PERIODICITY: WHY REPRESENTATIVES SHOULD NOT CONTROL THE UNITS OF REPRESENTATION

In most states in the United States, the legislature still has the dominant role in the contentious process of redrawing electoral districts to reflect shifts in population since the previous census. Even in most of those states that have redistricting commissions, political parties choose at least some of the members.¹

Typically, districts are drawn to protect the party in power and the seats of incumbents. The party fortunate enough to control the state legislature thus controls the redistricting. Less directly but usually no less effectively, parties in each state govern the process that draws the Congressional districts.

On its face, this practice seems objectionably undemocratic. It undermines popular control by giving the representatives who are to be controlled significant influence over the means by which they are controlled. The representatives or their party colleagues create the districts and, therefore, define the electorate. They determine the set of citizens who are to be represented and who are to judge the performance of the representatives. The process appears to be not one “in which the people select their representatives,” but one in which “the representatives have selected the people.”²

The problem could be avoided or at least mitigated by adopting a different electoral system. For example, district boundaries would be much less important in a system of multimember districts, which could be much larger than current districts. A multimember system would also permit the use of cumulative voting. Because in this system each citizen may cast as many votes as there are seats to be filled, the district would not be as likely to be dominated by one party. More radical reforms would abolish districts altogether and elect representatives by statewide or national forms of proportional representation. In many of these systems, voters in effect form their own districts, and each representative has a unanimous constituency. The various systems have comparative advantages and disadvantages.³

Yet none has been adopted on the national level in the United States, and in most jurisdictions in the United States the winner-take-all district systems are likely to continue to prevail. As long as districts exist in their current form, they will need to be redrawn periodically.

The current practice of granting redistricting authority to legislators has frequently been criticized, but none of the usual arguments against it is entirely satisfactory. First, it is claimed that redistricting should not be subject to a political process at all because the standards of a fair electoral process are objective. They express a form of perfect procedural justice, rather than pure procedural justice. The districts should match the population within certain limits, allowing for some variation to respect existing political jurisdictions. As far as possible, districts should respect the constitutional standard of “one person, one vote.” But this standard is not very helpful in making the hard choices involved in redistricting. Strictly interpreted as requiring every vote to count equally, it would prohibit patterns that give any category of voter more weight than any other; only a random distribution would be acceptable. Even in random redistricting, votes would not be weighted equally for most categories because people do not distribute themselves randomly among districts. Interpreted more permissively, as it has been in practice, the standard allows officials to take into account some of the factors that have traditionally determined the boundaries of districts—such as geographical features, historical communities, partisan advantage, and even incumbency protection. Armed with powerful computer technology scarcely imagined at the time the standard was proposed, politicians can now create equal districts that give more weight to any of a wide range of different categories.

Redistricting therefore involves making choices among a complex set of values. There is no objective or neutral standard that can determine these choices. Choosing is an unavoidably political process. The most recent criteria to be adopted by a state [Arizona constitution, art. IV, pt. 2, sec. 1(14)] are typical of those found in many other states: equal population, compactness, contiguity, community of interest, geographical saliency. Plainly, the criteria can yield conflicting results, and choosing among them requires deciding which should weigh more or less heavily. Interpreting and applying the criteria, as well as making the other choices between the various goals we have canvassed, calls for political judgment (Butler and Cain 1992). As long as minority rights are not violated, the political process—specifically the legislature—would seem to be the appropriate place to decide these inherently political questions. The argument from objectivity or neutrality thus does not support the case against legislative control of redistricting.

A second argument appeals to the importance of competition. Letting legislatures control redistricting, it is said, is likely to reduce the competition necessary for a healthy democratic system.⁴

Competition requires that voters have real alternatives. Partisan balance may shift from election to election, and in some states and the nation as a whole the division between the parties may be close. The 2002 midterm Congressional election was in that respect highly competitive. Even in safe districts competition in primaries can be keen. But the choice that most voters face is in the race in the general election in their own districts. That is the competition that matters most for their voting decision.

In most states citizens can vote only for the candidates in their own party primary. In most districts, incumbents almost always win, and by large margins. One of the most striking patterns in contemporary legislative politics in the United States is the high rate of reelection of incumbents not only in Congress but also in state legislatures (Ansolabehere and Snyder 2002a; Center for Voting and Democracy 2002; Cox and Katz 1996; Ferejohn 1977; Gelman and King 1990). The high rate of reelection may indicate that most voters are satisfied with their incumbents, or that incumbents have an unfair advantage from the resources of office, or simply that incumbents and strong challengers do not run when they expect to lose.⁵

Even if legislative control of redistricting is not the principal cause of the lack of competition, continuing to give the most power over the redistricting process to incumbents or to the currently dominant party would not seem to be the most effective way to encourage more of it.

Although more competition may be desirable in most district elections in the United States, the case against legislative control over redistricting cannot rest mainly on this goal. In the first place, electoral competition is not an end itself, but a means of allowing change that voters may want. It may also be a necessary means of discovering whether voters want change, or helping them recognize that they want it. But even as an instrumental goal, electoral competition can be misleading if it is conceived, as it usually is, on the model of economic competition.⁶

It neglects forms of contestation that do not readily fit within the categories assumed by market analogies. It gives more attention to the interplay of interests and preferences than to the interaction of principles and ideas. Furthermore, the preoccupation with making electoral contests more competitive distracts attention from other equally worthy goals, such as recruiting and retaining better qualified candidates and controlling the growth of the cost of campaigning. More generally, concentrating too much on promoting competition may distract citizens and legislators from attending to the business of government. The goal should be to find the optimal level of competition, not to maximize it.

In the second place, although there is still some dispute about the sources of the high rates of incumbent reelection, there is little evidence that partisan control of redistricting is the main cause (Persily 2002). States that have redistricted do not consistently have higher rates of reelection than states that have not done so. Incumbents also tend to be reelected in elections that have no districts (such as gubernatorial and senatorial races). Furthermore, politicians usually produce plans with districts that are more competitive than those that existed prior to redistricting. Redistricting tends to increase responsiveness (the degree to which the partisan composition of the legislature responds to changes in voter partisan preferences) and reduce partisan bias (the degree to which an electoral system favors one political party in the conversion of the total vote into the partisan division in the legislature).⁷

Two factors seem to account for this tendency: Parties have multiple and conflicting goals (protecting the party may require sacrificing incumbents), and the electoral effects of any redistricting plan are more uncertain than is often assumed.

The third argument against legislative control of redistricting comes closer to capturing its fundamental problem. Giving politicians or their partisan colleagues control of their own means of reelection, it is claimed, undermines accountability, one of the main purposes of elections.⁸

Politicians are accountable if they must face an electorate that can decide on the basis of their past performance whether or not to remove them from office. When politicians can adjust the boundaries of their district and manipulate other electoral rules to preserve or increase their relative advantage, they weaken the ability of voters to decide their fate. Some of these effects continue throughout their term until the next redistricting. When representatives believe that they are less vulnerable to defeat, they may while in office pay less attention to all constituents, and even less attention to constituents who are not in the majority that elected them. Also as a result of redistricting, voters who wish to decide on the basis of the past record more than future performance may find that the representative whose record they would judge is running in another district. Further, they may find themselves placed in a district where they are no longer in the majority and are now unable to discipline the party whose record they wish to judge.

Although control over redistricting does give politicians a tool to avoid being called to account, the argument against control cannot rest entirely on the value of accountability. First, even representatives in safe seats generally act as if their reelection is in doubt and therefore tend to be responsive to their constituents (Mann 1977). Second, some voters vote prospectively, rather than retrospectively (Fearon 1999; Powell 2000). The fact that your district has been altered, that you find yourself in a different district, does not matter so much if you are choosing on the basis of a prediction about how candidates may act in the future.

Third, the argument from accountability is incomplete because it does not recognize the significance of changes in the identity of the legislators to be held accountable, and the voters who are to hold them accountable. With respect to the identity of legislators, it does not cover many cases where politicians do not run for reelection. Some cannot be held accountable even if the districts remain the same. With respect to the identity of the voters, the argument does not take into consideration the shifts in the composition of the district. The majority who elected the representatives at the last election cannot be the same majority that is calling them to account at this election. New voters arrive, some voters move, and others die. The greater the interval, the more the change in the identity of the electorate, but some significant change is likely to occur between most elections. Consequently, an argument based on accountability that assumes a link over time between specific representatives and a determinate set of constituents cannot constitute a complete case against letting politicians control the process.

The more fundamental objection is that legislative control of redistricting does not adequately respect periodicity.⁹

Elections are not one-time events. An election is not like the ratification of a constitution or its amendments, a popular initiative or referendum, or even a typical plebiscite. Each election, though a discrete event, stands in an indefinite series. Because one purpose of periodicity is to enable voters to hold representatives accountable for their actions, the accountability objection is to that extent correct. But that objection depends on assuming that there is more continuity in the electorate and the elected from one election to the next than there usually is. Even if there were no continuity, periodicity would be important. It provides the means by which present majorities can escape the dead hand of past majorities. Elections must be held periodically so that, when new majorities replace old majorities in the electorate, they can then decide to what extent to replace an old with a new legislature. Periodicity keeps open the possibility of change—specifically, change in the membership of the legislature and thereby in the nature of the people's representation. In this respect, periodicity is an expression of the democratic value of popular sovereignty.¹⁰

To complete the objection to legislative control of redistricting, periodicity must be combined with a principle of authority, which specifies who should decide the rules that determine whether and to what extent change occurs in the legislature. The principle holds that decisions about such rules should not be made by people who have a preponderant interest against (or for) change in the membership of the institution in question (Thompson 2002, 133–34, 166–68, 179–80). The point is not that change is always desirable. The aim is to try to ensure that choices between change and continuity—in this case, decisions about redistricting—should be made on their merits, giving due weight to the many relevant factors that affect the democratic process. The assumption is that these decisions are more likely to be made on the merits if legislators and their parties do not make them. When an old legislature elected by the old majority sets the rules that define the new majorities that will elect the new legislature, this possibility of renewal and change in membership is less likely. Legislators, like members of most institutions, generally approve of the mechanisms by which they were selected.

The principle extends an idea originally expressed by James Madison during the debates on the Constitution. He sharply distinguished ordinary legislation from electoral regulation. We can trust the normal process of representation, he said, provided that the issue under consideration is one in which representatives share a common interest with their constituents, not one in which they “have a personal interest distinct from that of their constituents” (Madison [1787] 1966). But when representatives decide questions that affect their own status or that of their party, they will tend to give more weight to preserving their privileges and more generally perpetuating the practices of the institution. They will give less weight to promoting other values of the democratic process. On such issues, we should be “jealous” of assigning the representative body final authority.

We do not have to assume that the only problem is the “personal interest” of individual legislators in reelection, or even that this interest is always opposed to the interest of their constituents or the public interest on questions concerning the electoral process. Legislators may conscientiously (sometimes even correctly) believe that the best interest of the institution and the democratic process lies in preserving the current membership or at least the current rules for choosing members. But with regard to the rules that determine who is to be elected, citizens have good reason to doubt that legislators and their parties are impartial judges. If legislators do not pursue their individual or partisan self-preservation, they are likely to engage in institutional self-perpetuation. Whatever differences the parties in the legislature may have, they share a common interest in minimizing challenges from outside. They are likely to join in opposing term limits and campaign finance regulation, which reduce the influence of incumbency; and they are likely to work together to discourage fusion candidacies, which help third-party challengers.¹¹

Redistricting and other electoral measures—and the choice between change and continuity—are more likely to receive consideration on their merits if those who are deciding whether to adopt them are not the same as those whose electoral future they would determine.

The Madisonian principle refers to the circumstances of judgment rather than any particular conclusion that politicians might reach. It is concerned with the reasons that redistricting officials are likely to consider (and the institutional incentives that favor some reasons over others), rather than the results officials reach (and the institutional legitimacy of any particular result). The principle therefore does not necessarily require more (or less) competition in elections or more (or less) turnover in legislatures. Nor is it satisfied by showing that the competition in the system as a whole is greater than in most districts or that representatives in safe districts are responsive and accountable to their constituents.

Moreover, even if the high rate of reelection of incumbents is the result of strategic entry or exit decisions rather than any advantages incumbents may enjoy, the principle would still cast doubt on giving the dominant party control over redistricting. That party would still be deciding how to redistrict in circumstances in which it would be less likely to give due weight to all the relevant values at stake. For example, a party attempting to maintain control of the legislature could decide it had to pit two experienced and popular incumbents against each other, even if the decision caused one to withdraw from the race. Or a party seeking to create safe districts may need to produce more ideologically homogeneous districts, which could have the effect of increasing partisan polarization. A party could prefer a Democrat who is more liberal or a Republican who is more conservative than might be optimal for the political health of the process. A districting decision that has the effect of favoring less moderate and less centrist candidates may be justifiable in some circumstances. But as with other decisions that have systematic effects on the democratic process, the Madisonian principle implies that the parties are not the best judges of whether the results of such a choice serve constituents, let alone the public interest.

The value of periodicity combined with the Madisonian principle implies that the authority for governing elections in general and redistricting in particular should be located outside the ordinary legislative process. The need for independent judgment may suggest that courts should have this authority. Certainly, courts have an important role to play in regulating the electoral process. They are well suited to protecting individual rights, notably the constitutional requirements of equal protection and free speech, and are appropriately involved in redistricting cases that involve racial justice.¹²

But questions about change in the composition of majorities—to what extent districts should permit or encourage new majorities to form—raise issues that are less about individual voters than about institutional structures and the nature of representation itself. The conflicting demands of democratic majorities neither state legal claims that individuals usually have standing to bring nor permit remedies that courts usually have competence to apply. They involve the form of government—in constitutional terms, the guaranty clause—which courts are not well placed and are usually reluctant to adjudicate.¹³ Furthermore, courts are expected to respect precedents, and therefore are bound, more than other institutions, to the will of past majorities and to the decisions of supermajorities in constitutional ratification.

A more promising way to exercise control over redistricting (and potentially other aspects of the electoral process as well) is through an independent commission. Australia (since 1902) and Canada (since 1964) have relied with evident success almost completely on nonpartisan commissions for this purpose.¹⁴

In the United States, six states now authorize commissions to redraw their own legislative districts, and eight more states use commissions to revise congressional as well as legislative districts.¹⁵ The most recent state to join this movement is Arizona, whose voters in the 2000 election approved an initiative “creating an independent commission of balanced appointments to oversee the mapping of fair and competitive congressional and legislative districts.”¹⁶

A well-composed commission is better placed to give due weight to all the relevant factors that should be considered in drawing districts—equal population, compactness, contiguity, community of interest, geographical saliency, and even partisan politics. Members of commissions can view their role not as registering preferences of constituents, or even acting on their own conscience and convictions, but as working together to identify and express the will of the appropriate majority.

Most states acknowledge the unavoidably political nature of redistricting decisions by permitting former politicians to serve on their commissions (though most bar current and recent office holders). Commissions are properly political creatures, but they can cultivate a politics different from, or at least more wide-ranging than, the ordinary kind. They can be designed so that they do not serve entrenched majorities of any time or place.

But politics probably intrudes too much into the appointment process of most commissions now. Legislative leaders typically appoint the majority of the commission members, sometimes from a restricted pool of candidates (for example, nominated by a judicial committee). If the selection process were more independent of the legislature, commissions could better take into account all the political factors that legislatures consider but give greater priority to providing opportunities for changes in membership than to preserving the privileges of current members.

Commissions are less democratic than legislatures in the sense that they are less directly accountable. But the democratic authority of a commission could be maintained in various ways: by ensuring that voters have an opportunity to approve its constitutional role (its scope of authority and general procedures) and the nature of its membership (the qualifications of members and their method of selection). The loss of some direct democratic control in the present is surely worth the gain of greater democratic control in the future. Commissions have that negative virtue so critical to the democratic process: They are not self-perpetuating. In this way, they help fulfill a central democratic purpose of electoral periodicity.

SIMULTANEITY: WHY CITIZENS SHOULD VOTE AT ABOUT THE SAME TIME

On election night 2000, before all the polls had closed in Florida, the networks called the state for Al Gore. (By then, Florida's electoral votes were seen as critical to victory.) Some four hours later, the networks changed their minds, and declared George Bush the winner. About two hours later, they retracted that call, but only after Gore had phoned Bush to concede. The reports may have caused some people not to vote.¹⁷

Whatever the effects on turnout, the projections continued to shape the opinion and events in the month-long controversy that followed. In the House hearings devoted to election night coverage, Democratic representative Henry Waxman argued that Gore's “concession” prompted by the networks' projection of a Bush victory “set in motion a chain of events that were devastating to Al Gore's chances. And it immeasurably helped George Bush maintain the idea in people's minds that he was the man who won the election” (Political Staff 2001; U.S. House Committee 2001). Republicans also complained that the earlier call for Gore caused some of their supporters in the western states to stay home (Correspondents 2001; Political Staff 2001).

In the aftermath of this election, much of the criticism centered on the errors the networks made rather than the effects their projections had on turnout and legitimacy. In the House hearings, several network witnesses conceded that they had made errors and needed to change their methods but said they doubted that projections had any harmful effects (U.S. House Committee 2001). The president of the Associated Press heatedly declared that the mere act of holding these hearings was itself a threat to freedom of the press. Some of the networks seemed inclined to make modest reforms, but none wanted Congress to legislate in this area. An outside group recommended that the networks cease using exit polls to call elections, stop relying on one source to collect and collate data, and undertake organizational changes to ensure that accuracy in reporting takes priority over speed (Konner et al. 2001). Clearly these and similar reforms, whether legislated or not, would be desirable. If the media are to make projections and report polls, it is important that they do so accurately.

But the preoccupation with accuracy neglects the more general problem—the potential damage to the democratic process that results from reporting projections even when they are accurate. It is not simply inaccurate projections that we may wish to limit, but any projection that gives some people but not others information relevant to deciding how or whether to vote while the election is in progress. (In this respect, the greater the perceived accuracy, the greater the potential harm.) We can begin to develop the justification for limiting such projections by examining two common arguments against doing so.

The first appeals to freedom of the press. Although merely holding hearings on proposals to regulate projections does not of course violate freedom of the press, prohibiting the press from publicizing projections could do so. An outright ban on reporting exit polls would rightly be considered a violation of the First Amendment. Yet freedom of the press is not absolute in the electoral sphere. The courts have permitted greater regulation of the media in election-related speech than in many other areas.¹⁸

Electoral politics, as will be shown more directly in the next section, should be regarded as a special sphere, potentially subject to greater regulation than ordinary politics.

But even without directly regulating the media, the government could discourage electoral projections. There is no First Amendment objection to mandating uniform poll closings or embargoing the release of official returns (Ortiz 2002). Uniform closings would make the projections less significant. Embargoes on returns would make exit polls less reliable or more expensive.¹⁹

Furthermore, the networks could voluntarily agree to forgo making projections until all the polls are closed in all states. Finally, and most importantly, the main purpose of freedom of the press is to protect not the press but citizens and the democratic process.

If we can overcome the argument from press freedom, we still face the argument from voter freedom. Restricting election night projections, it is said, would deprive voters of the chance to vote on the basis of information some think they need to make a decision, and therefore would impair their free choice. A “true freedom to vote as one chooses” includes “the right to not vote at all or even to vote as everyone else does” (Barlow 1990). The assumption evidently is that free choice requires providing the option of deciding not to vote because you believe other voters have already determined the outcome, and the option of basing your vote on how others are voting. “Neither the networks, nor the lawmakers should be permitted to decide which information is useful to voters and which is not…. The choice must belong to the individual” (Barlow 1990).

This argument misconceives the nature of free choice. In a process in which projections are publicized, citizens may choose what information they wish to use, but they are denied some other kinds of choices. They cannot choose how information about their own choices and those of others is used. They vote in secret but they do not have the option of keeping the results of their votes secret until all of their fellow citizens have voted. They would have that option in a different system, but the choice of the system does not “belong to the individual.” Free choice is not simply a matter of individual decision, but also of institutional structure, which only citizens acting together can choose.

Although restrictions on publicizing projections would limit free choice in some respects, they could enhance it in others. Citizens may reasonably decide that the individual choices they make in an election are more valuable if no one has information about how they voted until the polls have closed. When citizens decide to limit the range of choice for the sake of improving the value of choice, they are choosing between different aspects of liberty, not between liberty and other values. Their decision may of course also take into account other values, such as the quality of the democratic process, but it is not correctly characterized as favoring collective over individual choice. Other countries that respect individual choice regulate not only publication of election results but also public opinion surveys in the period immediately before the election (Feasby 1997, 242).

The argument so far shows only that there is no presumption in favor of publicizing projections—that freedom of the press and voters would not be violated by restricting projections—but it does not provide any positive reason for such restrictions. Why should citizens choose a system that reports the results all at once? We can formulate a positive justification for restricting projections if we take notice of the second temporal property of elections—simultaneity.

In Congress and many other legislative bodies, members vote one at a time (as in a roll call vote) and then at the conclusion have the chance to change their votes after they see the provisional totals. How they vote is affected by how others vote. In an ordinary election, each voter acts as if he or she is voting at the same time as every other voter. No voter knows how others have voted until everyone has voted. No one can change a vote in response to how others have voted. And everyone votes with at least potential access to the same information. The constitutional requirement that all states hold elections on the same day (art. II, sec. 1) may be read as an expression of the idea that as far as possible the electoral experience should be the same for everyone. No more than 24 hours elapse from the casting of the first vote (usually just after midnight Eastern Standard Time in Hart's Location and Dixville Notch, New Hampshire) to the closing of the polls at 6 p.m. Hawaiian-Aleutian Time. Election-day customs similarly reinforce the idea. Candidates stop campaigning when the voting starts, and supporters are not permitted to demonstrate near the polling places. While voting continues, politics ceases.

The simultaneous character of an election is not an arbitrary or merely conventional procedural requirement. It has a normative rationale.We should distinguish two aspects of simultaneity, because each rests on somewhat different values. The first is voting at the same time; the second, voting without regard to how others vote.

The practice of requiring citizens to cast their ballots at more or less the same time, usually on the same day, rests partly on the value of popular sovereignty. The more that voting is concentrated in time, the more that the election expresses the will of a determinate majority (or plurality). The outcome can then be seen as the result of a single collective decision rather than the product of a series of decisions made by different majorities (or pluralities). Simultaneous voting in this way creates a more coherent popular sovereign.

Simultaneity rests even more importantly on the democratic value of fairness. If citizens vote at the same time (or have only information they would have had if they were voting at the same time), the value of each citizen's choice is no greater than that of any other citizen. All make their choices with equal access to relevant information. Election projections deny some voters information that other voters have. Also, to the extent that election projections discourage efforts by parties and candidates to mobilize voters in the western states, some citizens who might have voted do not make a choice at all. To the extent that information is unevenly distributed, an election is less fair.

The unfairness can be seen most clearly by considering cases in which a violation of simultaneity is more extreme. In the aftermath of the 2000 presidential election some people proposed to resolve the controversy by rerunning the election in several counties in Florida.²⁰

Even if a rerun had been practicable, it would have clearly violated the simultaneity norm. The Palm Beach county voters would have been voting with the specific knowledge that the election was very close. Many would have probably voted differently (shifting their votes from Nader to Gore, for example).

Such a violation of simultaneity should be distinguished from a different kind of strategic voting: the decision in a single election to vote for your second-choice candidate (e.g., Gore) instead of your first-choice candidate (e.g., Nader) in order to decrease the chances of your third-choice candidate (e.g., Bush). This kind of strategic voting is neither a violation nor a consequence of simultaneity. It is consistent with simultaneity because it takes place in a single election. It is a consequence of the winner-take-all rule that operates in most current systems in the United States. It would not be necessary, for example, in a system that uses forms of proportional representation or even one that employs the instant runoff method. The latter method provides a way of combining what would otherwise be sequential elections into a single simultaneous election.²¹

Fairness is not the only value at stake in simultaneous voting. When citizens go to the polls on the same day, publicly participating in a common experience of civic engagement, they demonstrate their willingness to contribute to the democratic process on equal terms. Going to the polling station and standing in line with one's neighbors may not rank among the most exciting moments in life, but election-day voting serves an important expressive purpose in a democracy. It enables citizens to express their commitment to a common project in which they participate on equal terms—the process of choosing the leaders who will represent them. When the media publicize projections, they undermine this sense of participating in an important civic activity. Voters in the western states lose the chance to take part with their fellow citizens in an event that is still in progress rather than one that is already on its way to the history books. The same media practice that causes the electoral process to treat some citizens unfairly also causes it to convey to all citizens the message that civic engagement is not all that important.

The expressive value of simultaneity has further practical implications. It strengthens the case for making election day a national holiday.²²

Because most people who are registered already go to the polls on election day, making election day a national holiday is not likely to increase turnout. It could even reduce turnout if people decide to travel instead of vote. However, the reform would enhance civic engagement in other ways. Holding the election on a national holiday not only strengthens the sense of participation in a shared civic activity, but also more concretely increases the number of election workers and polling locations. The improved facilities promote civic engagement on equal terms by making voting more accessible.

The norm of simultaneity also casts doubt on the increasingly widespread use of early voting—absentee ballots and voting by mail. Voting before election day has nearly doubled since 1980.²³

In Oregon all statewide elections have used a mail-in ballot since 1995. In the 2000 presidential election, a quarter of the California vote and two-fifths of the Texas vote were by absentee ballot. Early voting weakens the value of the experience of participating in a civic activity. The public affirmation of voting together openly on the same day is quite different in meaning from the private transaction of filling out and mailing in an absentee ballot. In addition to this expressive value, such activities can contribute in modest ways to the development of the capacities of equal citizenship. Participating openly with others is more likely to reinforce civic attitudes than is voting alone at home. Voting alone may be worse than bowling alone.

To be sure, the value of this kind of civic engagement is not so important that it should take precedence over fairness. Absentee voting is certainly justified for citizens who would otherwise have difficulty in going to the polls on election day—especially some older voters, disabled persons, and members of the armed forces. Absentee or other forms of early voting could also be justified if they substantially increased turnout especially among groups that are now underrepresented at the polls. But so far absentee voting has increased turnout only modestly if at all.²⁴

Unrestricted absentee voting and the further extension of other kinds of early voting may even reduce voter participation by discouraging efforts to mobilize voter turnout. Furthermore, the effects of any increase in turnout have not been equally distributed. Whites are twice as likely as blacks to vote by absentee ballot. Thus, while measures that erode simultaneity may be warranted if they make voting more fair, they are harder to justify if they are implemented only for the purpose of making it more convenient.²⁵

The second aspect of simultaneity is independence of voting—a norm that holds that citizens should not adjust their votes according to how others have voted. This partly rests on fairness. An election that enables citizens to adjust their votes in this way privileges strategic voting. Strategy of course has a place in the campaign, but the value of a vote should not depend on the strategic savvy of the voter.

The norm against strategic voting does not rest entirely on the value of fairness, however. We can imagine an internet voting system that enabled each voter to vote at more or less the same time (with access to the same information and with the opportunity for the same civic engagement), but with the option of changing one's vote after seeing the provisional totals. There would of course be significant practical difficulties with any such system. A limit on the number of times voters could change their votes would need to be established, and a deadline by which all voting would cease would have to be set. Protections against the use of “sniper” software would be required so that users could not exploit the advantage of switching at the last possible moment, as occurs in some online auctions. But assume that we could adequately protect against the many forms of potential fraud and manipulation, and also ensure that the technology is highly reliable and the devices readily accessible to all citizens. Is there any objection in principle against this kind of interdependent voting? What is wrong with adjusting your vote in light of how others are voting? It might seem that deliberative democrats especially ought to welcome this kind of interaction.

Although voters may certainly take into account the views of their fellow citizens, the kind of vote switching that would occur in a typical online system would be more likely to reflect conformist attitudes than deliberative interactions. This vote switching would undermine not only the norm that voters should make up their minds on the basis of what they hear in the campaign, but also the norm that voters should exercise independent judgment. Of course, many voters do not exercise much independent judgment in the current system, but online nonsimultaneous voting would create additional and specific incentives for voters to decide—suddenly and often irreversibly—to follow the crowd. It would invite politicians and their campaign managers to encourage just that kind of decision making. Whether or not voters or politicians generally acted in these ways, such a system would express in an especially pure form a denigration of independent judgment.

Simultaneity applies not only to individual voters but also to institutions. In the midst of the disputes after the 2000 presidential election, the Republican-led Florida House of Representatives resolved on December 12 to appoint its own slate of presidential electors, all reliably committed to George W. Bush (State of Florida 2000). The House majority feared that the courts or Congress would challenge the state's electors to the electoral college. To ensure that Florida's electoral votes would go to Bush, the legislators invoked this rarely used constitutional authority. Had the state Senate followed suit, the House would have decided the outcome of the presidential election for the nation. (As it turned out, the resolution proved unnecessary because the U.S. Supreme Court stopped the recounts, and the Florida Senate adjourned without taking action.)

The problem is not that legislators were acting as partisans. Indeed, it could be argued that this is exactly how they should have acted because they were elected on the basis of their party affiliation. The problem is rather that legislators were deciding the election in light of how other voters cast their ballots (as well as information about the vote totals in other states and the totals and margins of the candidates). Deciding the outcome under such conditions undermines simultaneity (Thompson 2002, 151–61).

The norm of simultaneity applies directly only to single elections and, therefore, does not necessarily rule out sequential elections such as the presidential primaries that the national parties hold in many states. Some of the same considerations (such as strengthening the sense of civic engagement) that support same-day voting would also favor a national primary or at least regional primaries. But sequential primaries do not raise fairness concerns to the same extent as does the strategic legislative selection of presidential electors. Although the outcome of the New Hampshire primary often has substantial influence on the subsequent primaries, the decision is not final in the way that the Florida legislature's would have been. In the absence of a compelling argument that the lack of simultaneity in successive elections is unfair, other considerations carry more weight. Preserving some version of the current mixed system of state primaries and caucuses may then be justified by pointing to the greater opportunities it provides for challenging candidates in many different settings, and the lesser scope for influence it gives money and the media in the nominating process (Polsby and Wildavsky 2000, 232–39).

Simultaneity in its various forms is thus important for reasons of fairness and civic engagement. Not merely voting, but voting at more or less the same time, plays an essential role in the democratic process. Democracies should therefore try to avoid the anomaly created by practices that permit citizens increasingly to choose the time they vote and should preserve as far as possible the practices that encourage voting together on a single election day. They should disfavor publishing the results of exit polls, expanding the use of absentee balloting, and granting legislatures the authority to select presidential electors on their own schedule.

FINALITY: WHY CAMPAIGNS SHOULD BE REGULATED MORE STRICTLY THAN ORDINARY POLITICS

In declaring the expenditure limitations of the 1974 campaign finance law unconstitutional, the Supreme Court, in Buckley v. Valeo, held that “only expenditures for communications that in express terms advocate the election or defeat of a clearly identified candidate” may be regulated (424 U.S. 1 [1976] at 44). The distinction is between financial support for advocating the election of candidates (which may be regulated) and support for advocating a particular policy or general program (which must be left unregulated). Candidate advocacy is part of a campaign, and is only partially protected under the First Amendment. Issue advocacy is part of ordinary political debate, and is fully protected.

But the line between the two kinds of advocacy was anything but clear in the Court's opinion, and subsequent attempts to clarify it have proved difficult. The problem goes beyond the disputes about the differences between contributions and expenditures, the distinctions between money and other kinds of influences, and the controversies about private and public financing (Issacharoff, Karlan, and Pildes 2001a, 533–45). The problem is not completely resolved by any reform proposed so far. It would remain even if Buckley were overturned. Although the problem of issue advocacy has received less attention than that of soft money, its practical effects have been no less pernicious, and its theoretical implications may be even more significant.²⁶

For many years after Buckley, the criterion for express advocacy was whether a communication used what became known as the magic words: “vote for,” “elect,” “support,” “cast you ballot for,” “Smith for Congress,” “vote against,” “defeat,” and similar terms. Subsequent court decisions broadened the test somewhat to include statements such as those that “provide in effect an explicit directive” to vote for named candidates.²⁷

Communications that merely discuss public issues, even if their purpose is to influence elections, could not be regulated at all.

The ambiguities of the distinction between express and issue advocacy began to be exploited in the 1996 election in a major way. By the time of the 2000 election the proliferation of sham issue advocacy ads (together with soft money) had for all practical purposes completely subverted the purpose of the regulations. Here is a now notorious example of a sham issue ad (broadcast in the final weeks of 1996 state election in Montana):

The campaign reform act signed into law in March 2002 seeks to deal with these sham issue ads by adopting a broader concept than express advocacy. It defines an “electioneering communication” (which is subject to regulation) as a broadcast ad that clearly identifies a federal candidate within 30 days of a primary or 60 days of a general election and is targeted to his or her constituency (U.S. Congress 2002). Evidently uneasy about the constitutionality of this definition (or eager to invite a challenge), the legislators added this provision: If the primary definition is held “constitutionally insufficient,” then an “electioneering communication” should be taken to mean any ad that supports or opposes a candidate and can be plausibly interpreted only as an “exhortation” to vote for or against a specific candidate. This fall-back definition is not quite the express advocacy test again (that test is expressly repudiated in the legislation), but it is so close to current practice that it would probably not deal with the problem of sham issue ads.²⁸

The primary definition (identifying a candidate in the specified period) is much better than the current practice, but it still calls for interpretation and is likely to continue to generate disputes.²⁹

The practical problem has been to formulate a criterion that is strict enough to rule out sham issue ads but loose enough to permit robust debate about issues. The theoretical challenge, which has received less attention, is even greater. It can be brought out by asking: Why should we regulate any kind of advocacy during a campaign? After all, no one would propose trying to control candidate advocacy all the time (even though it certainly takes place months before the official campaign period). The problem is not how to distinguish different kinds of advocacy, but how to distinguish electoral from nonelectoral politics. Why are elections special? This challenge goes beyond the advocacy problem. We have to justify a distinction between electoral and ordinary politics to defend even minimal campaign reforms, such as financial disclosure. (Presumably we would not want to require everyone who takes part in ordinary political activity to reveal all of his or her sources of financial support.)

The usual answers to the challenge are not adequate. First, some try to deny that the regulations assume any fundamental distinction between campaigns and other political periods. The aim of most campaign finance regulation is to control political fundraising, which takes place continually. Ads are regulated during a campaign simply because that is when the money is spent. But this reply does not sufficiently recognize that the purpose of the fundraising, whenever it occurs, is for the campaign and, therefore, that the regulation still assumes that there is something special about the campaign period. Funds raised for other purposes (even funds raised during the campaign) are not subject to regulation.

Second, some argue that the regulations in question do not create a significant difference between campaigns and other political periods because they do not prohibit any citizen from engaging in candidate advocacy. “No speech is ‘banned’ under the new Act…. The only restrictions apply to the sources of funding, not to the speech itself” (Potter 2002). It is true that under the law individuals can still advertise for or against a candidate as much as they wish—or can afford (U.S. Congress 2002). They have to disclose only the funding above certain thresholds. Corporations and labor unions may not use their treasury to sponsor electioneering ads, but even they can continue to run ads in the middle of the campaign, provided they use only funds from their PACs (which must be funded voluntarily by individual employees, shareholders, or union members). National political parties can sponsor electioneering ads if they use “hard” money (which is raised within strict limits and is subject to disclosure).

Although these considerations may adequately defend the new law against the charge that it bans important forms of free speech, they still do not explain why the regulation of the sources of funds target only campaign-related activities. Even if no one's liberty is seriously impaired and the integrity of the democratic process is significantly enhanced, the fact remains that the campaign finance law still constrains election-related activity more than other kinds of political activity. It imposes limits on campaigns that are not tolerated in other parts of political life. We are still left with the question: Why single out election-related activity?

The third set of arguments emphasizes the critical importance of the discussion that takes place in the campaign. Elections are periods of “heightened value debate and political opinion formation” and, likewise, periods of heightened opportunity for citizens to transmit those values and opinions to their representatives (Baker 1998). Citizens pay more attention to political communications during this period, and the amount and variety of political debate are greater than at most other times. Although many citizens may make up their minds before the campaign, undecided voters are more susceptible to political advertising during the campaign. Finally, the stakes are higher because voters are choosing their leaders, not simply trying to influence particular policies or decisions.

The trouble with these arguments is that they are double-edged. The very importance of campaign discussion provides a reason for restraining it less. The more attentive and the more receptive citizens are, the more vigorous and freewheeling the debate should be. In the discussion of some issues (such as decisions to take military action or to cut taxes), the stakes are no less high between elections. Furthermore, privileging noncampaign over campaign debate could give political advocates “a perverse incentive to conceal their electoral preferences in order to avoid burdensome regulations”—amounting to “the closeting of electoral speech” (Sullivan 1998). Encouraging citizens and politicians to disguise their electoral positions as general policy positions is “inconsistent with the notion of robust, uninhibited, and wide-open debate about the policy bases for choosing among our representatives.”

Despite the difficulty of separating electoral from nonelectoral politics, the need for some distinction remains. Without it, we are faced with the dilemma of either abandoning any attempt to control the excessive influence of money in elections or imposing overly broad prohibitions that risk stifling debate in the wider political process.

Beyond campaign finance reform, many other sensible rules depend on the assumption of the distinctiveness of the electoral process; we reasonably accept many regulations in the electoral realm that would be intolerable in the general domain of public discourse. There are limits on what voters are permitted to express at the ballot box, as well as requirements to disclose the identity of political speakers. There are content-based regulations of electoral speech, ranging from constraints on electioneering near polling places to selective bans on contributions from some speakers such as corporations (Schauer and Pildes 1999).

The suggestion that the stakes are higher at election time, while not correct in the form that it is usually presented, points in the right direction. What raises the stakes is not the relative importance of the questions that elections decide, but the relative finality of the result that elections produce. Finality, the third temporal property of elections, provides a more fundamental basis for treating electoral politics differently from other kinds of politics.³⁰

Finality is the flip side of periodicity. Elections occur at intervals but once an election takes place its results are final for the intervening period.

Electoral finality has two aspects. First, unlike ordinary political activity, elections (and the campaigns that precede them) come to a definite conclusion at a foreseeable time. An election takes place at a particular moment, which until the next election marks an end to the process of deciding who will hold office. The results of elections are not reversible until the next election. Electoral accountability is not continuous so that those elected can govern effectively for a period without the risk of immediate removal from office.³¹

Second, unlike ordinary political activity outside government, elections result in decisions that are binding on all citizens. Although noncampaign activity also often affects the outcome of elections, its purposes are broader. It is not directed exclusively at influencing a decision that, without any further political discussion or representational mediation, immediately obligates all citizens. The winner of an election gains the authority to act for all citizens.

These two aspects of finality together justify subjecting electoral politics to more stringent standards than those that govern the rest of politics outside government. Because of their finality, elections and the campaigns leading up to them should be considered more a part of government than a part of politics that influences government. Consequently, all citizens have not only an interest in the integrity of the process, but also a claim to participate in setting the standards that govern it. The standards that control the conduct of elections are therefore more appropriately determined by collective decision than by individual discretion. Citizens acting together should set the standards for the process that determines who will make and execute the laws by which they all are to be bound.

Finality thus provides in principle a normative basis for adopting stricter regulations for campaigns than for other political activity. But it does not supply a criterion for distinguishing between the two kinds in practice.³²

Like many other normative distinctions, such as that between private and public, the principled difference between electoral and other kinds of politics can be more readily established in theory than in practice. That is why any practical criteria—both those that specify the length of the campaign and those that identify campaign-related content—are arbitrary in the sense that there is no compelling reason to draw the line at the point they designate than at any of an indefinite number of other points. Why 60 days before an election rather than 90? Why not shorten the campaign period itself, as many European countries have done? Why is “referring” to a candidate necessary or sufficient test? To answer these and similar questions that arise in devising practical criteria, we must take into account a variety of considerations such as administrative convenience, enforceability and transparency. Yet the criteria need not be objectionably arbitrary if they represent a good faith effort to capture, within the limits of these practical considerations, the principled difference between electoral and ordinary politics.