Autonomy, equality, and decentralization. As Stephen Ansolabehere argues, the first two are “arguably essential” to any democratic scheme and the third is a deep feature of American political culture. Together they are appropriate guiding principles for any reform of the American election system. With that, I agree.

Stephen Ansolabehere has, however, specified these general values in a fairly particular way. According to Ansolabehere, we ensure autonomy “by requiring secrecy” and thus preventing vote buying; we ensure equality by “prevent[ing the] vote dilution” that counting fraudulent votes would cause; and we maintain decentralization in order to “achieve broad electoral control and technological innovation.”

These three general values, however, imply very different meanings to different people. To some, for example, autonomy means less the right to a secret ballot than the right to make an informed choice at the polls. To others, equality entails ensuring real access to the ballot more than preventing fraud. These loftily general values, in other words, hide some deep conflicts—conflicts that new technologies may not help us to overcome.

Consider how competing conceptions of equality might operate as constraints on technological design. For Ansolabehere, the “demand for equality…requires that all legitimate votes be counted, and that they must not be diluted by fraudulent votes cast by others.” Equality, in other words, largely entails counting every legitimate vote correctly and preventing voting fraud. It calls for a technology that accurately registers voter preferences and resists attempts at manipulation. Among existing technologies, in Ansolabehere’s view, paper ballots—either of the old-fashioned or optically-scanned type—perform these functions best and it may well be that no new technology can surpass them.

But this particular conception of equality leaves out a great deal. It ignores concerns about the relative accessibility and ease-of-use of a particular technology for different voting populations. Under the Voting Rights Act, for example, every jurisdiction must make voting materials available in the appropriate language to any language minority group that comprises at least 5 percent of the jurisdiction’s voting-age population. In the next decade, that means that Los Angeles County will have to provide county-wide voting materials in more than ten different languages. Traditional and optically-scanned paper ballots make satisfying this requirement very difficult. If every ballot, not to mention any supporting materials, has to be printed in more than ten different forms, the expense quickly becomes unsustainable and paper management alone becomes a nightmare. Paper architecture, in other words, frustrates this competing conception of equality.

It is no accident that Los Angeles and Riverside Counties, which are both linguistically diverse, prefer punch card and DRE technologies. Neither requires expensive preparation of physical materials to make the ballot accessible to language minority groups. With punch cards, for example, the same blank card works for all voters. Only the instruction frame on the voting machine, which points to where to punch, has to be available in different languages. Paper architectures, on the other hand, make satisfying the particular conceptions of equality that the Voting Rights Act embodies financially difficult.

The case of blind voters is even more interesting. Existing law clearly grants citizens with disabilities the right to be assisted in voting by someone of their choice. Many jurisdictions believe that providing such assistance exhausts their equality obligations to the blind; they make no effort to support technologies that would enable the blind to vote without assistance in secret. Although some advocacy groups have pressed claims that the Americans with Disabilities Act guarantees the blind the right to vote in secret (and such technology is currently available), the courts have been hesitant to require it.

The claims of the blind, though, pose the same conflict over competing conceptions of equality: the technologies that most easily promote equality in the sense of ensuring accuracy and preventing fraud do a lousy job of promoting equality in the sense of ensuring access to the blind on terms similar to those of other voters. Paper ballots of either the traditional or new sort throw up serious obstacles to this group’s full participation.

This same type of conflict occurs over different conceptions of autonomy. As Ansolabehere specifies that value, it largely concerns secrecy. Yet, as Oregon’s experiment with mail-in voting shows, secrecy can come at the expense of other conceptions of autonomy. Under Oregon’s system, voters fill in their ballots wherever they want and then mail them in or drop them off at central collection points. There are no traditional polling places. Some defenders claim that this way of voting allows voters to become better informed. People can fill out their ballots as they discuss the issues and candidates with those around them. Others argue that this informality makes it too easy for others to influence one’s vote. If choices are not made in secret, others will have a better opportunity to disturb the voter’s will. A strong family member, for example, might coerce the votes of weaker members of the family.

I do not mean to take either side in this debate. Both views are partially right. Mail-in voting will probably cause some voters to make more informed and better choices and will probably lead other voters to vote in ways that reflect preferences other than their own. Promoting autonomy as secrecy may make one form of coercion more difficult but also may degrade some voters’ decision-making. It is simply impossible to say whether secret voting or mail-in voting promotes autonomy in the abstract.

It is hopeless, of course, to expect any technology to escape these difficulties. The conflict is loaded into the values themselves. Some may accuse Ansolabehere, though, of advancing but not acknowledging one particular interpretation of these values. If autonomy and equality do not necessarily entail such an exclusive focus on secrecy and fraud prevention, Ansolabehere’s design constraints aid particular conceptions of democracy at the expense of others. Indeed, to some, this architecture is apt to appear somewhat partisan. In the current debate, Republicans tend more to see election reform in terms of fraud prevention while Democrats tend more to see it in terms of civil rights. These are two somewhat opposed ways of framing the debate, both of which resonate in equality. But to Democratic partisans at least, the assumptions behind Ansolabehere’s architecture are likely to appear Republican.