deploy machines in time for the May 16, 2006 primary election. No partial deliveries can be made, and no used or reconditioned machines are available. Danaher and ES+S predict a delivery date of June or July, with set up and training occurring for the November, 2006 election. Advanced stated that it would be “impossible to guarantee delivery by May 16.” A delivery date in late April is a possibility, but there would be no possibility to set up and provide training for the primary election.

Affordability

Cost proposals were received from the three vendors. The work group reviewed the costs prior to interviewing vendors and then discussed each line item with the vendor representatives. The proposals contained the costs for voting machines, consumable products, set-up costs, training, technical support, information services requirements, and computer hardware and software. The number of machines per precinct varies, depending on the type of system, but is never fewer than two per precinct.

Through this discussion and by comparing the cost proposal with the CoStars-10 contract (state purchasing agreement) and with the executed contracts from other jurisdictions, the work group was able to determine which elements of individual proposals could be modified or reduced. In some cases, the county’s existing departments (Information Services, Voter Registration, Voting Machines, and Public Works) are able to perform tasks that the vendor also offers. Where existing personnel can provide the service, the vendor proposal was modified.

Assessment of 10-year costs

A projection of 10-year costs was made, based on the information provided by the vendor and assuming two elections per year. Included in this projection are licenses fees for software, maintenance fees, consumable products, such as paper, printing, batteries, electronic media, seals, and so forth. Every potential cost item was examined to prepare the 10-year projected costs. In year one, the initial purchase covers some of the costs, so the costs are calculated for a nine-year period in this case. These are all itemized on the attached cost estimate sheets.

We have major questions on how some of the costs were arrived at.  We do not see an itemized cost estimate sheet, which would allow us to be sure all costs were compared fairly and accurately. 

III.    Profile of Three Systems

Advanced Voting Solutions

The Advanced system is a touch-screen machine. The entire ballot does not appear at the same time to the voter. The program scrolls through the ballot listings. The number of machines needed is based on the number of registered voters. For the purpose of our estimate, it was assumed that the state-recommended number of 1 machine per 300 voters would be needed.

The vote tallies are registered on the voting machine (which has a Smart Card specific to the precinct in which the machine is located), on a paper tape, and on a USB device. The USB device is used to transfer the vote tallies to election central. At the precinct level, the tallies from the USBs in use are totaled on one USB, which is delivered to election central, either in Doylestown or in Levittown.

Handicapped voters are accommodated with headset devices which can be fitted on the voting machines.

There is a VVPAT under development for the Advanced system; however, it has not been certified by Pennsylvania.

Voting machines are light weight and are stored in a precinct cart, which holds up to 10 machines. The machines can be plugged into the cart and the cart connected to a power supply.

ES+S

The ES+S system is an optical scan card reader system. Voters receive a paper ballot, which is 8 ½ inches by 14 inches (or larger if needed). The voter marks the ballot and inserts it into an optical scan machine. If a ballot contains overvotes, a display screen alerts the voter to a ballot problem. A ballot can be corrected before it is counted and put into the ballot box. After scanning it goes into a ballot box.

ES+S requires a separate type of system to accommodate handicapped voters. There are two options. One, called the Automark, uses a keypad and headphones and the machine marks the paper ballot. The voter can insert it into the optical scanner. The second option is a touch screen machine called the iVotronic, which also uses headphones. In either case, a separate unit would be required in every polling place for handicapped voters.

The optical scanner uses proprietary software and PCMCIA cards in each scanner to record votes. At the end of voting, the cards and a results tape are carried to election central. Election results are tabulated from the cards.

Paper ballots are required to be printed for every voter, at a cost of $0.31 per ballot.

This is a misleading statement, its intention seemingly to imply that ballot costs are an expensive issue. Also, especially for primaries, it is only necessary to print enough for a percentage of registered voters.  There is no mention that paper ballots (absentee and provisional) would also be required should a touchscreen system be purchased.  In the very likely event of machine failures, paper ballots would need to be on hand for voters.  Because of the volume discounts possible when printing for an entire county, the paper ballot costs would be a known, fixed cost with an optical scan system.  The costs of ballots needed with a touchscreen system would be considerably higher per ballot because there would be the same printer set-up costs, and no volume discount.  We would also need at least one scanner to read the absentee and provisional ballots, unless we opt to count those by hand.

The vendor said that optical scan results can be inaccurate, due to marginal marks on a ballot, folds in the paper, or smudges on the paper. He expressed surprise that the county was considering optical scan because the touch screens are more accurate. If a ballot cannot be read by a scanner, it can be hand-copied onto a new ballot and then scanned. High-speed scanners are available for central locations, but their accuracy is less reliable than the scanners proposed for each voting precinct.

Here, the report seems to imply that because the vendor himself seems to be reluctant to tout this product, it must be inferior, especially since a fault was intentionally pointed out.  First of all, all the voting machine vendors want to push their touchscreen systems because their profit margins are so much larger than what they get for the optical scan systems.  Many more touchscreens are required (3-6 times as many), they make vast amounts in escalating and hidden “maintenance fees,” and other possible reasons. 

Also, how a vendor can claim that touchscreens are more accurate when there is absolutely no way to verify their results is outrageous.  One will notice there is no mention of the hundreds or thousands of touchscreen malfunctions, breakdowns, lost votes, elections thrown into chaos, and no way to recapture the voters’ intent….only a “surprise” that optical scans are being considered with errors occurring because of folds or smudges on paper.  Absolutely incredible.   

Danaher

The Danaher machine is a full-screen touch screen system. It replaces the lever machines, one for one, and has the same appearance as the machines with which voters are familiar. The entire ballot is displayed on the machine face.

Each unit comes with an audio keypad so it can be used by handicapped voters. Results are tallied on memory cartridges, which are carried to election central. They use proprietary software. There are redundant reporting mechanisms, with aggregate result totals being stored in six locations that are cross-checked against each other.

Many of these security problems (data stored on small chips that can be stolen, lost, etc., and private and secret software used to run the machines) are shared with the optical scan machines.  However, the optical scans have a secure, redundant, voter-verified hard copy of the ballots to use for recounts and audits of an election should problems or disputes arise.  These alleged redundant reporting mechanisms stored in numerous locations and cross-checked with each other are useless because they are all generated by the computer!  If a vote is incorrectly cast, no matter how many times you recount it or print it out, it will still be wrong, and you will have no way to check.  Should the entire machine crash, you have no way to recover the votes.   

In addition to the memory cartridge, Danaher has developed a VVPAT module that can be added, but it is not certified and cannot be used in Pennsylvania.

Danaher has never used a VVPAT module in any election…this is a vague promise of something that may never come about, and has no guarantee of ever being available or certifiable in Pennsylvania.  Should Pennsylvania ever join 38 other states in the USA that currently or in the very near future will require a VVPAT of some kind, we will either be at the mercy of whatever printer system Danaher comes up with (and pay whatever they demand), or toss out the whole system and start over (most likely going to an optical scan system, like entire states are currently doing!). Any scenario will prove very expensive to Bucks County taxpayers.

IV.    Preferences of Voters and Poll Workers at Voter Forums

Voters and poll workers who used the machines expressed preferences. The results, which are attached, indicate that the Danaher machine received the highest marks in all questions asked.