On March 17, 2005, several representatives from TrueVoteCT met with with Anthony Householder, Director of Community Affairs, of Senator Christopher Dodd's Wethersfield office. Mr. Householder arranged a conference call with Ms. Kennie Gill, a member of Senator Dodd's Washington staff and his primary assistant on the issue of electronic voting. Mr. Householder described Ms. Gill as Dodd's HAVA expert and the person who drafted HAVA in the first place.
The conference call lasted for more than an hour and a half. It was a frank and spirited and at times contentious discussion, which touched on a range both technical and political issues. No formal minutes were taken. What follows is a compilation of impressions based on the personal recollections of individual participants in the conference call.
Ms. Gill thanked us for our efforts to support SB 55 and was especially complimentary of our support of accessible voter verified paper ballots---i.e., VVPBs that would be accessible to persons with disabilities.
She noted that Senator Dodd was one of primary architects of the HAVA act and feels very strongly about making voting accessible to persons with disabilities. One of the first questions that came up was whether Senator Dodd would support SB55, given its amended language to require an accessible VVPB. She replied that Senator Dodd makes it a policy not to take a position on state laws. No one at the meeting raised the point that Senator Dodd was instrumental in killing last year's version of SB55 (SB388), which had received widespread support in the Connecticut General Assembly and looked like it was headed for easy passage.
There was much discussion of HAVA. It was obvious that Ms. Gill (and perhaps the Senator himself) are very much committed to HAVA, to the point of coming across at times as somewhat defensive about its shortcomings.
There was discussion about whether CT's lever machines would need to be upgraded or replaced in order to be in compliance with HAVA. Although we would have expected Ms. Gill to cite the requirements for a permanent paper audit trail as the reason why lever machines would not comply, she instead cited the requirement that a voter have the opportunity to verify and correct the ballot before casting it. She said, "How can a voter verify the ballot on a lever machine? They can't."
This response seems somewhat puzzling to us because it is now widely acknowledged that despite HAVA's "verifiability" language, there is no voter verification with paperless DREs. Yet rather than simply acknowledging this point, Ms. Gill seemed instead to try to argue that all voting technologies have verification problems. A comment that she repeated at several points in the discussion is that "we can't let the perfect become the enemy of the good." What she meant by this, apparently, is that despite HAVA opening the door to completely unverifiable, paperless DREs, this is a big improvement in our voting systems and we should take it as a step forward.
On the contrary, we pointed out that although many of us were optimistic about HAVA, its implementation has been a huge disappointment. One of us mentioned that HAVA's main accomplishment so far seems to have been to allow large, partisan-affiliated voting machine vendors to gain increasing control over US elections at the expense of the taxpayers.
When the phrase "privatization of voting processes" was used to characterize the trend that HAVA seems to have fostered, Ms. Dill took exception and explained that corporations have always made voting equipment and we wouldn't want the federal government to take on the manufacture of such equipment. We agreed with these points but noted that, whereas in the not too distant past, votes were primary collected and counted by volunteer, bi-partisan elections officials, they are now increasingly being collected and counted by computers, manufactured by partisan corporations, at the voting machine, precinct, county, state, and federal levels. We pointed out that the American public is growing increasingly skeptical of this practice.
Another issue concerned the 2006 target date in the HAVA legislation. It is our understanding that the dates in HAVA were chosen to avoid initial use of the new machines during a presidential election. Thus 2006 was chosen to avoid roll out of new equipment during 2008. Thus, the decision by the Connecticut Secretary of the State to use the machines in November 2005 was never intended by HAVA and represents an unrealistic deadline.
There was much discussion of the need for VVPBs. Despite her apparent support for VVPBs at the start of the meeting, Ms. Gill never missed an opportunity to question the whole idea paper ballots. At times her points came very close to what one typically hears from vendors of paperless DREs. She raised nearly every stock objection: there has been a long history of fraud and ballot stuffing associated with paper ballots; printers jam and disenfranchise voters; no system is perfect; paperless DREs will be tested by independent Federal laboratories; paperless DREs have the lowest error rates and highest rates of accessibility; and even lots of voters don't bother to look at the printed record (as if the electronic record, which they can't look at, is any better). In addition to these stock objections, Ms. Gill then went on to claim that the printer might put the duplicate records into the ballot box without the voter verifying it. We had no idea what she was talking about. She then asked, "How will the voters verify the paper record?" and one of us responded, "By opening their eyes and looking at it." This exchange left us wondering whether Ms. Gill truly understands how VVPB systems worked.
One stark reminder of how far apart our views were regarding the issue of VVPBs is that Ms. Gill's views of the 2004 elections in Maryland and Nevada were diametrically opposite to our views. She claimed that the Nevada (accessible VVPBs) election was fraught with difficulties and that the Maryland election (100% paperless DREs) went off without a hitch. We believe her interpretation of the 2004 election is simply out of touch with the reality that occurred. There were over 50,000 documented instances of irregularities, many associated with DREs.
Much of Ms. Gill's objection to VVPBs focused on whether they should be considered the official ballots. She thinks not, and this apparently is reflected in Senator Dodd's bill (S. 17). A main reason seems to be that VVPBs are not accessible to the blind and other persons with disabilities. She explained that this was the motivation behind Dodd's new bill, which would require 4 different kinds ballot, including paper, audio, pictorial, and digital. When we pointed out that there are machines now that allow AVVPB, she reiterated many of her objections to using paper in an election.
We explained that Dodd's bill is not supported by VerifiedVoting.org because, among other things, it doesn't require voter verification until 2009, which is unacceptable to us.
What was most dismaying about the discussion is that Ms. Gill apparently does not acknowledge the basic lack of verifiability and security that goes with paperless DREs. She seems to regard the impossibility of verifying what goes on in the computer's memory as just another technological problem that can be addressed by the right kind of engineering. Her view seems to be that all voting systems have problems, and DRE voting has advantages that far outweigh its problems.
The computer scientists and computer professionals among us tried to explain that there is a qualitative difference between DRE technology and technology that uses VVPBs, pointing out that it is theoretically and practically impossible to determine that the software inside the DRE is correctly counting votes. Her response was something like, "come on, you mean we can land people on the moon and fly millions of people in airplanes every day and we can't verify that the DRE machine is trust worthy?" In other words, in Ms. Gill's view, the problems we perceive with paperless DREs are just engineering problems that can either be worked out by better design or guarded against by better independent software testing and certification.
We tried to assure her that there was near unanimity among computer scientists and computer professionals (except of course among those who are working for the DRE vendors) about this fundamental problem, but to no avail. We claimed that with paperless DREs there was no way to achieve an independent audit and that a necessary element for an independent audit is a ballot that had been verified by the voter. At one point Ms. Gill seemed to dismiss our points by suggesting hat computer experts have one opinion on the technological issues and Senator Dodd has another.
To illustrate how far apart we were on this issue, consider our difference over the following detailed example raised by Ms. Gill, apparently to knock down the idea that VVPBs should be the official ballots. She asked us to imagine a case where the number of voters who signed the registry at the poll was less than the number of paper ballots counted. Surely, in such a case, we wouldn't want to rely on the paper ballots. We responded that indeed, in such a case, there appeared to be evidence of vote tampering and that an investigation should be conducted. But we pointed out that the evidence that something was amiss was due to the fact that there were paper ballots to use as prima facie evidence. And we offered the following counter example. Suppose in a paperless DRE system, the software merely switched votes from candidate A to candidate B, so that the total votes matched the number of voters. Without a VVPB, there would be no way to determine that tampering occurred in this case, which is why VVPBs are necessary. We truly expected that this would convince her of the difference between paperless DREs and VVPBs. However, to our astonishment, she refused to concede the point and instead raised the irrelevant issue that in Nevada some voters reported that they didn't look at the paper ballots. It was a frustrating exchange.
The impression that we were left with is that Ms. Gill (and we would assume Senator Dodd himself) are for some reason not supportive of VVPBs. We speculated after the meeting that perhaps their commitment to the disability community is what is behind their reluctance to offer unequivocal support for VVPBs.
Perhaps the most positive conclusion that can be drawn from the meeting is that Ms. Gill agreed that what happens at state level needn't wait for these issues to be resolved at the federal level. She said repeatedly that HAVA only imposes minimum requirements and the states are free to require higher standards. She also said that the intention of HAVA was to give the states great flexibility to "experiment" and to use HAVA help to meet their own needs. She strongly implied that CT could use its HAVA funds for lever machine replacement, but added that it would be subject to what is written in the state HAVA plan. She also acknowledged that the plan could be revised by following the appropriate "sunshine" procedures and that several states had already revised their initial HAVA plans.
However, when push came to shove about questions (1) can HAVA funds be used to replace lever machines, and (2) are lever machines compliant with HAVA's 2006 requirements, she deferred to Federal auditors (for (1)) and to the Justice Department (for (2)). She seemed to saying that what HAVA has accomplished is something akin to: Write an ambiguous law, fail to tell people what it means, and then threaten legal action against them if they fail to comply. She did say that no action had been taken against anyone for violating HAVA provisions. She also indicated that the likely consequence of failing to meet the January 1, 2006, implementation deadline would be the necessity to work out an agreement with the Justice Department (a consent decree?) that would spell out what actions the state must take to become compliant.
The point that we derived from this part of the discussion is that the consequences to the state of failing to meet the deadline seem to be much less severe that the SOTS has implied in her public statements.
The meeting ended with Ms. Gill trying to focus on the areas on which we all apparently agree. However, except for agreeing that HAVA needs to be fixed, it is difficult for us to see substantial common ground. Ms. Gill did try to download Michael Fischer's cost analysis report from the TrueVoteCT web site but failed. We had a printed copy of it with us which we left with Anthony, who promised to FedEx it to her the same day. Hopefully, the report will have some impact on her and Senator Dodd's views on this issue.
To summarize, the main points that we can take away from this meeting are:
It is important that we continue to work for the passage of SB55, regardless of what happens to the various bill's being considered in Congress.
Ms. Gill (and presumably Senator Dodd) have a very troubling understanding of the problems associated with DREs. They seem to believe, against the views of almost all independent computing experts, that paperless DREs can be trusted to count our votes, as long as they are properly engineered and administered.
Ms. Gill (and presumably Senator Dodd) seem to have a fundamental antipathy against VVPBs, much of it based on erroneous assumptions about how VVPB technology works and on how well VVPB machines worked in Nevada.
One of the summarizing points that we tried to convey to Ms. Gill was that she and other supporters of DREs would do well to take a look at the big picture. When HAVA was introduced in 2002, it was greeted with optimism that it would help to improve confidence in our elections. In the elections since then that have involved DREs, that has not been the case. The first wave of alarms about DRE technology have largely been raised by computer experts and activists. But the wave is growing and unless these very important issues and concerns are addressed, we will be facing a tsunami of skepticism and cynicism about our election processes in the near future. As more people come to understand how these machines work and how our votes are being counted and possibly miscounted, they will increasingly come to see HAVA not as the law that helped America vote, but rather as the law that was used by the electronic voting industry to undermine confidence in our elections and to transfer millions of tax payer dollars into their coffers. That would be an unfortunate legacy for HAVA.