FFEC :: Accessibility vs Verifiability in Volusia County

The current debate about whether to buy Diebold paperless touch screen machines, sometimes called DREs (direct recording electronic ballot stations,) or the AutoMARK ballot marker to meet requirements for disabled access to voting machines under the Help America Vote Act (HAVA) has been framed as accessibility versus verifiability. Yet the features that make both of these machines disabled accessible are unrelated to whether they mark a paper ballot or a memory card. The "us" vs "them" nature of this discussion is predicated on the assumption that, absent verifiability concerns, one would chose the Diebold touch screen. That is emphatically not so. From the perspective of disabled access alone, the AutoMARK is clearly superior to the Diebold equipment.

Let me be clear about our findings:

1. More disabled voters are able to vote privately and independently on the AutoMARK than on the Diebold Accu-Vote TSX.

2. The AutoMARK is easier and quicker to use than the TSX for all disabled voters, including the blind.

3. Testers from Florida Fair Elections Coalition who used both pieces of equipment unanimously preferred the AutoMARK. In fact, we do not know of a single disabled person, including blind people, who has used both pieces of equipment who does not prefer the AutoMARK.

4. In a recent survey of nearly 500 disabled people who used the disabled accessible voting machines at an Oregon vendors' fair, AutoMARK received the highest overall ratings of the ten machines tested.

Table 1 at the end of this document gives a side-by-side comparison of the AutoMARK and the Diebold Accu-Vote TSX.

The AutoMARK was developed specifically to provide private and independent voting for disabled voters. Like the TSX, it also allows blind voters to put on a headset and listen to a voice that gives them directions and then guides them through the ballot. But instead of nine buttons, it has five (up, down, left, right, and a center button to execute commands), a sip and puff control, a joy stick, and a foot pedal, all of which make it accessible to voters with problems using their limbs.

The AutoMARK allows the voter to verify his or her ballot in several ways. The machine will read back the voter's choices as he goes along and will repeat all the choices before the ballot is marked. Voters can also put the ballot back in the machine, and it will read back the voter's choices. This feature provides a true verification since it will read back any ballot put in the machine, not just the last one marked.

Opponents have had to look hard for something to complain about. They have decided that the AutoMARK doesn't allow manually disabled voters to vote independently because a poll worker might have to load the ballot in the machine for those with the most severe impairments. Very few disabled voters will need such assistance, but admittedly, a few will not be able to load the ballot by themselves. Of course, all disabled voters will need the ballot loaded by the poll worker on the DRE with a smart card. None of them will be able to load it themselves. Those with severe manual impairments can't vote at all on the DRE hence they won't have to worry about independently loading the ballot.

Another complaint is that the paper ballot could be seen by others, thus compromising the privacy of the disabled voter's ballot. That issue is addressed in several ways. First, the ballot comes out into a privacy sleeve, much like the folder given other voters. In addition, the machine has a feature that allows the ballot to be printed without names and other text so that it cannot be read by poll workers if their assistance should be needed at some point.

Detractors also have alleged that people with severe manual impairment would not be able to carry the ballot to the scanner and put it in. Most disabled voters, including those with manual impairments, should be able to accomplish this fairly easily since it does not require fine motor skills. But for those few voters who cannot carry the ballot themselves, it can drop directly into a locked ballot box—not a high tech option, but certainly a simple one. Remember, this is not a problem on the DRE because they can't vote on it at all without assistance! Further, the disabled voter will not be able to remove the smart card from the machine and return it to the poll worker.

Florida Fair Elections Coalition sent testers to Pasco County for the specific purpose of seeing how the machines really serve disabled voters. They uniformly found the AutoMARK to be easier and quicker to use than the Diebold DRE. Their report is attached as Exhibit A.

In addition, a survey of users at a vendor fair in Oregon gave the AutoMARK the highest overall rating among the accessible voting machines tested for its accessibility features. Although the Diebold Accu-Vote TSX was not among the disabled accessible machines demonstrated at the Oregon vendor fair, many of the tested machines were DREs. See Exhibit B for details. The purpose of the demonstration in Oregon was to give guidance to the state in purchasing in its disabled accessible equipment.

The Diebold Accu-Vote TSX is nothing but a regular touch screen voting machine that can be used with a kit that makes it accessible for blind voters. First, a poll worker must load the ballot for the disabled voter using a "Smart Card." Blind voters put on headphones and listen to a voice that instructs them in how to use the equipment and then directs them through the ballot. They navigate through the ballot using a nine-button keypad, similar to that on a telephone. The keypad is detachable so that the voter can place it in whatever position works best for him or her. Those are the only disabled accessible features available on the TSX.

When board members of the Florida Fair Elections Coalition used the Accu-Vote TSX, they discovered a number of significant problems with the machine. They found that it did not have any features for meeting the needs of people with severe manual dexterity issues—no foot pedal, no sip/puff, and no joy stick. The absence of these features means that this equipment will not serve a large segment of the disabled community and thus will not comply with the provisions of the Help America Vote Act (HAVA), which mandates that the equipment serves all disabled voters.

Also, as demonstrated at the Volusia County Supervisor of Elections office, the Diebold DRE did not have the simultaneous audio-visual feature that allows visually impaired voters to use the audio to augment what they can see. In addition, members of our board who used the equipment at a demonstration in Pasco County found that it would not allow review of the ballot as required by HAVA. It would only inform the voter that his or her choice had been selected. Ballot review is required by HAVA and would be necessary for reimbursement.

Perhaps the most surprising finding of our testers was the incredible length of time that it took to vote on the DRE using the audio component. According to our testers (and confirmed by a Diebold technician at the Pasco site), it took them about 30 minutes to get through a ballot, more than three times longer than on the AutoMARK.

This finding is significant for a number of reasons. First, people are more likely to make mistakes, skip races, or fail to review their ballot if the process takes an extraordinary amount of time. Plus, many disabled voters in Volusia County, especially the elderly, may not physically be able to wait in long lines. Consider how long the waits could be: If a carload of voters (4) came to the polls together and no one else was already there, the last person in the car would have to wait at least two hours to vote and probably longer. If a van arrived with eight voters, the last person in the van to vote would have to wait at least four hours to vote. And that allows for no time between voters.

But again, this assumes the best case scenario. Recent tests show that disabled voters who tested the DREs still required assistance to vote. An article in Access World, published by the American Federation of the Blind, reported that a test by disabled users of the audio voting features of five different DREs found that almost all needed some assistance with each of the voting machines. The most difficult task for the thirteen disabled users was attempting to change their selection in a race. Eight of the thirteen needed assistance accomplishing this task on the Diebold Accu-Vote TSX.

In all the confusion surrounding this issue, it is a good idea to remember just why we are buying this equipment: to meet HAVA requirements that we have at least one disabled accessible machine in each precinct by the first federal election after January 1, 2006. We will only be reimbursed if the equipment is HAVA compliant.

On its website, the manufacturer of AutoMARK has posted a legal opinion that their equipment is HAVA compliant. They are willing to guarantee their equipment and put that in their contract. Diebold has not offered a guarantee, and we are confident that they cannot do so because their equipment is not HAVA compliant. Diebold's proposed contract with Volusia County does not even mention HAVA nor does it guarantee compliance. As a matter of fact, it explicitly disclaims all representations and warranties, express or implied.

Section 301 of HAVA contains the standards for voting systems, including the requirements for disabled accessible equipment. In paragraph 301(a)(3)(A), it states "the voting system shall be accessible for individuals with disabilities, including non-visual accessibility for the blind and visually impaired, in a manner that provides the same opportunity for access and participation (including privacy and independence) as for other voters."

Overall, the AutoMARK allows the disabled person's voting experience to be as nearly equal to that of the non-disabled person as possible. The disabled voter marks the same ballot as everyone else and has it cast and counted on the same equipment. In addition, the disabled person's vote is just as likely to be counted as anyone else's, and is just as verifiable. In fact, once the ballot is marked, there are no distinctions between the disabled person's ballot and anyone else's. The AutoMARK comes closest to providing what HAVA requires—the same opportunity for access and participation.

Another important consideration for Volusia County is how the equipment will interface with its current optical scanners. While many have questioned whether the AutoMARK can work with our Diebold Accu-Vote OS, most have assumed that the Diebold touch screen will work with the Diebold optical scanner since it is the same vendor. But this is not the case; it is the DRE that will make a huge impact on our current system and procedures for holding elections.

The AutoMARK will work with our equipment without any additional hardware, software, or upgrades to our system. The AutoMARK is a stand-alone ballot marker. It can mark any piece of paper. Once the ballot is marked it goes into the scanner, and that is the end of the impact on the system.

Unfortunately, as Diebold representative Robert Pickett told the county council, Diebold has "made a business decision" to push its DRE for disabled access and therefore will not permit the AutoMARK to be used with its equipment. It sent a letter to the state certification department to that effect. Shortly thereafter, the certification department made a decision that it would not test for certification the AutoMARK features that allow it to be programmed from the ballot. This means that the AutoMARK will have to be programmed from the Diebold software used to create the ballot. In short, the certification department aided Diebold in its attempts to force us into buying its equipment instead. They thought they had us over the proverbial barrel. They didn’t count on the outrage of the citizens as well as the county council at the obvious attempts at coercion.

Using the touch-screens would mean considerable changes in our election procedures, as well as the acquisition of equipment, some of which has not even been developed, much less certified. At present, the optical scan accumulator adaptor listed as optional equipment on our contract is not operational. Diebold is still trying to figure out how to get this to work so that all the data can be put together before transmitting it via modem to the central tabulator. This is crucial because our central tabulator, as a security feature, will only permit a single upload from a precinct. Therefore, as the acceptance testing shows (exhibit 6 of the proposed Volusia county contract with Diebold,) the plan is to download all the optical scanners via modem from the precincts as usual, and then carry the memory cards from all 179 touch-screen machines back to the Supervisor of Elections office. The cards will be put in the machine one at a time and uploaded a total of 179 times (if nothing goes wrong.) If we assume that each card will take about five minutes for the upload, we can expect this procedure to add at least fifteen hours to our process.

Of course, that assumes that it can be done at all. There is nothing in the acceptance testing to show how we will be able to upload these cards at all once the optical scanner results are in from the precincts since only one upload per precinct is permitted. The acceptance testing cleverly skirts this issue by testing only the interface of the DREs and the tabulator and omitting testing with the scanners altogether. If we assume that a planned upgrade to the central tabulator will make this possible, we have to ask if that upgrade is currently certified. If it isn't, then our entire system is uncertified, and the council by law cannot buy it.

Even though it was clearly a carefully orchestrated show put on by the Supervisor of Elections, the June 29 special meeting of the Volusia County Council was heart-wrenching. People with disabilities crowded the chamber. Many of those who spoke for the purchase of the DREs were bused in from Jacksonville, one hundred miles away. Members of HAVOC (Handicapped Adults of Volusia County) spoke against the purchase and argued for a verifiable as well as accessible vote. Many of the disabled seemed puzzled about why, after all this time, they still weren't assured of a private and accessible vote.

Unfortunately, they were talking to the wrong people. It was not the county council that was seeking to deprive them of their right to vote privately and independently. Indeed, Volusia County was among the first counties in Florida to vote to get disabled accessible machines. They should have addressed the State of Florida, which had made them pawns in a battle over accessibility—that is, access to billions of dollars in federal funds.

In reality, Volusia has been given no choice at all; it is expected simply to accede to what the state has already planned—to make sure that big corporations that made big donations to political campaigns get the lion's share of the spoils. The disabled are given the choice between poor equipment or none at all. Everyone loses—except the vendors, of course.

This battle is an essential one in the fight to preserve verifiable elections. The state has long indicated its wish that all Floridians vote on DREs. Internal memos from Diebold staff show that it would like to phase out the optical scanners and move all its customers to DREs. Once the machines are being used for disabled access, it hopes that its customers will be pushed to move to a single system because of the many problems (both legal and technical) with operating two systems.

Volusia County has had the audacity to consider its obligations to its citizens—to provide machines that will truly allow a private and independent vote for disabled voters and to maintain the verifiability of everyone's vote—to be a higher consideration than blind obedience to the state. We are grateful for their courage.

Comparison of AutoMARK and Diebold Accu-Vote TSX
by Accessibility Features and System Impact

Accessibility features	AutoMARK	Diebold Accu-Vote TSX
Audio & headphones	Yes	Yes
Keypad	Yes, 5 keys	Yes, 9 keys
Sip/Puff	Yes	No
Foot pedal	Yes	No
Joy stick	Yes	No
Simultaneous audio & visual presentation	Yes	Not available at Volusia Co. demo
Audio review of ballot	Yes	Not available at Pasco Co. demo
Time to vote (at Pasco demonstration)	Less than 10 minutes	More than 30 minutes
Impact on system and procedures
Allows reporting totals from precinct	Yes	No
Requires purchase of additional equipment	No	Yes. Upgrade to central tabulator, optical scan accumulator adaptor for precinct reporting (not yet certified)
Additional poll workers	One per precinct	One per precinct
Additional procedures on election night	None	Print out tapes from DRE, remove disk, transport to SOE's office, download 179 disks one by one to central tabulator, accumulate results.
Additional time for election night reporting	None	Between 10 and 15 hours (if nothing goes wrong)
Certifications
Federal (NASED)	Yes	No
State	Pending	Not all components required for operation

Why the AutoMARK is Better
for Volusia County

Note: This report was prepared in July 2005. Much more information is now available on the performance of both machines. Our research, however, continues to show satisfaction with the performance of the AutoMARK by election officials and disabled voters and dissatisfaction with the TSX.

Accessibility versus Verifiability?

Why the AutoMARK is Better for Volusia County

Note: This report was prepared in July 2005. Much more information is now available on the performance of both machines. Our research, however, continues to show satisfaction with the performance of the AutoMARK by election officials and disabled voters and dissatisfaction with the TSX.

Accessibility versus Verifiability?

Why the AutoMARK is Better
for Volusia County