Reliability of US Voting Systems:

Followup on the Oct 7, 2004 hearing

Part of the Voting and Elections web pages, http://www.cs.uiowa.edu/~jones/voting/
by Douglas W. Jones
THE UNIVERSITY OF IOWA Department of Computer Science

A response to the October 7, 2004
Congressional Black Caucus
Hearing on election preparedness
October 11, 2004

Introductory Note

I would like to thank the Congressional Black Caucus for the opportunity to speak at the October 7 hearing, and I would like to complement the Caucus on the variety of interesting and insightful questions raised by Caucus members at that event. A number of issues were raised at the hearing that we did not have an opportunity to address at the hearing. I will address these here, formulating each comment as a response to an issue raised by one of the caucus members.

Uniformity of our Voting System:

Representative Robert Scott of Virginia asked why we have so many different voting systems in the United States. Why is there not one uniform national system?

This same question was asked by members of the House Science Committee in their hearing on May 22, 2001, and I stand by my answer to that committee. If we were to adopt a uniform system of voting, nationwide, we would put ourselves at risk unless we were successful in selecting a genuinely perfect system.

Consider what would have happened if we had decided to adopt a uniform nationwide voting technology back in 1968. At that time, the most promising voting technology on the marketplace was the Votomatic punched card. Unlike the mechanical lever voting machines of past generations, punched cards offered the possibility of a voter-verified paper trail, they offer computerized vote tabulation, and they are economical. Not only that, but a Votomatic voting booth folds down to a package one person can easily carry, while the old mechanical lever voting machines weighed a significant fraction of a ton.

I believe these benefits would have easily convinced the nation, back in 1968, that the Votomatic system was the best choice. We would have standardized on this system, and by the time we found out its serious flaws, we would have been stuck with it. Standardization would have removed the incentive to invest in alternative vote counting systems.

The only way to avoid this risk of having a national standardized voting system is to commit to a long term program of federally funded research on voting systems at the same time the decision is made to standardize. This federal funding must be sufficient to compensate for the drying up of private capital that will be a natural consequence of eliminating the competitive marketplace for new voting systems that we have today.

If we are unwilling to fund voting system research and evaluation at the level I just suggested, we must maintain a competitive marketplace for voting systems, and to do this, we must continue to allow jurisdictions to purchase new systems independently.

Give that we keep the current basic framework, however, we can do considerably bettern than the status quo! Right now, if you look at the likelihood of your vote being counted, comparing two different counties that use identical voting equipment, there can be a considerable range! In election 2000, for example, voters in Washington and Escambia counties, in Florida, had almost 4 percent of their presidential votes overlooked, while voters in Clay County and Santa Rosa counties had only 0.7 percent of their presidential vote overlooked. All 4 of these counties used the same basic voting technology, in this case, Optech precinct-count mark-sense ballot tabulating systems. Where was the difference? It must have been in election administration, ballot design or voter instruction!

I do not mean to pick on Florida, but the intense scrutiny Florida faced after election 2000 gives us the data we need to find these examples. I suspect that the situation is quite similar in other states. I also do not wish to pick on the Optech ballot scanner. The statistics from Election 2000 in Florida for the Global (now Diebold) precinct-count optical mark-sense scanner show a similar range of variation.

I propose that we can and must reduce this range of variation, but since the variation we see does not appear to depend that much on technology, we must focus on promulgating and perhaps mandating best practices in election administration, best practices in ballot design, and best practices in voter education to solve this problem. This requires that we invest in a serious search for these best practices. We cannot just hold hearings to do this, but we must do research, digging into the source of variation with far more care than has yet been done. This will take research funding at a level higher than is currently available.

Partisan Election Administration

Representative Scott also asked why we tolerate strongly partisan statements from voting system vendors. Representative Maxine Waters of California raised a closely related issue when she asked why we allow elections to be administered by partisan political officials instead of demanding a nonpartisan election administration.

I believe that, as citizens of a democracy, we all have a civic obligation to study the issues facing our country and make judgments about the policies of our government. Any person who does this will draw conclusions that will align themselves with some candidates for public office and with some proposals for referenda, and against other candidates and referenda. In sum, I do not believe that it is possible for a responsible, intelligent, aware and informed citizen of a democracy to be nonpartisan. It is not wrong to say that a good citizen in a democracy has not only a right to be partisan but an obligation.

Because of this, the requirement that elections be run by nonpartisan authorities amounts to nothing more than a requirement that those who build the voting systems or those who administer elections refrain from making any statement of their partisan alignment. Since I know for a fact that they are extremely likely to have partisan preferences, asking that they avoid telling me where they stand denies me the ability to easily determine partisan bias so I can balance it.

What we must insist on is balance. For every partisan election worker leaning left, we must insist that they be balanced with someone leaning right. This is easy in areas where there is a healthy opposition, but it is difficult in areas where the opposition party has withered. Because of this, I support efforts to import members of the opposition party into jurisdictions where the local opposition party has an insufficient number of members to provide the necessary balance in election administration.

So, in areas of Chicago where there are too few Republicans, or in areas of downstate Illinois where there are too few Democrats, we ought to be importing election workers in order to guarantee that there are bona-fide representatives of both parties on the panels of election workers who run the precincts.

It may be appropriate to place some limits on partisan activities by election commissioners or members of boards of canvassers, but we must be careful about this. We could, for example, forbid state and county eleciton commissioners from serving as officers of campaign committees. Furthermore, it seems fairly obvious that election commissioners and members of canvassing boards should not administer elections in which they stand as candidates. This would be akin to a judge trying, in his own courtroom, a case in which he himself is also the plaintiff or defendant.

When it comes to bias among the owners or executives of voting equipment vendors, I believe the real issue is transparency in the conduct of elections. If we demand a sufficiently high standard of auditing and transparency, it should not matter if the devil himself made the voting equipment. (I am indebted to David Dill of Stanford University for that quote.) We can only achieve this if we require that all steps in the processing of ballots be genuinely open to observation and sufficiently well explained that the observers can understand what they are seeing, and if we require that the design and certification of voting systems be similarly open and observable. Unfortunately, across most of this country, we do not meet either of these requirements.

Federalizing Election Administration

Representative Shelle Jackson Lee of Texas asked about the possibility of federalizing the conduct of elections for Federal office.

It is noteworthy that Canada has long done this. Elections Canada conducts all federal elections in that country. Provincial elections are conducted by provincial elections offices, such as Elections Ontario, and local elections are conducted by the municipal or county election authority, as the case may be.

This has consequences! It means that there is nothing similar to a general election in Canada. The Federal election ballot is trivial! In the United States, this would never have more than three races, Congressman, Senator and President. As a result, paper federal election ballots are easy to hand-count in a very short time, and there is essentially no justification for using computer technology anywhere in the process.

If federal elections are conducted by a different office than the one that conducts state and local elections, we have the real possibility of duplication of infrastructure, with the federal election office buying one set of voting booths, ballot boxes and other election infrastructure, while the state and local election authorities buy duplicate equipment.

This change could also lead to a significant political shift. Having different election authorities for federal and state and local elections would destroy the institution of the general election. This would destroy the well known American institution of one candidate running on another's coat tails. With no federal candidates at the top of the ticket, turnout for state and local races could decline markedly, and linkage between national and state politics could be broken. I think this would radically change the nature of political parties in the United States.

In sum, making this change could have large numbers of unintended consequences if it is not thought through carefully. We have been conducting general elections for the past 200 years in the United States, with races as far apart as president and municipal constable appearing on the same ballot, and I would not recommend abandoning this institution without very careful thought!

What Can We Learn from the Rest of the World

Representative Chaka Fattah of Pennsylvania commented on the apparent success India has had with its new nationwide electronic voting system and wondered what else we could learn from the rest of the world.

This is a very important question! The United States is not the only democracy in the world. Brazil and India have moved aggressively to computerize the conduct of elections nationwide. Ireland made such moves but this spring, abandoned this move in response to public concern about the integrity of the paperless voting system being used. Argentina has experimented with voter-verifiable paper ballot printers in Tierra del Fuego. Several Cantons in Switzerland have moved to Internet voting in the last several years, and the Council of Europe has just issued recommendations for electronic voting systems.

There is one important factor that distinguishes our election system from almost every other democracy in the world. In the United States, we vote on everything from President to county soil and water district conservation board. An old joke among American voting officials is that a typical California voter casts more votes in one election than a typical British voter casts in a lifetime. This is true in contrast to most parliamentary democracies! In many parts of the world, you vote only for your member of parliament; the parliament then selects the prime minister and the cabinet, and the prime minister and cabinet select the regional governors, who, in turn, select the local government. In some of the world's democracies, this extends all the way down to the municipal level, while in others, local and provincial governments have some elective offices.

The Indian and Brazilian models of electronic voting are designed to support parliamentary elections with just one or two races on the ballot. The Brazilian model allows for hundreds of candidates per race, while the Indian model assumes smaller races. Neither model would generalize easily to a US-style general election. Generalizing the Indian model, in particular, would require the use of an electronic voting machine the size of an old-style lever voting machine.

It is noteworthy that, in both India and Brazil, there are reports of problems with electronic voting. In Brazil, despite protests to the contrary from the Tribunal Superior Electoral, there are charges from dissident groups that the Tribunal has effectively prevented meaningful oversight of election conduct by election observers. In India, gangs of thugs have begun to modify their traditional approach to rigging elections by shifting to new tactics that exploit the physical vulnerabilities of the new lightweight electronic voting machines.

I have been following the work by the Council of Europe on recommendations for E-voting. These recommendations were adopted on September 30, 2004 by the Committee of Ministers of the Council of Europe. Like the 1990 and 2002 Federal Election Commission standards, these standards are officially voluntary, but, as has happened with the FEC standards in the United States, there is a reasonable likelihood that they will be written into law by member states. Unfortunately, as with the FEC standards in the US, the CoE recommendations are deeply flawed. I strongly recommend paying attention to the evolution of these standards, but it would be a mistake to view them as a great improvement on anything done in the United States.

The most comprehensive statement of dissent about the Brazilian e-voting systems can be found on the web site of Pedro Rezende; note particularly this article:
  http://www.cic.unb.br/docentes/pedro/trabs/cryptobytes3.pdf

For those curious about the Council of Europe voting system recommendations, see:
  https://wcm.coe.int/ViewDoc.jsp?id=778189&Lang=en

My critique of the Council of Europe document is available on the web at:
  http://www.cs.uiowa.edu/~jones/voting/coe2004.shtml