Stephen Ansolabehere's proposals for a federal initiative to research secure and reliable voting technologies and to develop, in particular, a four-stage architecture to promote voting accuracy, secrecy, and efficiency, all seem to me quite valuable. I am concerned only by his assertion of "decentralization" as a desirable, indeed necessary criterion for voting reform. His arguments for that criterion are inadequate and contradictory to his findings and his proposal. Those failings matter, because governmental decentralization is being widely espoused and implemented in American politics and law today, especially by Republicans—except when decentralization proves inconvenient for Republicans, as in Bush v. Gore. But though decentralization can have certain democratic benefits, in the U.S. decentralized voting systems have always been and remain impediments to the more extensive democratization that the American political system still needs.

Let me use an anecdote to illustrate some features of American politics that are largely invisible in Ansolabehere's discussion. Long ago, when I was a teenage Republican activist in my hometown of Springfield, Illinois, I was employed by the county government as a "runner" on election nights. Large boxes containing the punch card ballots from precincts would be brought into election headquarters and set on a table. Then each precinct's ballots would be fed into a computer, which would print out about twenty copies of that precinct's results, as well as the totals up to that point. I would take these copies and distribute them to the representatives of the media, the parties, and the campaign staffs who all sat waiting for me at little desks. For this I received twenty-five dollars a night and easy entry into the victors' celebrations, in which I participated on a discreetly non-partisan basis.

This experience reveals a major source of America's attachment to decentralized voting systems that Ansolabehere neglects. I got this cushy election night job, as well as many others, as a patronage position in return for service to my political party. Virtually everyone else working to administer the elections was also a patronage appointee. The reasons for that are rooted in the Constitution, but not in basic constitutional values. The Constitution's embrace of decentralization arose from pragmatic compromises rather than shared principles. And the constitutional system was actively hostile to the creation of political parties, which have, nonetheless, proved indispensable in making the system work. Parties have therefore been built up with difficulty, often as state and local patronage machines; and party leaders and the local officials they elect have long fought to keep much of the running of government and elections in their own hands—with the result that the parties often effectively control election-related jobs.

And they control them in more ways than one. When I was a lad, Springfield was closely divided between Republicans and Democrats, as it had been since the days of Lincoln and Douglas. Hence the elections room crowd was a bipartisan one that made overt voting theft difficult. Even so, as the precinct boxes piled up, one or two would occasionally disappear for a while, and when they reappeared, no one could know for sure whether they had more, less, or the same number of votes as they'd had before. We all knew, moreover, that in heavily Republican and Democratic parts of the state, party hacks moved up by making sure that their boxes produced robust vote totals for their side. Neither the patronage-appointed election officials, nor the local media (who were in bed with local elites or else locked out of what was going on), provided much check on these extracurricular activities. And in turbulent times like the late 1960s when I did this work, votes from predominantly black precincts were still especially likely either to disappear or to end up being surprisingly loyal to the reigning party. Local control has generally been a vehicle of not only partisan but also racial, class, and gender hierarchies throughout U.S. history.

Ansolabehere presents "decentralization" as a basic criterion for American electoral reform without addressing its origins and ongoing complicity in systems of patronage, corruption, and preservation of local elite dominance. He suggests inaccurately that it is a constitutional mandate. He cannot explain why; instead he acknowledges that Article II does not actually require such a decentralized system as we now possess, and that various other constitutional provisions such as the equal protection clause point toward greater federal control, given the widespread inadequacies of current procedures. He contends that we cannot hope to elect lots of different officials on the same day without local control of elections, but it is just not that hard to add offices to ballots for different locales. He invokes the old chestnut that decentralization provides for experimentation and innovation. But he also provides evidence that in fact the oldest voting system, paper ballots, is least likely to lose votes, a result that makes the value of past experimentation questionable. And in fact we do not need decentralized elections to test different voting systems, as Ansolabehere recognizes in his own proposal. He calls, after all, for a federal research initiative. And though his proposed four-stage architecture for voting would permit some local variations in the particular technologies employed, in fact it is a brief for federal action to make voting systems across the country much more fundamentally uniform, as well as reliable, and secure, than they are now.

That's why I like it. I don't like the contradictory, essentially rhetorical endorsement of "decentralization" as a voting system criterion. I suspect that endorsement represents an effort to placate the entrenched political interests that have long worked to prevent American elections and American politics from being responsive to the great bulk of our less advantaged citizens, particularly the poor, people of color, and women. But it is dangerous to placate those entrenched interests, dangerous to reinforce the sense that "decentralization" in electoral institutions is a key American principle—because Ansolabehere's technological proposals, useful as they are, represent only small steps toward the greater democratization of American political institutions that we need. American registration systems, election day schedules, districting systems, rules governing parties' and candidates' access to ballots, campaign financing systems, and much more, all need to be reformed. If we continue to genuflect to decentralization as a fundamental criterion for running U.S. elections, we make it much harder for such reform efforts to achieve true democracy in America.