The Progressive Report

The Constitution provides for a federalist system of self-government with judicially enforceable constraints on the exercise of legislative and executive power to guarantee a core set of fundamental personal rights. The Constitution could be amended, as it has multiple times, to provide greater protection for voting rights and democratic participation. But on the assumption that for the foreseeable future there will be no new constitutional amendments, the focus here is to determine which legislative or other policy changes would best serve the goal of securing Article IV, Section 4’s guarantee of a “republican form of government,” where citizens choose who will hold power on their behalf rather than tyrants ruling without regard to popular will.

The organized effort to repudiate the result of the 2020 presidential election, which culminated in the insurrection at the Capitol on January 6, 2021, revealed the Republic’s vulnerability to a partisan effort to retain power regardless of the popular verdict. The election results had been legitimately determined through the procedures established by law precisely for the purpose of enabling the citizenry to render judgment on whether the incumbent should receive another term. Of course, because the idea that partisans would deny the people the electoral choice they actually had made is so anathema to the most elementary precepts of constitutional democracy, these partisan usurpers had to pretend that they were the ones defending democracy against a malevolent usurpation of the people’s will. Hence, the fabrication and incessant repetition of the “Big Lie” that the election was stolen, in order to provide a justification for repudiating the result. But it was all an Orwellian effort to eradicate truth, brazenly substituting a concocted alternative, that was too much for even some of the incumbent’s top advisers— like the Attorney General at the time— to stomach.

If the Republic is to be protected against a repeated attempt to perpetrate this kind of plot, it will need to redress four deficiencies. First, and most obvious, is the urgent task of eliminating weaknesses in the procedures for determining and declaring electoral result0in other words, the basic infrastructure of counting votes and certifying winners. Second, and at least equally important, is changing electoral procedures to reduce the risk of elected officials predisposed once in office to retain power regardless of new electoral choices the voters make, which could be to replace the incumbent officeholder with someone offering a different vision. Third, and most challenging but worth pursuing, is changing the rules to make unlawful conspiracies to negate electoral outcomes through deliberate falsehoods. Fourth, looking long-term but perhaps most significantly for the ongoing sustainability of the Republic, the nation needs a reinvigoration of civics education that encompasses not only the history and norms of self-government, but also the mathematics of democracy. (George Szpiro’s book Numbers Rule, published by Princeton University Press in 2010, is a very accessible introduction to the historical development of what its own subtitle describes as the “mathematics of democracy” from early days to contemporary times.) Electing officeholders to represent citizens based on the will of the voters requires collecting the diverse preferences among the voters into a coherent whole, and this additive process is more complicated than many citizens understand because their civics education has not adequately schooled them on the topic. If American democracy is to be truly safeguarded from threats to undermine it, citizens will need to have a better understanding of the numerical underpinnings of how voting can either succeed or fail to secure self-government.

The analysis and recommendations here are not intended to be exclusive. See, for example, Richard L. Hasen, Identifying and Minimizing the Risk of Election Subversion and Stolen Elections in the United States, 135 HARV. L. REV. 265 (2022), on the risks of election subversion and how best to combat them. That valuable contribution calls for improved chain-of-custody requirements, risk-limiting audits, and related measures to enhance vote-counting integrity. This paper embraces those proposals, while endeavoring to go further in addressing structural elements of the electoral system that make American democracy vulnerable to subversion because of increased conditions of partisan polarization, including especially the rise of authoritarian populism.

There is no higher imperative than reaffirmation of the principle that ballots be counted honestly and accurately according to what objective evidence (including the ballots themselves) transparently demonstrates, using procedures and rules established in advance of the election; then, as a key corollary to this basic principle, whatever the result of the count may be, defeated candidates and their political allies accept the result as the valid verdict of the electorate.

Protecting Presidential Elections

In the context of presidential elections, reinvigoration of this principle requires revising the Electoral Count Act of 1887. This statute provides the procedures for the special joint session of Congress that, under the Twelfth Amendment of the Constitution, officially declares the result of a presidential election. As tragically became apparent on January 6, 2021, some senators and representatives were willing to abuse the existing procedures in an effort to repudiate the result of the 2020 election, despite judicial decisions rejecting similar efforts in the various states whose vote tallies had been challenged. Although this abuse of the process caused violence only in 2021, the same kind of abuse was committed by the opposing political party in the aftermath of the 2000, 2004, and 2016 elections. In all three of those cases, members of Congress objected to counting electoral votes cast solely because they believed that the wrong electors had been appointed in various states, although Congress has no power to second-guess

a state’s appointment of its electors pursuant to its own laws, but only to make sure that the electoral votes it receives were cast by individuals who were indeed appointed electors pursuant to state law (as was undoubtedly the case in all these instances of improper objections after the 2000, 2004, 2016, and 2020 elections).

The best way to revise the Electoral Count Act, to prevent the same kind of abuse from occurring again, is to make unambiguously clear that the task of joint session of Congress— in the event of any dispute over a state’s electoral votes— is to count the votes cast by electors whose appointment is based on state law adopted in advance of their appointment, as confirmed by the courts to the extent necessary. Under Article II of the Constitution, state legislatures have the authority to choose the “manner” of appointing the electors in each state, and all state legislatures have chosen (by ordinary means of legislation, which includes a role for a gubernatorial veto) to hold a popular vote as their “manner” of appointment. Presumably, state legislatures will continue to use popular votes as their chosen means of appointing electors and, having done so, the courts will continue to require that these popular elections be conducted according to the laws that state legislatures have enacted for this purpose. (Whether, or the extent to which, state constitutions can constrain state legislatures in the exercise of this power under Article II of the federal Constitution is an issue of some controversy and potentially subject to U.S. Supreme Court adjudication.

But however the Court decides that issue, it remains true that electors must be appointed pursuant to law, and therefore state officials are not permitted to lawlessly repudiate the rules that the state itself has established in advance for their appointment.) Thus, as long as Congress requires itself to count the electoral votes cast by electors whose appointment the courts have confirmed as based on the popular votes mandated by state laws, there will be no available avenue to repudiate the result of the election based on the judicially confirmed counting of the popular vote in each state.

While the exact details of the procedures that states have created for adjudicating disputes over counting votes in an election differ from state to state, it is undoubtedly true that— as both 2000 and 2020 show— campaigns will go to court, using whatever judicial procedures arguably are available, in an effort to secure a result in their favor. Thus, whatever is the ultimate resolution of this litigation, whether in state or federal court, that resolution will be the rule-of-law answer as to which candidate won the popular vote in that state. Reform of the Electoral Count Act should be guided by the overriding goal of requiring that rule-of-law answer to ultimately prevail in the joint session of Congress held pursuant to the Twelfth Amendment.

No senator or representative, nor any combination of them, will be able to repudiate the judicially confirmed result of the popular vote— as long as in a revised ECA Congress will have unambiguously foreclosed the power of its joint session to invalidate these judicially confirmed results. Nor will it be possible for any rogue official within state government, not the state’s governor nor even the state’s entire legislature, to nullify the result of the state’s popular vote. Going rogue in this way would be in opposition to what the courts declare the result of the popular vote in the state to be. Thus, if Congress is required to count what the courts declare to be the result in each state, then any attempt by a governor or other state official to contravene what the courts declare would be null and void upon arrival in Congress.

A state legislature might attempt to take back using a popular vote to appoint the state’s electors; but under Article II, Congress— not the state legislature— has the power to set the “time” for appointing the state’s electors. Congress has exercised this power by setting Election Day in November as when electors get appointed. Thus, once electors have been appointed by the date designated by Congress, using a popular vote as previously chosen by the state legislature, the state legislature itself no longer can undo the appointment of the state’s electors for that year’s presidential election. It can switch to a different “manner” of appointment for the next presidential election, four years later. But the legislature itself is stuck with its previously chosen popular vote— and the result of that popular vote according to its own laws for counting, and (when necessary) recounting, that vote. If a state legislative attempts to redo the appointment of electors just because it doesn’t like the outcome of the popular vote, that attempted redo is null and void, and the revised Electoral Count Act should make this constitutional point crystal-clear. (Nor can a state legislature, after the popular vote has already been held, claim for itself a role in the process of remedying alleged defects, like fraud or irregularities in the vote-counting process, unless beforehand the legislature by statute had given itself that role.)

To the extent there is any confusion on this point, especially confusion generated by invocation of the so-called “independent state legislature” doctrine, it is imperative to dispel it as part of Electoral Count Act reform. Whatever the appropriate contours of the controversial “independent state legislature” doctrine may be— for example, to what extent does the Article II power of state legislatures to choose the “manner” of appointing presidential electors constrain the state’s own judiciary from engaging in flexible interpretation of state statutes enacted pursuant to this power?— that doctrine has no capacity to negate the separate power of Congress, also explicitly delineated in the same Article II, to set the “time” for appointing a state’s electors for any specific quadrennial election. Thus, however plenary and unconstrained the state legislature’s power to dictate the rules for appointing electors in advance of their appointment, once they have been appointed at the “time” specified by Congress, their appointment cannot be undone, not even by the same state legislature that chose the method of their appointment in advance.

Any statutory confirmation of this basic constitutional point as part of a revised Electoral Count Act would undoubtedly be within Congress’s legislative power under the Necessary and Proper Clause, as it would simply implement with greater precision what the Constitution itself already provides. (While some have argued that the current Electoral Count Act exceeds the power of Congress, a suitably tailored revision of the Act should be seen to fall within the Necessary and Proper power.)

Protecting Congressional Elections

It is not only a presidential election that is at risk of being nullified. The same thing could happen in a congressional election. Congress, however, has the power to prevent that from happening.

Article I, Section 5 of the Constitution provides: “Each House shall be the judge of the elections, returns and qualifications of its own members.” This provision, with historical roots in the British battle between Parliament and Crown, is a double-edged sword. On the one hand, it means that as long as the ballots in a Senate or House election have been preserved in a way that has kept their integrity intact, they can be recounted by a committee of the Senate or House, as the case may be, in order to assure that the tally is honest and accurate. On the other hand, it also means that the party that controls the congressional chamber in question can abuse that power to manipulate the counting of ballots to give the party an extra seat that it did not actually win.

This power has been abused in the past, including at times when partisanship in Congress was not as polarized and vehemently contentious as it currently is. Therefore, one necessarily worries about the possibility of abuse again, whether in the aftermath of the 2022 midterms or in another future election. Both the Senate and the House would be well-advised to consider in advance of the midterms what additional steps they might take to minimize the risk of such abuse.

The problem of partisanship in congressional ballot-counting presents an especially acute danger to democracy

in the particular context of the single seat in question tipping the balance of which party holds the majority of seats in the chamber. It is bad enough when the party already in power gives itself an extra seat beyond what it actually earned from the voters. It presents an entirely different kind of peril when the chamber itself cannot be organized at the beginning of the new session into majority and minority parties, with committee assignments and leadership positions determined accordingly, unless and until the winner of the decisive seat is determined.

In the rare cases when this has occurred before, deadlock has set in, and the chamber has been unable to function until the logjam is broken by some precipitous means. (In one instance, John Quincy Adams— the only former president to serve in Congress— had to assert an extra-constitutional authority to preside over the House as its chair, making consequential parliamentary rulings as necessary, until the electoral dispute was settled sufficient to permit a rollcall vote on the installation of the next Speaker.) With the Senate currently at 50-50, it is not at all unlikely that the question of whether the Democrats retain majority control (by virtue of Vice President Kamala Harris being the Senate’s tiebreaking president), or instead switches to a 51-49 Republican majority, could turn on the outcome of a close and deeply disputed election in one of the midterm’s races.

While the results of the two Georgia runoffs on January 5, 2021 (which similarly flipped control of the Senate from one party to the other) were not disputed, those runoffs were overshadowed by the rioting at the Capitol on the next day. Moreover, the culture of electoral denialism— which rejects the validity of a vote tally that shows one’s own side as having lost— has taken root in the aftermath of January 6, 2021, thereby posing a greater risk of a consequential midterm Senate election becoming seriously disputed. With all the controversy over the changes to Georgia’s election laws since January 6, 2021, it is not hard to imagine the state’s Senate election this year becoming mired in litigation, even though the state’s 2020 Senate races did not.

The same point applies to Arizona’s current Senate election, or indeed any of the ones being fought in “battleground” states like Pennsylvania or Wisconsin. If a dispute over one (or more) of these races reaches the Senate itself in January of 2023, and especially if resolution of the dispute will determine which party holds the majority in the Senate, then the two party caucuses— and especially the two leaders, presumably still Senators Schumer and McConnell— will need to settle upon a mutually agreeable plan for handling the dispute. In any circumstance, that would be an exceedingly difficult challenge. In the current toxic political climate, it would put the Republic in even more precarious and potentially ungovernable circumstances. While there is likely little appetite for preparing in advance for a circumstance that may not materialize, it would be easier to establish a set of shared procedures and principles for handling this kind of situation before it actually arises. Assuming that kind of advanced planning does not occur, then one must hope that Senate leaders can summon the kind of bipartisan leadership that will be essential in the heat of the moment. (The courts likely will be of no assistance if and when this kind of dispute reaches the Senate precisely because the Constitution so explicitly designates the Senate itself as the forum for settling the matter; therefore, under the political question doctrine, the judiciary will see its role as keeping itself out of what the Constitution has allocated to a different body.)

Protecting Gubernatorial and Other State or Local Elections

Some of the most important elections in the United States are not for federal office, but for state office instead. This is true now as much as it has ever been. Think of the governorships of major battleground states, like Michigan or any of the others previously mentioned: Arizona, Georgia, Pennsylvania, or Wisconsin. Or secretary of state races in these and other states.

Given federalism, there is only a limited role that federal law can play in protecting the integrity of counting votes in state and local elections. One must hope that state courts will do their part in enforcing state laws to protect to honest and accurate counting of ballots in these elections. More Americans should know about the heroism of state judges that have done so in the past, including for example the members of the Wisconsin Supreme Court who in 1857 refused to permit the state’s governor to manipulate the counting of ballots to give himself another term that he did not win from the voters.

Until recently, one might have even thought that there was no role whatsoever for federal courts to play in this context and that it was necessary to rely entirely on state courts. To be sure, “the great dissenter” John Marshall Harlan lamented in a case of blatant ballot-box stuffing in a Kentucky gubernatorial election, Taylor v. Beckham (1900), that the Fourteenth Amendment should be construed to constrain this kind of deliberate subversion of democracy. But being the great dissenter, his sole voice for this view did not prevail at the time.

Bush v. Gore (2000), along with a myriad of lower-court cases in the same vein, however, open up the possibility of federal courts preventing flagrant manipulation of vote-counting in gubernatorial and other state or local elections. The same Equal Protection Clause requires that all ballots cast in the same election to be counted equally, which is the principle that the Court— indeed seven of the nine justices— upheld in Bush v. Gore (with two of the seven dissenting on the appropriate remedy in the context of that contentious case), applies just as much to a gubernato- rial or any other election as it does to the popular vote that a state employs for appointing its presidential electors. Thus, if in any of this year’s gubernatorial or other state-level elections, like for secretary of state, there is any attempt to negate the true choice of the voters and to impose a dishonest outcome instead, in principle it should be possible for the genuinely winning candidate to go to federal court and, pursuant to the Fourteenth Amendment, secure an injunction against this denial of equal treatment.

The risk of negating election results, just because politicians in power dislike the verdict of the voters, requires bolstering the rules and procedures for counting ballots and declaring results, as described above. But it is not enough to bolster these rules and procedures. Ultimately, it matters who is in office with the final authority to determine an election’s winner. It’s the members of Congress sitting in the joint session required by the Twelfth Amendment who will control whether or not they abide by the requirements of the Electoral Count Act in settling any dispute over the winner of a presidential election. Likewise, it’s the senators or representatives already in office who will decide whether they let partisanship affect their decisions on who will prevail in any dispute over a seat in their chamber.

Thus, it makes a difference whether members of Congress themselves are faithful to the principle that the will of the electorate, as determined by an honest and accurate count of ballots cast by valid voters, should be respected— or instead whether members of Congress are willing to repudiate the will of the electorate solely for the sake of partisan political power. Obviously, if the voters themselves want to elect members of Congress who are willing to overturn subsequent election results just so their party holds power, then there is nothing that the electoral system itself can do to stop this self-subversion of democracy. The electoral system needs to put into office the candidates whom the majority of voters want to win, and if these candidates in turn seek to destroy the system itself, the protection must come from collateral constitutional constraints enforced by courts (like the Bill of Rights), rather than from the electoral process itself.

But the current electoral process unnecessarily compounds this kind of danger. Currently, election structures tend to facilitate the election of extremist (tear-the-system-down) candidates, rather than more consensus-oriented (play-by-the-rules) candidates. At least in the current political climate, those extremist candidates are also more likely to be election deniers. In other words, the present danger takes a compound form: the existing electoral procedure enables candidates to win seats in Congress who are not truly the preference of a majority of the electorate,

and these minority-faction winners are those who tend to be willing to repudiate future election results that reflect what the majority genuinely wants. Thus, electing members of Congress who will tend to respect election results requires reform of election procedures, so that they are designed to elect the candidates that the majority of voters most want to win. This requires changing the rules for both primary and general elections and how the two stages of the overall process are related to each other.

Madison and Condorcet: Factions and Elections

The challenge for republican government is to translate the multiplicity of preferences that exist within the citizenry into public policy choices that the government implements. The Madisonian philosophy that underlies the federal Constitution is to use separation of powers to prevent electoral choices from turning into either a tyranny of the majority or a tyranny of the minority. But the Madisonian philosophy was premised on the expectation that the complex constitutional architecture of separated and federated powers would keep the multiplicity of interests within society from coalescing into organized and ongoing political parties that would manage to capture control of government and contravene the general public interest. As we all know, however, this Madisonian expectation was almost immediately dashed, when intense party competition formed between the Federalists and the Jeffersonians; and by 1792 James Madison himself was defending his allegiance to Jefferson’s party in opposition to the Federalists. Ever since, the United States has been endeavoring to make its anti-party Constitution function effectively given incessant and intense partisanship.

Around the same time that Madison was developing his constitutional philosophy, across the Atlantic in France the Marquis de Condorcet was pursuing his own theories of how best to combine diverse preferences into a single social choice. Condorcet’s key insight was that, whenever there are three or more candidates (or options) from which to choose, the best choice for society as a whole is the one preferred by a majority of voters when compared individually against each of the other alternatives. It isn’t always the case that there is a “Condorcet winner” in this way; sometimes, as Condorcet himself showed, the preferences of individual voters, although each perfectly rational on its own, inevitably will create an indeterminate cycle— A beats B, B beats C, and C beats A (rather like the game of rock, paper, scissors)— when these preferences are attempted to be aggregated together in a group decision. But when a Condorcet winner does exist, it is best for the collective choice to be the Condorcet winner because no other option could be more preferred by the group as a whole. (If challenged by another candidate, the Condorcet winner necessarily will prevail— because more voters prefer the Condorcet candidate when compared to any other alternative.)

Elections in the United States are not designed to identify the Condorcet winner even when one exists. While each jurisdiction relies on its own electoral design, the prevalent approach employs party primaries followed by a general election where only a plurality of votes is required to win. As a result, a minority of votes often is able to block the election of the Condorcet winner. Consider the current divide in the Republican party between its more moderate and traditional wing, on the one hand, and its more polarizing and populist wing, on the other. The latter may have become dominant in many states, causing its candidates to win party primaries against more moderate and traditional candidates. But in many of these states, the more moderate and traditional Republican would be the Condorcet winner in the general election: this candidate, when compared against the Republican primary winner, or against the Democratic primary winner, or indeed against any other candidate, would be preferred by a majority of the general election voters. (The same point can be true on the Democratic side of the divide: a more progressive Democrat might win the party’s primary, but depending on the particular state, a more centrist Democrat who loses the party’s primary would actually be the Condorcet winner in the general election.)

This effect is problematic even without the threat of election denial. Yet in the current political climate, eliminating the Condorcet winner in a party primary also contributes to the risk that the eventual winner of the general election, once in Congress, may be predisposed to repudiate future election results. In a party primary, an election denialist may defeat a Condorcet winner (because the primary’s electorate is a narrow slice compared to the entirety of the general electorate). If the election denialist were to lose in November to the opposition party’s nominee, the risk to democracy would diminish. But, depending on the overall political profile of the state (or district, in the case of a House election), the voters may prefer the election denialist to opposing party’s nominee, and this is true even though these November voters would have preferred the Condorcet winner to either the election denialist or the opposing party’s nominee. In this way, the member of Congress may end up being an election denialist who was less preferred by the entire electorate than another candidate who was not an election denialist. If this kind of result is replicated in race after race, it is precisely the circumstance that will endanger the future of democracy, and it will be a consequence of not electing the candidate most preferred by voters overall.

Thus, a promising, and perhaps necessary, way to protect democracy from the risk of election subversion is to alter the electoral system itself so that it elects Condorcet winners. This can be done with a version of Ranked Choice Voting that can be called “round-robin voting” because it resembles a round-robin tournament in which each competitor is matched one-on-one against every other competitor. By enabling voters to rank their preferences among all candidates running for the same office, regardless of party affiliation, the electoral process can conduct the direct comparisons between each pair of candidates necessary to identify the Condorcet winner (assuming one exists). For each pair of candidates, one will be ranked higher than the other on each of the ballots cast in the election (and, when an individual voter does not rank all the candidates listed on the ballot, a ranked candidate can be deemed preferred by that voter to any unranked candidate). In this way, ranked ballots can show whether one candidate is preferred by more voters when compared directly with every other candidate. If so, that candidate— the Condorcet winner— can be elected.

If Congress were populated by Condorcet winners, the risk of Congress repudiating future election results because members did not like the electoral choices voters made would be much lower than currently exists, with many members of Congress having secured their seats despite not being Condorcet winners and, more disconcertingly, making abundantly clear their allegiance to election denialism and their willingness to repudiate valid results solely for the sake of partisan power.

Proportional Representation (or “self-districting”) as a Way to End Gerrymandering

Using ranked-choice ballots to elect Condorcet winners, if employed to fill Senate seats, would tend to produce senators more oriented towards bolstering democracy rather than undermining it. But this reform, by itself, would not solve the problem of gerrymandering, which afflicts the House of Representatives (but not the Senate, given that senators are elected statewide). House districts are so misshapen that in some districts even the Condorcet winner may be a candidate with a confessed commitment to “the Big Lie” and the willingness to reject valid election results. Gerrymandering, moreover, creates a political culture in which it is standard practice for politicians to manipulate elections in order to maximize power for one’s own party— no matter what the voters actually want. No wonder so many members of the House were willing to nullify the 2020 presidential election even though all the courts had upheld the outcome as the true choice of the voters: who cares what the voters genuinely decided; the point is to retain power for one’s party, including by manipulating the rules and procedures if necessary. A party that gerrymanders its way to retain power regardless of the electorate’s preferences finds itself willing to abuse the Electoral Count Act for the same purpose— or at least so it seems to some observers attempting to account for the deterioration of democratic norms.

Thus, it is imperative that Congress find a way to adopt bipartisan reform that eliminates the evil of gerrymandering in congressional elections. At the state level, there have been efforts to adopt measures, including the creation of new redistricting commissions, to do this. While some of these moves have had some success, others have not. Ohio, for example, has arguably become a travesty as the commission empowered to curtail gerrymandering has simply insisted on perpetuating the practice, despite four decisions from the state supreme court ordering it to comply with anti-gerrymandering standards newly enacted into state law. (Reflecting a design flaw in the commission’s composition, its members are elected officials— including the state’s governor and secretary of state— and they have acted as if they are compulsively addicted to the drug of partisanship, unwilling even to obey judicial decrees authored by the chief justice who despite being from the same political party is endeavoring to interpret the law in accordance with a nonpartisan conception of the public interest.)

The situation has gotten to the point where it is worth contemplating elimination of districts completely, at least as traditionally conceived, in order to prevent districts being gerrymandered. (Filling legislative seats on a statewide basis would require proportional representation in order to avoid minority vote dilution in violation of the Voting Rights Act.) The Constitution does not require single-member districts, or even any geographic districts, to elect members to the federal House of Representatives. Congress could repeal its statute imposing this requirement— on condition that states experiment with alternative proportional representation systems, all of which would eliminate gerrymandering. (As required by the Voting Rights Act, a state would not be permitted to abandon traditional single- member, geographic-based districts unless it did so pursuant to some form of proportional representation.)

One such system, perhaps easiest for Americans unfamiliar with proportional representation to understand and accept, would be for a process of “self-districting” in which voters themselves would choose the constituency in which they belonged. These constituencies could be based on geography. For example, in Ohio voters could sign up if they wished to be included in a Cleveland-based constituency. But constituencies could also be based on other attributes. Baby boomers, Gen-X-ers, or other age groups could organize constituencies. So too could Farmer or Labor groups, or Voters of Color, if they wished.

In this system, the first stage of the electoral process would be for voters to choose which constituency to join, from a list of all that have been organized by their fellow citizens. A constituency would need to attract enough voters to earn one or more House seats. For example, Ohio is entitled to 15 seats in the federal House of Representatives. Under this system, at least one-fifteenth of the electorate would need to join one of the listed constituencies in order for that constituency to receive one of Ohio’s House seats.

If a constituency received enough votes for two or more seats, there would need to be a method for allocating the seats among all the constituency’s voters. The simplest way to do this would be to let the constituency itself do this according to whatever criteria the constituency established in advance, so that voters were aware of the criteria before joining the constituency. For example, a constituency could use geography to divide itself into several districts. Or a constituency could divide itself into different age groups, with one seat for each age group. Or a constituency could assign its voters randomly to each of its several seats.

The second stage of the process would be for candidates to run in each of the constituency’s seats. Under this system, the second stage would tend to look more like party primaries currently. In a constituency filled entirely of self-selected Clevelanders, or Farmers, the candidates competing to win that specific House seat would be vying for the supports of voters who voluntarily grouped themselves according to the attribute they considered most politically significant. (Because in choosing constituencies voters would not divide evenly into the state’s number of House seats, there would need to be at least one remainder seat for all those voters whose choice of constituency did not attract enough votes to receive a seat of their own. For these voters, the second stage of the electoral process would resemble a more traditional general election, as candidates of varying ideological perspectives could compete to win the seat that represents this amalgamated and heterogenous constituency. Also, the size of the remainder district would need to comply with the constitutional principle of equally populated districts, as articulated in Westberry v. Sanders (1964), assuming the current Court continues to adhere to that longstanding precedent; compliance with Westberry could be handled in different ways, including giving voters the option of joining either an “oversubscribed” district of their choice or an “undersubscribed” remainder district, depending on the voter’s own prioritization of population equality or composition of the constituency.)

This system of “self-districting” bears some resemblance to the various forms of “party-list” proportional representation that exist in Europe. Of course, any of those could be imported directly in the United States. But structured as self-districting, where voters choose the constituency from which they will elect a candidate to represent them in the House, enables political parties to compete both in the formation of the constituencies and, if they wish, also in nominating candidates to run in the constituencies. This kind of constituency-based form of electoral competition may seem less foreign to American voters than European-style proportional representation. Either way, however, would eliminate the evil of gerrymandering.

Moving to proportional representation, moreover, would solve the problem of drawing districts to comply with the Voting Rights Act. In a case from Alabama, the Supreme Court is poised to make it impossible to draw districts to protect minority voters from underrepresentation through vote dilution. A system in which minority voters could self-select their own constituency would guarantee their ability to elect a portion of seats roughly equal to their share of the state’s total electorate. The concerns that would cause the Court to invalidate race-based districts drawn by the government would not apply to self-organized constituencies voluntarily chosen by voters themselves.

To be sure, in a proportional representation system, there is the risk of an anti-democratic political party winning a percentage of legislative seats equivalent to its share of the total electorate’s votes. But as long as this anti- democratic party remains in the minority, it cannot control the legislature as a whole (although there is the risk that forming a governing coalition will require inclusion of an extremist party). What is especially dangerous about America’s gerrymandered House of Representatives, is through the combination of distorted districts and partisan primaries, an anti-democratic faction within one of the two major political parties can leverage control over that party and then win the distorted districts, so that it captures a far larger percentage of seats in the House than its share of the electorate as a whole. This electoral leveraging then enables a minority faction to exert control over the legislative chamber and potentially perpetuate that control contrary to the will of the majority within the citizenry itself.

Eliminating Plurality Winners in Primaries

The two major reforms described above— proportional representation for House seats and Condorcet-based “round-robin” voting for the Senate— would go a long way in protecting the future of Madisonian republicanism from the threat of an authoritarian faction led by a populist demagogue. But both reforms would be ambitious even though they are fully consistent with the Madisonian premises of the federal Constitution. A more modest reform with the same aim would be to reform conventional party primaries by eliminating plurality winners.

Plurality winners in primary elections contradict the idea of a two-party system, in which each of the two parties chooses a nominee who represents the dominant view of that party. The plurality winner of a multi-candidate party primary might actually be a “Condorcet loser”— meaning that a majority of voters in the primary would have preferred any other candidate when compared one-on-one with the plurality winner. A recent GOP primary for an Ohio seat in the U.S. House of Representatives illustrates the problem. An especially extreme candidate won the primary with only 36% of the vote because two more mainstream candidates split 60%— one getting 31% and the other getting 29%. It is likely that either of the more mainstream candidates would have beaten the plurality winner if the primary had been a two-candidate, rather than multi-candidate, race. Thus, it hardly can be said that the plurality winner is the preferred choice of the party as a whole or a majority of its voters in the district. (Not every three-way split in a partisan primary will fit this description, but the fact that Nebraska’s GOP gubernatorial primary came very close to repeating what happened just a week earlier in Ohio suggests that the risk of this problem is not trivial.)

What is worse, the “Condorcet loser” in a party primary may go to win the general election. If so, it doesn’t mean that general election voters preferred this plurality-winning “Condorcet loser” to the other candidates in the party primary. On the contrary, the voters in November also may prefer either of the more mainstream candidates who split the vote in opposition to the extremist plurality winner of the primary. But the November voters may dislike the other major party’s nominee even more. In this way, the extremist “Condorcet loser” of the primary may end up seated in Congress, even though both a majority of the primary voters and a majority of the general election voters would have preferred any other of that party’s primary candidates to the primary’s plurality winner. This example thus illustrates how permitting plurality winners in primary elections can undermine democracy, by seating extremists in Congress who do not represent the true preferences of either the party or the electorate as a whole.

The Constitution imposes no obstacle to state or federal legislation that would eliminate plurality winners in primary elections. To be sure, the First Amendment as interpreted by the U.S. Supreme Court in cases such as California Democratic Party v. Jones (2000), imposes some constraints on legislation that attempts to regulate party primaries. But the First Amendment concern in these cases is that the government not dictate to a political party, as an association of citizens, the procedures by which the party endorses a candidate to be its nominee in an election. At the same time, the Court has recognized that the state has valid interests in regulating its general election ballot, including which candidates are entitled to appear on the ballot along with a designated party label. Thus, if a state law provided that a candidate could not appear on the general election ballot identified with the label of a particular political party unless that candidate won a majority of votes, either in a primary election or some other electoral procedure (like a party convention), this kind of law should survive First Amendment scrutiny. The political party would be free to use whatever procedure it wished in order to make the endorsement of its nominee, but that candidate could not have the party label on the general election ballot unless the party’s procedure satisfied the state’s majority, not plurality, requirement.

In Maine Republican Party v. Dunlap (D.Me. 2018), the court upheld the requirement that political parties used Ranked Choice Voting as a means to achieve majority, rather than plurality, winners in their primary elections. Distinguishing California Democratic Party v. Jones among other precedents, the court held that the state’s interest in having candidates achieve an adequate level of support in order to earn a position on the state’s general election ballot justified the imposition of Ranked-Choice Voting in the party primary. The court also observed that because the state was entitled to require use of a primary election as the means for securing a spot on the general election ballot, the state could also select the procedure for calculating the primary’s winner based on the ballots cast.

Congress, too, could impose this kind of requirement for congressional elections, pursuant to its power under Article I, Section 4 to regulate the time, place, and manner of these elections. But for reasons of federalism, Congress might wish to adopt a more limited requirement: that states require a majority rather than plurality of votes to win either the primary or the general election. It is the double-plurality nature of the current system that is so dangerous to democracy. If a majority is required to win the general election, then a plurality-winning “Condorcet loser” in a primary could not prevail in the general election without demonstrating majority support from the electorate as a whole, even if a third-party or independent candidate attempted to present an alternative to both this “Condorcet loser” nominee and the nominee of the opposing major party. Likewise, as long as a majority is required to win a party primary, the winner of the general election is likely to be the majority’s preference between the two major- party candidates given the prevailing two-party nature of the nation’s electoral system. Thus, a new federal law that would eliminate just the double-plurality feature of the existing system would be a modest adjustment that would significantly help reduce the risk of extremist winners in Congress.

Richard Hofstadter famously described McCarthyism as an episode of the “paranoid style in American politics,” the implication being that McCarthyism was not an altogether new phenomenon. Today’s “Big Lie”— the belief that the 2020 election was stolen from Trump and that therefore Biden illegitimately occupies the Oval Office— can be seen as the most recent recurrence of the same intermittent phenomenon. Both McCarthyism and the Big Lie involve charismatic demagogues able to convince a large portion of the U.S. populace to embrace a fabricated falsehood as if it were a demonstrable fact.

But there are two features of the Big Lie that make it something entirely new and thus significantly different from McCarthyism or any previous version of this paranoid style. First, while the fabricated falsehood in McCarthyism concerned Communists in the government— the “Red Scare”— the fabricated falsehood at the heart of the Big Lie concerns the result of a presidential election. While the former is bad, the latter may be worse because it undermines the exercise of self-government through the electoral process. No previous episode of the paranoid style involved the counting of ballots and an unwillingness to accept what recounts plainly demonstrated to be an accurate tally of authentic votes cast by valid voters. If this new form of “electoral McCarthyism” persists in convincing a large portion of the electorate that vote tabulation is inherently unreliable, then representative government by means of democratic elections may no longer be feasible.

The second new feature of the Big Lie, in comparison to McCarthyism and earlier examples of the paranoid style, is the dissemination of intentional disinformation by the hyper-transmissible means of internet-based social media platforms. In particular, the algorithms that prioritize inflammatory content spread deliberate lies— including lies about election results— with an unparalleled efficiency, far beyond the capacity of previous methods of communication, whether word-of-mouth, pamphleteering, telephone or even talk radio. For democracy to survive, it must be protected against destruction by disinformation. Yet there is understandable fear that any effort at this protection will violate the First Amendment and bring about the demise of democracy that it’s trying to avoid.

The answer is to sever the current connection between the algorithms and immunity from liability for culpable content: the social media platforms can keep their algorithms, but then they lose their immunity; or they can keep their immunity, but then they lose their algorithms. If we are going to safeguard our democracy, however, they can’t keep both—which is what federal law in a provision known as “Section 230” now provides.

The Need for a Well-crafted Congressional Statute Criminalizing Deliberate Electoral Lies

“[T]he use of a known lie as a tool,” Justice Brennan said for the Court in Garrison v. Louisiana (1964), is “at odds with the premises of democratic government.” This passage comes in a case where the Court applied the same constitutional standard adopted in New York Times v. Sullivan (1964)— knowing or reckless falsity— to criminal as well as civil defamation: in the Court’s own words, “the knowingly false statement and the false statement made with reckless disregard of the truth, do not enjoy constitutional protection.” The Court rejected the proposition that just because the speech in question was political in nature, concerning the conduct of government officials, it therefore was absolutely immune from criminal prosecution. In his opinion for the Court, Justice Brennan explained:

Although honest utterance, even if inaccurate, may further the fruitful exercise of the right of free speech, it does not follow that the lie, knowingly and deliberately published about a public official, should enjoy a like immunity. At the time the First Amendment was adopted, as today, there were those unscrupulous enough and skillful enough to use the deliberate or reckless falsehood as an effective political tool to unseat the public servant or even topple an administration. … Calculated falsehood falls into that class of utterances which ‘are no essential part of any exposition of ideas, and are of such slight social value as a step to truth that any benefit that may be derived from them is clearly outweighed by the social interest in order and morality.’ Chaplinsky v. New Hampshire, 315 U.S. 568, 572.

The incessant perpetuation of the Big Lie shows that the nation has perilously strayed from this principle, and to protect democracy from the corrosive consequences of rampant propagation of deliberate disinformation about the functioning of democracy itself, there will need to be a resuscitation of the principle in the particular context of messages aimed at subverting the valid result of an election.

The behavior of leading figures associated with President Trump’s reelection campaign, including President Trump himself, demonstrates that they believed they were entitled to deliberately lie about the outcome of the election without any risk of criminal punishment. This belief likely stemmed from a change in political culture created by a misinterpretation of United States v. Alvarez (2012), the so-called “stolen valor” case. In that fractured decision, a plurality of the Supreme Court held that an Act of Congress criminalizing false claims of winning the Medal of Honor violated the First Amendment. But the plurality decision did not immunize all deliberately false statements from any form of criminal liability. Instead, the decision was premised on the particular statute being poorly drafted. The plurality opinion made clear that a narrowly tailored statute, justified by an adequate evidentiary predicate of the serious harm it aimed to address, would pass constitutional muster.

The opinion cited a myriad of precedents imposing liability for “defamation, fraud, or some other legally cognizable harm associated with a false statement, such as an invasion of privacy or the costs of vexatious litigation.” As additional examples of permissible statutes criminalizing intentional falsehoods, the plurality listed the Acts of Congress that ban: 1) lying to a government official; 2) perjury; 3) impersonating a government official. The concurrence was even more clear about the power of Congress to criminalize deliberately false speech when there is an adequate predicate and the statute is well-crafted in relation to its objective: “a more finely tailored statute might, as other kinds of statutes prohibiting false factual statements have done, insist upon a showing that the false statement caused specific harm or at least was material, or focus its coverage on lies most likely to be harmful or on contexts where such lies are most likely to cause harm.”

Given a proper understanding of Alvarez, it should be possible for Congress to enact a new statute aimed at criminalizing the kind of deliberately false speech that the Big Lie exemplifies. Without attempting here to draft the statutory language exactly, it could be something along the following lines (although careful consideration of all the relevant pros and cons would need to be assessed before moving forward with any new legislation of this nature):

“It shall be unlawful for any person associated with a campaign for federal office, including the candidate of that campaign, to disseminate a deliberately false statement about the result of the election for the purpose of preventing the true winner of the election from taking office or otherwise altering the official outcome of the election so that it contravenes the electoral choice actually made.”

Crafting the statute in this way limits liability only to persons involved with the campaign; ordinary citizens would not be subjected to criminality for repeating the lie perpetrated by the campaign. Only the dissemination of falsehoods would be covered; stray comments in casual conversations by campaign officials would not. There would need to be proof that the deliberate falsehood was made for the purpose of subverting the election’s outcome. Lies intended to air grievance or sooth bruised egos, while deplorable, would not be criminal. If a statute were carefully limited in this way, it ought to easily survive Alvarez-like analysis. After all, given the horribly pernicious consequences of the Big Lie itself to the ongoing operation of American democracy, there should be no doubt whatsoever that a statute along these lines was aimed at countering a serious harm of a type Congress is entitled to redress. Far from legislating loosely with respect to falsity generally, Congress would be targeting any acutely dangerous category of deliberate dishonesty designed to destroy democracy.

Even with this kind of statute on the books, enforcement would be difficult. The Justice Department would need to weigh heavily the various competing considerations on whether to pursue a prosecution under this statute, especially against a leading figure of the opposing political party. Despite overwhelming evidence to support a prosecution, there would be the question of whether a unanimous jury would convict or, instead, view the prosecution as unwarranted political revenge. Nonetheless, having this statute on the books would have some value as a deterrent effect. In contrast with the current situation, where there was little fear of criminal prosecution for the Big Lie, enactment of this statute would cause candidates and campaign officials to think twice in the future. (The hearings held by the January 6 Select Committee has raised the question whether President Trump, Rudy Giuliani, and others involved in propagating the Big Lie— even after learning from Attorney General Bill Barr that there was no basis for it— might be subjected to criminal prosecution for a conspiracy to defraud the United States under existing statutory law. Some scholars, however, have observed how difficult such a criminal prosecution would be, particularly against the former president.).

If ever again an Attorney General tells a President that the claim of a stolen election is “b s ,” and the President insists on repeating that claim incessantly in an effort to convince members of Congress, or anyone else involved in the process of bringing the election to a conclusion, to overturn the outcome in order to obtain a second term contrary to the actual results of the election, having this kind of carefully crafted statute— tailored precisely to this problem— would create much stronger conditions for bringing a criminal prosecution of the President— as difficult as that trial would be for the country. Moreover, any of the President’s allies who participated in disseminating a known falsehood of this nature would be even more at risk of criminal prosecution and thus presumably would think long and hard before assisting a President (or indeed any elected officeholder) with such a deliberately fabricated scheme to subvert an election’s outcome. While in any specific case the Justice Department still might decline to prosecute for the benefit of the nation, the law itself should make clear that this kind of deliberate dishonesty for purposes of election subversion is criminal behavior and in no way protected by the First Amendment.

The Difference Between Twitter and Texting

The Constitution had a Postal Service Clause before it had a Free Speech Clause. Article I, Section 8 of the original Constitution empowers Congress “to establish post offices and post roads.” This enumerated power recognizes the need for government to provide the infrastructure to enable to distribution of communication. The government can authorize private companies to perform this postal service, or even turn a previously government-operated postal service into a private-sector entity. But whatever entity performs this postal service is still performing a government function.

The letters, books, magazines, and other writings that get distributed through the postal system are forms of expression protected by the First Amendment, but the postal service itself is not engaging in its own protected First Amendment activity. In fact, the postal service has no choice over what communications flow through its system. It must accept all lawful mailed content willing to pay the applicable postal rate. This is true even if the postal service is run by a private-sector company rather than the government itself.

The same principle applies to modern technologies that the authors of the original Constitution could not have envisioned. The telephone company is a “common carrier” obligated to transmit all lawful phone calls, regardless of content, even if the telephone company were to find the content distasteful. It is the parties to the phone conversation who are exercising their First Amendment right to freedom of speech by their choice of what to say to each other in the phone call. The phone company itself is not engaging in its own First Amendment protected expression by transmitting the conversation.

As a common carrier, a phone company is immune from liability for the speech it distributes through its system. A phone call might contain defamatory content, for example, but the phone company would not be liable for the defamation. In this respect, the phone company differs from a radio station that transmits a defamatory broadcast.

Texting, as the term itself applies, is a written form of a cell phone call. Therefore, it makes sense to apply the “common carrier” principle to texting services as much as to cell phone services. The company that provides the service, as a common carrier, is obligated to distribute the text message regardless of its content (as long as the customer pays the applicable rate for the texting service). In return, the company is not liable if the text message is defamatory or otherwise tortious.

Given modern technology, it’s possible to send group texts, not just one-on-one texts. These, too, can be subject to the same “common carrier” rules: as long as the parties sending and receiving the text messages are willing to pay whatever applicable rates, the texting service provider can be obligated to transmit these texts on equal terms regardless of content. Likewise with email, the modern equivalent of old-fashioned mail (as its name implies); email service providers can be obligated to transmit all email on equal terms, regardless of content, including bulk email.

Twitter is different. While a tweet may resemble a text, its dissemination does not follow “common carrier” principles— at least not when Twitter is operating in its default mode. (Twitter users have the option of having tweets arrive in a purely chronological order, a format that more resembles a content-neutral “common carrier” basis. This toggling between two different ways Twitter operates raises the question whether its legal status should be dualistic, with “common carrier” protections from content liability when users turn on the purely chronological flow of tweets but lacking the same kind of “common carrier” status when Twitter operates in its regular mode, prioritizing tweets based on their content.) The company makes its own editorial judgments about what tweets to promote or demote. In this respect, it is closer to a radio station or newsstand (or bookstore) than it is to a phone company or postal service.

As a consequence, it does not make sense to provide Twitter, or other social media companies, the same immunity from liability provided to common carriers. If Twitter truly became a common carrier, converting itself essentially into a form of text-messaging service— with no algorithms to prioritize some tweets over others based on content— then retaining Twitter’s immunity for the content of tweets would be appropriate. But as long as Twitter is in the business of hawking some messages more than others, just as a newsstand prioritizes what magazines to display, the Twitter company should be on the hook if its business decisions cause injury to others.

Some have argued that social media companies, like Twitter, cannot survive if they lose their special statutory immunity for the speech they disseminate. But they have no right to survive if the only way they can do so is by causing harm for which they are unwilling to provide compensation. That’s like saying that factories should not be required to pay for the pollution they cause because making them pay would force them out of business. The business should exist only if it can cover the costs it causes.

Therefore, social media companies should be put to a basic choice: convert to common carrier status in order to re- tain their immunity from liability, or else lose their immunity from liability in order to retain their right to make judg- ments about what speech to prioritize based on content. To be clear, social media companies should not be forced to be common carriers if they would prefer to retain editorial judgment over the prioritization of the content they distribute. Just as newspapers and bookstores should not be required to be common carriers, so too with social media companies. (The Supreme Court has before it a case from Texas raising this basic point.) But social media compa- nies can and should be put the choice between the benefits and burdens of exercising editorial judgment on the one hand, like a newspaper, versus the benefits and burdens of being a common carrier without editorial judgment, on the other. What social media companies should not be able to insist upon is a right in effect of having their cake and eat it too: to have all the benefits of editorial judgment with none of its burdens and all the benefits of common carrier status with none of its obligations.

Putting social media companies to this basic choice would not by itself guarantee the preservation of democracy. But it would create more favorable conditions for democracy to function. Just like newspapers and television news channels must exercise some editorial care about the content of their messages— as some news stations have relearned in the aftermath of 2020, given their exposure to liability for defamatory statements about voting machine vendors— so too would democratic discourse be less polluted with disinformation if social media companies had to bear the cost for damages caused by the defamatory or other wrongful messages they promoted.

Congress may not wish to legislate in the public interest in a way that would remove a windfall that the social media companies currently receive from their special immunity from liability. Historically Congress has been reluctant to make factories pay for their pollution. Still, the question is what should Congress do. The answer that democracy itself demands: make social media companies pay for the harmful verbal pollution they spew. Social discourse would still be robust; it just would be cleansed of the toxicity that currently exists only because it’s getting a free ride and not required to pay for its corrosive social disutility.

Recently, there has been a widespread recognition of the dangerous decline in civics education during the last half-century and thus the corresponding need for its revitalization. One aspect of this need, however, has been overlooked. The focus has been on civic literacy: knowing the nation’s history, including the fundamental norms that underlie the Constitution and the development of democracy in the United States. This civic literacy is indeed essential for the perpetuation of self-government. But civics education must be more than just achieving civic literacy; it needs to include another component, which we can call civic numeracy.

As shown above, in the discussion of ranked-choice voting and Condorcet winners, as well as the problem of plurality-winning primaries and Condorcet losers, the math of elections is not always straightforward. On the contrary, the challenges of designing a democratic electoral system— so that the winner of an election is the candidate most preferred by a majority of voters— are daunting. There are important tradeoffs between alternatives. For example, is it better to have two rounds of voting, or a single round using a ranked-choice ballot? Even if electing Condorcet winners is desirable, are there countervailing reasons why it would be better to design an electoral system that determines the winner based on other criteria (like sequential elimination of the least popular candidate)?

The level of mathematical aptitude necessary to consider these questions is not beyond what is ordinarily taught in K-12 math classes, like trigonometry and algebra. But the mathematics of elections is not part of the standard K-12 curriculum in the United States. It should be. Citizens would have a much better understanding of how the electoral process is designed to work, and how it might be reformed to work better, if they were taught in school about the basic mathematical principles involved. Even something as simple as the difference between plurality and majority winners is not well understood by most voters, and this lack of education on the fundamentals of electoral science is an impediment to self-government. Yet teaching these key concepts could be made accessible to students, using analogous examples from sports tournaments, chess competitions, and the like. Elections, after all, are competi- tive events not dissimilar from the kinds of competitions that students themselves undertake as part of their own overall schooling.

Educating for civic numeracy should not be controversial. The idea of a Condorcet winner is simply a numerical concept, given the inherent nature of a multi-candidate election. Teaching citizens to understand the concept, so that they can then form their own opinions on what type of electoral system would be best, is not an attempt to impose any values on students. Just like teaching students what an interest rate is does not impose any view on what policy the government should adopt regarding interest rates— this teaching simply provides students with an important mathematical concept they need to understand in order to be able to make their own policy judgments. So too teaching students the analytical tools involved in electoral system design is merely to provide citizens with the foundation they need to make their own informed judgments on this core topic.

To be sure, the polity’s choice about what electoral system to adopt is itself necessarily a collective value judgment. Whether an electoral system should be structured to elect a Condorcet winner, even when that candidate’s appeal to a majority of voters does not reflect significant first-choice enthusiasm among any of them, is inevitably a tradeoff among alternative approaches to the philosophy of democracy itself. A well-designed K-12 civics education would not attempt to indoctrinate its students on any particular point of view with respect to these philosophical matters. But if undertaken sensitively to the freedom of students to make up their own minds on these matters, civics education can help prepare citizens to understand the tradeoffs that they themselves must make in deciding what conception of democracy they think best. Moreover, giving citizens the basic tools of civic numeracy is itself an essential prerequisite in order for them to have even an understanding of what is at stake in these tradeoffs.

While introducing civic numeracy into the curriculum will not achieve immediate improvement in American democracy, the benefits will accrue over time. If we can keep democracy going between now and then, future generations may be able to make it work even better and to make its guardrails even more robust.