Chapter 37: Democratic Erosion — Institutional Norms, Constitutional Hardball, and the Stress Test of the 2020s

A Note Before You Read

This is the most ideologically sensitive chapter in the book. It also is the one that, by most measures, has changed the most between the date a textbook like this would have been written ten years ago and the date it is being written now. Comparative political scientists have been studying democratic erosion — the gradual weakening of institutions, norms, and checks rather than their sudden collapse — for thirty years. Until roughly 2016, the case studies were almost entirely outside the United States: Hungary under Viktor Orbán, Turkey under Recep Tayyip Erdoğan, Venezuela under Hugo Chávez, the Philippines under Rodrigo Duterte, Russia under Vladimir Putin. Since 2016, that literature has turned its instruments on the United States itself, and the readings have been mixed and contested.

The chapter takes three commitments seriously. First, it states empirical findings as findings, including findings that some readers will find politically uncomfortable. Second, it steel-mans the major positions across the political spectrum on what those findings mean. Third, it does not predict the future, declare the United States a failed democracy, or declare American institutions invulnerable. The trajectory is uncertain, the interpretation is contested, and a textbook is not the place to resolve a debate that the country has not yet resolved. The chapter equips you to think clearly about the question. You will reach your own conclusions.

A second note. Wherever the chapter uses charged terminology — "insurrection," "stolen election," "patriotic protest," "rigged," "common-sense reform," "voter suppression" — it does so in quotation marks and identifies which speakers used the term, why, and what the descriptive (non-team-jersey) institutional language for the same event would be. The Balance Guide in the instructor companion sets this rule for the whole textbook. It is most important here.

37.1 Democratic Erosion: What the Concept Names

Most Americans, asked when a democracy "fails," picture something dramatic: tanks in the capital, a televised announcement that elections are canceled, a uniformed officer behind a desk where a president used to sit. That picture is democratic collapse — a sudden, often violent transition from a constitutional regime to an authoritarian one. Examples in the twentieth century include Chile in 1973 (the Pinochet coup), Argentina in 1976 (the Proceso military junta), Greece in 1967 (the Colonels' coup), and a long list of African and Asian post-colonial transitions. Collapse is the case classical political theorists of authoritarianism — from Hannah Arendt to Juan Linz — primarily had in mind.

Democratic erosion names a different phenomenon. It is the gradual weakening of the institutions, norms, checks, and civic capacities that distinguish a democracy from its alternatives. It happens through formally legal moves: contested but lawful executive orders, lawful but norm-violating legislative actions, lawful but politicized prosecutions and pardons, lawful but partisan judicial appointments. There is no single moment at which a democracy ends; instead, the institutions that were supposed to constrain political power are progressively hollowed out, sometimes over a decade or more. The end-state, if erosion is unchecked, is a regime in which elections are still held but the playing field is no longer level — the ruling party controls the courts, the press, the electoral machinery, the agencies that investigate corruption, the universities, and the means of dissent. Hungary is the canonical recent case in the comparative literature. Russia of the 1990s into the 2000s is another. Turkey from 2002 onward is a third. None of these governments staged a coup. All of them, comparative scholars argue, eroded democratic institutions to a degree that significantly constrained competitive politics.¹

The distinction matters because erosion does not present itself as erosion. Each individual move can be defended as legal, as a response to provocations from the other side, or as a corrective to past wrongs. The institutions look like they still work — there are still elections, courts, newspapers, opposition parties — and the effects of erosion accumulate slowly enough that no single news cycle reports them as a regime change. Whether erosion has happened, in any specific case, is partly a measurement question and partly a matter of analytical judgment. There are technical instruments for measuring it; there are also disagreements about what the instruments measure and how much weight to give them.

Two things to keep in mind throughout this chapter. First, "erosion" is not a synonym for "policy outcomes I dislike." A government can pursue policies you strongly oppose without eroding democracy in the institutional sense — and a government can erode democracy while pursuing some policies you support. The category names a specific institutional phenomenon, not a substitute for partisan disagreement. Second, "erosion" is also not a synonym for all norm violations or all aggressive uses of formal power. Politics has always involved hardball. The empirical question is whether and where the magnitude, frequency, and trajectory of hardball has shifted in ways the long-running comparative-democracy literature flags as concerning. The chapter takes that question seriously without prejudging the answer.

37.2 Measuring Democracy: V-Dem and Freedom House

How do you tell whether a democracy is eroding? The two most-cited measurement frameworks are the Varieties of Democracy (V-Dem) project, headquartered at the University of Gothenburg in Sweden, and Freedom House, an American nonprofit that has been measuring political rights and civil liberties globally since 1972.²

V-Dem distinguishes among five conceptions of democracy — electoral, liberal, participatory, deliberative, and egalitarian — and constructs separate indices for each. The flagship measure, the Liberal Democracy Index (LDI), combines electoral fairness with rule of law, judicial independence, civil liberties, and constraints on executive power. V-Dem's coding is performed by panels of country experts (typically five or more per country per year) and aggregated through a Bayesian item-response-theory model, with confidence intervals reported. The dataset goes back to 1789 for many countries, allowing long-run comparisons.

Freedom House publishes an annual Freedom in the World report that classifies every country as Free, Partly Free, or Not Free, based on a 100-point composite score (40 points for political rights, 60 for civil liberties). The methodology is less granular than V-Dem's — fewer indicators, less complex aggregation — but the coverage is similarly comprehensive, and the data goes back to 1972, providing another long time series. Both projects are funded by a mix of foundations (V-Dem heavily by European foundations and the European Research Council; Freedom House historically by the U.S. State Department and various private donors), and both have been criticized — Freedom House more often than V-Dem — for the geopolitical orientations of their funders. The criticisms run in both directions: Freedom House has been accused both of being too soft on U.S. allies and of being too hard on countries the U.S. opposes. V-Dem, with European funding and a more academic orientation, attracts less of this critique but is not immune to it. For the purposes of this chapter, what matters is that both indices show a similar pattern for the United States since the mid-2010s, which makes the pattern more credible than if only one index showed it.

The pattern is this. Both V-Dem and Freedom House show the United States declining on their composite measures from a peak in the late 1990s and early 2000s, with the decline accelerating after 2016. V-Dem's Liberal Democracy Index for the United States dropped from approximately 0.88 in 2015 to approximately 0.73 by 2024, the largest decline among advanced industrial democracies in that period.³ Freedom House's score for the United States dropped from 94 in 2010 (out of 100) to 83 in 2024, with most of the decline concentrated in three areas: integrity of elections (loss of points related to the contested 2020 transition and subsequent state-level changes to election administration), functioning of government (loss of points related to executive ethics norms, congressional oversight failures, and politicization of the Justice Department), and freedom of expression (loss of points related to harassment of journalists, polarization of media, and government pressure on speech platforms). The U.S. is still classified as Free; it is no longer at the top of the Free category, where it sat throughout most of the post-Cold War period.

You should resist two opposite over-readings of these numbers. The first over-reading is to treat them as definitive: "V-Dem says the U.S. democracy is collapsing, so it is collapsing." V-Dem and Freedom House are imperfect instruments, the underlying coding choices are debatable, and a multi-point drop on a 100-point scale is a real signal but not an unambiguous one. Reasonable scholars who accept the methodologies disagree on how to interpret the magnitude. The second over-reading is to dismiss the numbers as politicized: "V-Dem is European-funded and reflects European political preferences." There is some truth in the observation, but a coordinated decline across multiple independent measurement frameworks (V-Dem, Freedom House, the Economist Intelligence Unit's Democracy Index, the Pew Research Center's cross-national data, the Bertelsmann Transformation Index) is harder to dismiss as artifact. The honest reading is: something is being measured, it has been moving in one direction for a decade, and the direction is consistent with what the comparative-democracy literature would predict for a country experiencing the kinds of stresses the United States has experienced since 2015–16.

What are those stresses? The chapter now turns to them.

37.3 The "Stress Test" Framing

The American political system was designed for disagreement. Theme 1 of this book, repeated since Chapter 1, is the Madisonian observation that the Founders did not believe in consensus and did not design institutions for it. They expected factions, expected rivalry, expected ambition to "counteract ambition" (Federalist No. 51), and built a structure of separated and overlapping powers precisely so that no single faction or branch could dominate. By that design logic, the question is not whether American institutions are under stress — they always have been — but whether the current stresses are exceeding what the system was designed to handle.

A useful analogy: a bridge engineered to carry trucks of a certain weight is designed with safety margins. It can carry heavier loads than the rated maximum, sometimes by a factor of two or three, before failing. Crossing the rated maximum a few times does not mean the bridge will collapse on the next trip. But if the loads are rising over time, if the redundancies are being eroded, if the steel is showing fatigue, then "still standing" is not the same as "operating within design tolerance." The right question is not whether American democracy has collapsed (it has not) or whether it is operating well (it is not, by most observers' lights, including conservative observers like Yuval Levin and Bill Kristol who write from inside the institutional tradition). The right question is whether the system is closer to its operating tolerance than it was twenty years ago — and that question is empirical.

Several stresses have come together since the mid-2010s, none of them entirely new, all of them intensifying:

• Polarization. Mass-public polarization has risen substantially since the 1990s, but elite polarization has risen faster (Chapter 17). The two parties are sorting into ideologically homogeneous coalitions to a degree the post-WWII period did not exhibit. The political scientists Christopher Hare and Keith Poole, using Congress's roll-call voting record, find that the ideological distance between the median Democrat and median Republican in both chambers is at the highest level since Reconstruction.⁴

• Decline of trust in institutions. Gallup's annual confidence-in-institutions battery shows declining public trust across nearly every American institution measured since the 1970s — Congress, the presidency, the Supreme Court, the press, the medical profession, organized religion, the military. The military and small business retain the highest levels of public trust; Congress and television news the lowest. The decline is not partisan in any simple sense — it is a long-term feature of American public opinion across both parties — but the partisan distribution of trust has scrambled in important ways since 2016, with each party's voters now distrusting the institutions associated with the other party much more than was historically true.⁵

• The fragmentation of shared facts. Chapter 24 (Media) treated the rise of partisan-aligned news, the decline of broadcast television's role as a shared-fact gatekeeper, and the role of social-media algorithms in promoting engagement-optimizing content. The empirical finding is that Americans on the left and right increasingly draw their political information from non-overlapping sources, hold non-overlapping factual beliefs about contested events, and discount the same evidence to different degrees depending on its source. This is not symmetric in every way — research by Brendan Nyhan, David Rothschild, and others has found some asymmetries in misinformation exposure across partisanship — but it is also not purely a one-side phenomenon. Both information ecosystems contain elements that reinforce in-group beliefs and dismiss out-group evidence.⁶

• Institutional norm violations. Section 37.4 takes this up in detail. The headline finding is that since roughly the late 1990s — and accelerating sharply since 2016 — informal norms that previously constrained political behavior have been violated more frequently, by both parties, with the pace of norm violation increasing.

• Election-administration disputes. The 2000 election (the Bush v. Gore decision; Chapter 22) raised the salience of election administration. Voter-ID laws, mail-ballot expansion, registration-list maintenance, and election-night procedures have since become objects of partisan litigation in nearly every cycle. The 2020 election (Section 37.6) marked a significant escalation; the post-2020 wave of state-level changes (Chapter 36) has further intensified disputes.

• The Capitol breach of January 6, 2021. The events of that day are treated as an institutional fact in Section 37.6 and Case Study 1. Whether they were a one-off shock or part of a longer pattern is among the most contested questions in this chapter.

Take the stresses together, and the question political scientists are asking is whether the cumulative load on the institutions — the cumulative demand for the institutions to mediate disputes, transfer power peacefully, hold officials accountable, and restrain executive overreach — is closer to or further from the system's operating tolerance than it was a generation ago. The answer most observers give is "closer." The disagreement is about how much closer, in which directions, and what (if anything) should be done about it.

37.4 Norms vs. Rules

The U.S. Constitution is a set of formal rules. It is also remarkably short — about 4,400 words in the original, with another 3,000 words across the 27 amendments. It does not specify how the federal government actually operates day to day. It is silent on the existence of political parties, on the structure of congressional committees, on the cabinet (only vaguely sketched), on judicial review (established by Marbury v. Madison in 1803, not by the document itself), on the size of the Supreme Court (set by statute, not by the Constitution), on filibusters (a Senate rule, not a constitutional provision), on the federal civil service (Pendleton Act of 1883, not the Constitution), and on hundreds of other features of the system that any modern citizen would describe as part of "American government."

The gaps in the formal rules are filled, in any working democracy, by informal norms — patterns of behavior that political actors observe even when they could legally violate them. The political scientists Steven Levitsky and Daniel Ziblatt, in How Democracies Die (2018) and Tyranny of the Minority (2023), argue that the two most important norms in any democracy are mutual toleration (the recognition that political opponents are legitimate competitors, not enemies to be eliminated) and institutional forbearance (the disciplined non-use of legal powers that, while available, would damage the system if pushed to their limits). On their account, democratic erosion proceeds primarily through the erosion of these two norms — opponents are increasingly framed as illegitimate, and powers that were previously held in reserve are increasingly deployed.⁷

A partial inventory of pre-2016 American political norms — many of which had been observed for decades by both parties:

• Peaceful transfer of power. A defeated incumbent concedes after the votes are counted (or after legal challenges are exhausted), participates in the inauguration of the successor, and allows an orderly handover of executive authority. With the brief exception of the 1860–61 secession crisis (a category of its own), every American presidential transition between 1797 and 2017 followed this pattern, including in the deeply contested 1876 (Hayes-Tilden), 1960 (Kennedy-Nixon), and 2000 (Bush-Gore) elections.

• Presidential disclosure of taxes. From 1976 (Gerald Ford voluntarily releasing partial returns) through 2012 (Mitt Romney releasing two years of returns under criticism for not releasing more), every major-party nominee for president released personal income-tax returns, generally for the most recent several years. This was norm, not law.

• Presidential separation from Justice Department investigations. After Watergate, an informal but firm norm developed under which the White House would not direct the Justice Department on the conduct of investigations involving the President, the President's family, the President's businesses, or the President's political opponents. Attorneys General from both parties (notably Edward Levi under Ford, Janet Reno under Clinton, Michael Mukasey under Bush 43, and Eric Holder under Obama) treated this norm as foundational, even when they took political heat for doing so.

• Senate "blue slips" for judicial nominees. When a federal judicial nominee was named for a court within a senator's home state, the Senate Judiciary Committee historically deferred to that home-state senator's approval (a returned "blue slip") before proceeding with confirmation. The strict version of the practice eroded over decades but was largely observed until 2017.

• Restraint on the use of executive power for personal benefit. Presidents have always exercised executive power in ways that benefit their political coalitions. The norm — observed unevenly but as a general expectation — was that executive power would not be used for the President's personal financial or legal benefit, distinct from the political coalition's interests.

• Bipartisan respect for election results. Once the votes were certified, an election was treated as having produced a legitimate result. Defeated candidates and their supporters could (and frequently did) argue that the result was unfortunate or that the electorate had erred; they did not dispute the result's validity in the institutional sense.

• Restraint on impeachment as a partisan tool. Impeachment was reserved for serious offenses — "high crimes and misdemeanors" in the constitutional language. Articles of impeachment introduced as routine partisan moves were uncommon for most of American history; the Andrew Johnson impeachment (1868) and the Bill Clinton impeachment (1998) were treated as exceptional events even by their critics.

• Restraint on filibuster use. The Senate filibuster was used relatively sparingly until the 1970s, and even after the rule changes of 1975 (which lowered the cloture threshold from two-thirds to three-fifths), most legislation could clear a 60-vote threshold without extended floor debate. Filibuster use accelerated sharply in the 1990s and 2000s and reached unprecedented levels in the 2010s.

• Restraint on court-curbing. Proposals to expand the Supreme Court, to abolish or restructure lower courts, or to discipline judges through legislation were treated as rare and exceptional. The most famous case — Franklin Roosevelt's 1937 proposal to expand the Court to 15 justices to overcome New Deal vetoes — was rejected by his own party in Congress and remained the standard cautionary tale.

• Pardons. Presidents have always issued pardons, including controversial ones (Ford's pardon of Nixon in 1974, Carter's pardon of Vietnam-era draft resisters, George H. W. Bush's pardon of Iran-Contra figures, Clinton's pardons of Marc Rich and others on his last day in office, Obama's commutation of Chelsea Manning's sentence). The norm was that pardons would be reviewed by the Justice Department's Office of the Pardon Attorney, would generally follow a process and standards, and would be balanced by overall restraint on the use of the power for personal or political-coalition allies.

• Inspectors General. A 1978 law established statutory Inspectors General in major federal agencies, whose mission is to investigate fraud, waste, and abuse, and to report independently to both the agency head and Congress. By informal norm, IGs were typically allowed to complete their tenures, were dismissed only after notice to Congress and for stated cause, and were given operational independence to pursue investigations without political interference.

This list is not meant as nostalgia. The norms were not always observed; their operation was sometimes conveniently flexible; some of them protected outcomes (white male political dominance in the early twentieth century, for example) that the country has explicitly moved beyond. The list is offered descriptively: these were practices political actors followed even when not legally required to do so, and they are part of what made the formal Constitution operate the way it operated for most of the twentieth and early twenty-first centuries.

The key empirical claim of the comparative-democracy literature is that norms can be violated faster than they can be restored. Once a norm is broken, the cost of re-establishing it is high — political opponents will (rationally) hesitate to return to it unilaterally if they fear the other side will not reciprocate. Norms thus operate as a coordination equilibrium, and once the equilibrium is disrupted, it can be very difficult to coordinate back. Levitsky and Ziblatt call this the "guardrails of democracy" thesis and argue that the United States, since roughly 2016, has been living through a sustained stress on the guardrails.

37.5 Constitutional Hardball

The legal scholar Mark Tushnet introduced the term constitutional hardball in a 2004 article to describe a category of political moves that share two features: they are legal (consistent with the formal rules of the Constitution), and they violate norms (depart from how the same powers had previously been exercised, in ways that benefit the party using them and damage the institutional system).⁸ Examples are bipartisan and span decades. The point of this section is not to declare which party has been worse — that is a contested empirical question, treated below — but to demonstrate that hardball is a pattern, not the property of one side, and that the trajectory has been escalation across the political spectrum.

A bipartisan inventory:

Filibuster expansion. The Senate's Rule XXII (cloture, current threshold 60 votes for most legislation) has been used with rising frequency by both parties. From 1917 (when cloture was first established) through 1970, cloture was filed fewer than 60 times total. Through the 1980s and 1990s, frequency rose. From 2007 to 2014, with Democrats controlling the Senate and Republicans in the minority, cloture was filed at then-record rates. From 2015 to 2020, with Republicans controlling the Senate and Democrats in the minority, cloture continued at high rates. Both parties, when in the minority, have used the filibuster more aggressively than the prior generation of senators did. Both parties, when in the majority, have proposed reducing or eliminating the filibuster — and have followed through partially when the costs of inaction were sufficiently high.⁹

The "nuclear option." In November 2013, a Democratic-majority Senate, led by Majority Leader Harry Reid, invoked the so-called nuclear option to lower the cloture threshold for executive-branch nominations and lower-court judicial nominations from 60 to 51 votes. Reid argued that Republican filibustering of Obama's nominees had reached a level that prevented the executive branch from functioning. In April 2017, a Republican-majority Senate, led by Majority Leader Mitch McConnell, extended the same change to Supreme Court nominations, allowing Neil Gorsuch's confirmation by simple majority. McConnell argued that Democrats' filibuster of Gorsuch — a continuation of the Garland dispute (below) — required the move. Both Reid and McConnell took heat from their own caucuses for the changes; both argued that the other side had forced their hands. The asymmetric-hardball literature debates whether the 2013 move or the 2017 move was the more significant escalation; the positions in that debate track partisan affiliations imperfectly but predictably.¹⁰

The Bork rejection (1987). In July 1987, President Reagan nominated federal appellate judge Robert Bork to the Supreme Court. Bork was a leading conservative legal scholar — a former Yale Law School professor, a former Solicitor General — and his judicial philosophy was unmistakably originalist. Senate Democrats, led by Judiciary Chairman Joe Biden, organized a sustained opposition campaign, ultimately defeating the nomination 42-58 in October 1987. The Bork rejection has been read in several ways: (a) as a legitimate exercise of the Senate's "advice and consent" role, in which a nominee's judicial philosophy was scrutinized and found wanting; (b) as the moment at which judicial nominations became fully politicized in the modern sense, with media campaigns, advocacy-group mobilization, and ideological framing; (c) as an early case of constitutional hardball, with the term "borking" entering political vocabulary as a verb.¹¹

The Garland blockade (2016). In February 2016, Justice Antonin Scalia died unexpectedly. President Obama nominated Judge Merrick Garland of the D.C. Circuit, a moderate respected on both sides of the aisle, to fill the vacancy. Senate Majority Leader Mitch McConnell announced within hours of Scalia's death that the Republican-majority Senate would not consider any Obama nominee, holding the seat open for the next president. The Senate held no hearings and no vote. After Trump's election, Justice Neil Gorsuch was nominated and confirmed in April 2017. McConnell defended the move on the grounds that no Supreme Court vacancy had been filled in the final year of a presidency since 1932, and that the principle of "letting the voters decide" was a legitimate procedural rule (later termed the "McConnell Rule"). Critics argued that the precedent McConnell cited was inaccurate, that Garland's qualifications were unchallenged, that no formal Senate rule supported the blockade, and that the move represented a categorical violation of the institutional norm under which Supreme Court vacancies were filled by the sitting president.¹²

The Barrett confirmation (2020). In September 2020, Justice Ruth Bader Ginsburg died approximately seven weeks before the November presidential election. Within days, McConnell announced that Senate Republicans would proceed with a Trump nomination. President Trump nominated Judge Amy Coney Barrett of the Seventh Circuit; the Senate confirmed her on October 26, 2020, eight days before the election. The confirmation vote was 52-48, with Senator Susan Collins (R-Maine) voting against on the same procedural grounds McConnell had cited in 2016. Critics argued that the confirmation directly contradicted the "McConnell Rule" of 2016 — if it was inappropriate to fill a vacancy in an election year in 2016, it was inappropriate in 2020. McConnell responded that the relevant procedural distinction was unified party control of the Senate and presidency (which Republicans had in 2020 but not 2016), not the timing within an election year. Whether this distinction is principled or post-hoc is one of the most-cited examples in the literature on asymmetric hardball.¹³

Court expansion proposals (2020–21). During the 2020 presidential primary and after the Barrett confirmation, several Democratic candidates (Pete Buttigieg, Kamala Harris in some statements, Elizabeth Warren after caveats, Beto O'Rourke explicitly) proposed expanding the size of the Supreme Court from nine to thirteen or fifteen justices. After the November election, with Democrats holding the presidency and narrow majorities in both chambers, the Biden administration appointed a Presidential Commission on the Supreme Court of the United States to study court reform options, including expansion. The commission's December 2021 final report did not endorse expansion, and the Biden administration did not pursue legislation. Court-expansion proposals also did not pass through Congress during the 117th or 118th. Critics argued that even floating expansion crossed a long-standing institutional norm — the same one that had defeated FDR's 1937 plan; defenders argued that a Court arguably stocked through the 2016 and 2020 hardball moves represented its own departure from norm and that proposing reform was a legitimate response.¹⁴

Recess appointments. Both parties have used recess appointments — the constitutional power to fill vacancies during a Senate recess without confirmation — more aggressively than the early American practice contemplated. The Obama administration's 2012 recess appointments to the National Labor Relations Board and the Consumer Financial Protection Bureau were unanimously struck down by the Supreme Court in NLRB v. Noel Canning (2014), with the Court holding that the Senate's pro-forma sessions counted for purposes of the recess clause. The Trump-1 administration explored the use of recess appointments at several points but did not deploy them aggressively after Noel Canning. The Trump-2 administration, beginning in 2025, has signaled greater willingness to test the limits of the post-Noel Canning doctrine.

Extreme partisan gerrymandering. Chapter 35 treated this in detail. Both parties have engineered sharply biased maps when they controlled the relevant state legislative process. Republican-drawn maps after 2010 in North Carolina, Wisconsin, Ohio, Michigan, and Pennsylvania produced the largest aggregate partisan asymmetries in the country; Democratic-drawn maps after 2020 in Maryland, Illinois, and (attempted) New York produced asymmetries that were narrower in number but, where drawn, severe. The Supreme Court in Rucho v. Common Cause (2019) held that partisan gerrymandering claims are non-justiciable in federal court, leaving the question to state courts and state constitutions, with the result varying by state.

Pardon power. Both Trump and Biden have used the pardon power for politically connected recipients in ways that earlier administrations would have considered exceptional. President Trump issued pardons to Roger Stone, Paul Manafort, Steve Bannon, and others — political allies who had been convicted of offenses tied directly to the President's own administration's conduct. President Biden, in his final weeks in office, issued preemptive pardons to family members (including his son Hunter Biden, after stating he would not pardon Hunter), to Dr. Anthony Fauci and General Mark Milley, and to members of the House Select Committee on January 6 — a category of preemptive pardon for which there is little modern precedent. Both pardons have been criticized from both directions: defenders argue each was a legitimate use of constitutional authority for plausible reasons; critics argue each represented a normalization of pardon-as-political-instrument. The asymmetry, here, is that the underlying questions are different (Trump pardons were typically for already-convicted defendants tied to his administration's conduct; Biden's preemptive pardons were for individuals who had not been charged), but the direction of norm departure runs in both cases.¹⁵

The pattern is clear: the trajectory across both parties is toward more aggressive use of formal powers in ways that violate the institutional norms that had previously constrained those powers. Where the parties differ — and the differences matter to the asymmetric-hardball debate — is in the frequency, magnitude, and strategic significance of the moves. We turn to that debate after first establishing the institutional facts of the most acute period.

37.6 The 2020 Election, January 6, 2021, and the Trump-2 Period

This section states the institutional facts of the most contested period in modern American political history. It does so without editorializing about Trump or about the people who participated in the events. The chapter's commitment is to surface what occurred and what the institutional consequences were; the moral and political judgments are yours to form.

The 2020 election. On November 3, 2020, Joseph R. Biden Jr. defeated incumbent President Donald J. Trump in the popular vote (81,283,501 to 74,223,975, a margin of approximately 7 million votes) and in the Electoral College (306 to 232). The election was administered by state and local officials in all fifty states, with significant expansions of mail-in and early voting due to the COVID-19 pandemic. State election officials of both parties certified the results. Federal courts heard and rejected more than sixty post-election lawsuits brought by the Trump campaign and aligned parties; many were dismissed for lack of standing or evidence, and several were heard on the merits and decided against the Trump position. The Department of Homeland Security's Cybersecurity and Infrastructure Security Agency (CISA), then headed by Trump appointee Christopher Krebs, issued a statement on November 12, 2020, calling the 2020 election "the most secure in American history." Krebs was dismissed from his position on November 17. Attorney General William Barr, also a Trump appointee, told the Associated Press on December 1, 2020, that the Department of Justice had "not seen fraud on a scale that could have effected a different outcome in the election." Barr resigned shortly afterward.¹⁶

The empirical record across more than two years of investigation, litigation, audits (including partisan audits in Arizona, Wisconsin, and other states), and academic analysis has not produced evidence of fraud or systemic error sufficient to have changed the 2020 outcome in any state. State-by-state recounts, signature audits, and forensic reviews — including those commissioned by the losing campaign — have repeatedly confirmed the results. President Biden won the 2020 election. This is not a contested empirical claim at the level of the data. It is, however, a claim that a substantial share of Republican voters has continued to dispute, with poll numbers in the 30 to 60 percent range across various surveys depending on question wording.¹⁷ The persistent gap between the empirical record and a meaningful share of the public's belief about the election is itself one of the most-cited examples in the comparative-democracy literature of how mass-public belief about election integrity can decouple from institutional findings — a pattern with parallels in some other recent comparative cases.

The peaceful-transfer commitment. During the 2020 campaign, President Trump declined on multiple occasions to commit to accepting the result if he lost or to a peaceful transfer of power. After the election, he did not concede and asserted publicly and repeatedly that the election had been "stolen." The Trump campaign and aligned parties pursued litigation, recount efforts, and public-pressure campaigns aimed at preventing certification of state election results in several states (Arizona, Georgia, Michigan, Pennsylvania, and Wisconsin in particular). On December 14, 2020, state electors cast their Electoral College votes, with Biden receiving 306 to Trump's 232. Subsequent events included pressure on state legislatures and state election officials to set aside the results, the "fake electors" plan in seven states (in which alternate slates of pro-Trump electors signed certificates and submitted them in parallel to the certified Biden electors, an action subsequently the basis of state criminal indictments in Georgia, Michigan, and elsewhere), and public and private pressure on Vice President Mike Pence to refuse to count the certified electoral votes in his ceremonial January 6 role.¹⁸

January 6, 2021. Case Study 1 of this chapter presents the events of the day in detail. In summary: a joint session of Congress was scheduled to convene at 1:00 p.m. on January 6, 2021, to certify the Electoral College votes — the constitutional procedure under which the certification is largely ceremonial but legally binding. Several House and Senate Republicans had announced they would object to the certification of certain states' electors. Earlier that morning, President Trump addressed a rally on the Ellipse, calling on supporters to march to the Capitol. The crowd at the Capitol grew through the morning. At approximately 12:53 p.m., perimeter barricades were breached. By 2:00 p.m., portions of the building had been entered. Members of Congress were evacuated. Vice President Pence was moved to a secure location after declining a Secret Service offer to leave the Capitol grounds. Members of the rioting crowd entered the Senate chamber, the Speaker's office, and other portions of the building. A Capitol Police officer, Brian Sicknick, suffered medical events later that evening and died on January 7; the medical examiner determined the death was due to natural causes (strokes), with the Capitol breach as a contributing context. Capitol Police Officer Howard Liebengood and Metropolitan Police Officer Jeffrey Smith died by suicide in the days following. Other officers reported sustained injuries. One protester, Ashli Babbitt, was shot and killed by a Capitol Police officer while attempting to climb through a broken window into a corridor near the Speaker's Lobby. Other protesters and bystanders died from medical events during the day.

The Capitol was secured by approximately 5:40 p.m. with the assistance of the D.C. Metropolitan Police, the National Guard (deployed after a delay that subsequent reviews investigated), and other federal agencies. Congress reconvened on the same evening, with the Senate resuming at 8:06 p.m. and the House at 9:02 p.m. The certification of Biden's election was completed at approximately 3:40 a.m. on January 7. The total certification delay was approximately six hours.

In the immediate aftermath, public condemnation of the breach was bipartisan and broad. Senator Mitch McConnell, on the Senate floor on the evening of January 6, called the rioters' conduct "criminal" and stated that "the United States Senate will not be intimidated." Senator Lindsey Graham (R-S.C.), a long-time Trump ally, said on the Senate floor that night, "Count me out. Enough is enough." Vice President Pence completed the certification. House Republicans split on the second impeachment that followed, with ten voting in favor of articles of impeachment against the President for "incitement of insurrection." The Senate impeachment trial, in February 2021, resulted in a 57-43 vote to convict — falling short of the 67 votes required for conviction. Seven Republican senators voted to convict (Burr, Cassidy, Collins, Murkowski, Romney, Sasse, Toomey), the largest number of Senators in either party voting against their President in any impeachment trial in American history.¹⁹

The Department of Justice prosecuted approximately 1,500 individuals for offenses related to the Capitol breach, ranging from misdemeanor trespassing to seditious conspiracy. The vast majority pleaded; many were convicted at trial. Members of the Oath Keepers and the Proud Boys, two militia-aligned organizations, were convicted of seditious conspiracy in 2022 and 2023, with leaders of both organizations sentenced to terms of 18 to 22 years. On January 20, 2025, his first day in office, President Trump issued pardons or commutations for nearly all of the approximately 1,500 individuals convicted of January 6-related offenses, including those convicted of violent crimes against police officers and those convicted of seditious conspiracy. The pardons were among the broadest mass pardons in American history.

Subsequent legal proceedings. Special Counsel Jack Smith, appointed by Attorney General Merrick Garland in November 2022, brought two federal indictments against former President Trump: a four-count indictment in the District of Columbia (August 2023) related to alleged conspiracy to obstruct certification of the 2020 election; and a 40-count indictment in the Southern District of Florida (June 2023) related to retention of classified documents at Mar-a-Lago after the end of his first term. The federal D.C. case was substantially affected by the Supreme Court's July 2024 ruling in Trump v. United States, which held that former presidents enjoy presumptive (and in some cases absolute) immunity for "official acts" performed during their tenure, and required that the lower court reconsider the indictment in light of the new doctrine. After Trump's victory in the November 2024 election, the Department of Justice in November 2024 moved to dismiss both federal cases, citing the Office of Legal Counsel's policy against prosecuting a sitting president. State cases — including the Fulton County, Georgia, indictment relating to alleged interference with state election officials — proceeded on different timelines, with mixed outcomes. The Manhattan District Attorney's case relating to alleged falsification of business records produced a guilty verdict on 34 counts in May 2024; sentencing was deferred and ultimately handled with no prison time after Trump's election victory.²⁰

The 2024 election. Donald J. Trump defeated Vice President Kamala Harris in the November 2024 presidential election by a margin of approximately 1.5 percentage points in the popular vote (49.8% to 48.3%) and 312-226 in the Electoral College. The election was administered by state and local officials, certified by state authorities, and certified by Congress on January 6, 2025, in a ceremony presided over by Vice President Harris. The transition was peaceful. Trump was inaugurated on January 20, 2025. The election outcome is not contested at the level of the data. Both Republican and Democratic election officials, the Cybersecurity and Infrastructure Security Agency, and independent observers reported that the 2024 election was administered without significant fraud or irregularity sufficient to affect outcomes.

The Trump-2 period. Since January 20, 2025, the Trump administration has taken a series of executive-branch actions that political scientists and legal scholars across the political spectrum have flagged as institutionally significant. These are stated below as institutional facts; their normative significance is contested.

• Schedule F revival and federal-workforce changes. A 2020 Trump-1 executive order had created Schedule F, a new classification for federal employees in policy-related positions that would have removed them from civil-service protections. The Biden administration revoked the order before its operative effect. Trump-2 reissued the order on January 20, 2025, with implementing guidance over subsequent months. Estimated affected positions are in the tens of thousands. The administration also offered "deferred resignation" programs to substantial portions of the federal workforce; estimates of departures are in the range of 100,000 to 200,000 federal employees as of late 2025.

• Inspector General firings. On January 24–25, 2025, the administration removed 17 Inspectors General from federal agencies in a single weekend without the 30-day notice to Congress that the relevant statute requires. The administration's position has been that the statute's notice requirement is constitutional only as a non-binding requirement on executive personnel actions. Litigation is ongoing as of the writing of this chapter.

• January 6 pardons. The mass pardons of January 20, 2025, are described above.

• Personnel actions at the Department of Justice and FBI. The administration removed several career prosecutors who had worked on January 6 cases; replaced the Director of the FBI; and signaled a willingness to investigate political opponents. The administration's Attorney General has stated publicly that the Department's mission includes investigation of perceived political bias in prior investigations. Some of these moves are routine inter-administration personnel changes; others depart from the post-Watergate norm of separation between presidential political agenda and Justice Department independence.

• National Guard deployments. The administration has deployed National Guard units (including federalized state National Guard, which requires a different legal authority than state-controlled deployment) in connection with civil unrest in several U.S. cities. The deployments have been the subject of ongoing litigation, with state governors of the affected states arguing in some cases that the deployments exceed constitutional authority.

• Calls for prosecution of political opponents. The President and senior administration officials have publicly called for the investigation or prosecution of named political opponents on a series of occasions. Some have been characterized as rhetorical; some have been followed by referrals or investigative actions. The tradition under post-Watergate norms was that presidents did not publicly call for the prosecution of named individuals.

• Tariff and emergency-powers actions. The administration has invoked the International Emergency Economic Powers Act (IEEPA) for a series of tariff actions against Canada, Mexico, China, and other trading partners. IEEPA's text on the scope of permissible action under emergency powers is contested in current litigation; the administration's position is that the powers are broader than the Biden and Obama administrations had treated them.

• Executive orders on agency authorities. The administration has issued executive orders that significantly restrict or restructure the operations of several agencies, including the Department of Education, the Environmental Protection Agency, and the Department of Health and Human Services. Several of these orders are subject to ongoing litigation under the Administrative Procedure Act and other statutes.

These actions are, individually and collectively, the most contested actions of any modern American administration in the eyes of legal scholars and political scientists. Whether they represent a difference of degree from prior administrations (which engaged in their own assertions of executive power) or a difference of kind is the central question of the analytical literature. The chapter takes that question up in the next section.

37.7 Difference of Degree, or Difference of Kind?

Begin with what is shared across most of the analytical literature: presidential power has expanded substantially over the long sweep of American history, particularly from the New Deal forward, and most modern administrations have engaged in moves their predecessors would have considered exceptional. Lincoln suspended habeas corpus during the Civil War (a controversial move that the Supreme Court would later say required congressional authorization); Wilson prosecuted political dissidents under the Espionage Act of 1917 (a record that includes the imprisonment of socialist leader Eugene V. Debs); Franklin Roosevelt interned approximately 120,000 Japanese-Americans by executive order in 1942 (later acknowledged as a profound civil-rights violation in the Civil Liberties Act of 1988); Lyndon Johnson significantly expanded the war in Vietnam under the Gulf of Tonkin Resolution; Nixon engaged in extensive abuses culminating in resignation; Reagan's administration secretly traded arms for hostages and funded Nicaraguan Contras in violation of congressional appropriations (Iran-Contra); George W. Bush authorized warrantless surveillance of U.S. persons (later challenged but partially codified); Obama issued DACA by executive action after Congress declined to pass the DREAM Act, and significantly expanded the use of armed drones for targeted killing of U.S. citizens abroad; Biden issued substantial student-loan forgiveness through agency authority, despite a Supreme Court ruling that found the original major proposal exceeded executive authority. Each of these is a real episode in American constitutional history; each has been criticized at the time and subsequently. The point is not that prior administrations were innocent. They were not.

The empirical question is whether the cluster of actions in the 2017–26 period is meaningfully different from the prior pattern. Several respected analytic positions exist. The chapter steel-mans four of them.

Position 1: "There is a meaningful asymmetry, and the trajectory is concerning"

The position associated with Levitsky and Ziblatt, with Mounk, with Ginsburg and Huq, with most political scientists who specialize in comparative democratization, and with a significant tradition of conservative-defector intellectuals (Liz Cheney, Bill Kristol, Adam Kinzinger, The Bulwark, the Niskanen Center). The argument runs as follows.

The comparative literature identifies four warning signs of democratic erosion: (a) rejection of (or weak commitment to) democratic rules of the game; (b) denial of the legitimacy of political opponents; (c) toleration or encouragement of violence; and (d) readiness to curtail civil liberties of opponents, including the media. By those criteria, the events from late 2020 through January 2025 — a sustained refusal to accept an electoral defeat, the "stolen election" framing, the Capitol breach and the subsequent broad pardons, the elevation of Jan-6 defendants in political rhetoric, and the public calls for prosecution of opponents — represent a constellation that the comparative literature has reliably flagged as concerning in other countries.

This position does not deny that prior administrations exceeded norms. It argues that the magnitude of recent norm violations exceeds the prior baseline; that the violations cluster on one side of the spectrum more than the other in the post-2016 period; and that the trajectory is one of escalation rather than stabilization. Levitsky and Ziblatt's Tyranny of the Minority (2023) marshals comparative data to argue that the United States is currently inside the same pattern that preceded democratic erosion in Hungary, Turkey, and other recent cases — and that the durability of American institutions, while real, should not be confused with invulnerability.

This position does not predict American democratic collapse. It argues that the warning signs warrant action by both elite and mass-public actors to defend institutional norms, that the burden of action falls heaviest on the side whose norm violations have been most acute, and that the long-term trajectory will depend on whether elite and mass-public actors choose to defend norms when their own party's interests are inconvenienced.

Position 2: "The institutional system is responding correctly; the alarmism is overstated"

The position associated with several scholars at the Federalist Society and Heritage Foundation, with some legal commentators at National Review and Reason, and with portions of the New Right intellectual tradition. The argument runs as follows.

American institutions are doing exactly what they were designed to do. In 2020, the courts ruled against the Trump campaign's election-fraud claims at every level. In 2021, the Senate held an impeachment trial. In 2024, the Supreme Court issued substantive rulings on presidential immunity that did not give the President carte blanche but did define the constitutional limits. In 2024, Trump won the election fairly and the transfer of power was peaceful. In 2025, the courts are continuing to rule on executive-branch actions, with mixed outcomes — some upheld, some struck down, some pending. This is constitutionalism working. The alarmism of the "democratic erosion" framing reflects, on this view, a partisan reaction to electoral outcomes rather than a clear-eyed assessment of institutional functioning.

This position emphasizes that prior administrations' executive overreach was equally or more concerning to its critics at the time, and that the bar for crying "democratic erosion" should be higher than "the policies of the administration in office offend me." It points to the 2024 election outcome as evidence that ordinary democratic processes are operating: voters voted, votes were counted, the result is being respected. It argues that the most likely outcome of the current period is a return to normal political competition, with the courts continuing to operate as a check, the next election continuing to be contested, and American democracy continuing to function.

A subtler version of this position — associated with some legal academics across the spectrum — emphasizes that the post-2016 period's most concerning judicial actions came from both sides: the Trump v. United States immunity ruling extends presidential power in ways many scholars find troubling, but court-expansion proposals from the other side are not less troubling, and a clear-eyed view sees the institutional pressures running in both directions.

Position 3: "The real democratic erosion is on the left, in the administrative state and in soft-speech-suppression"

The position associated with portions of the conservative legal academy, the New Right, and some libertarian-aligned commentators. The argument runs as follows.

The traditional definition of democratic erosion focuses on executive overreach, election denial, and political violence. These categories are real, but they are not the only categories. A different category — and on this view, the more empirically significant one for the United States in the 2020s — is the unaccountable expansion of administrative-state authority over political speech, economic activity, and educational and cultural institutions. The "Twitter Files" releases of 2022–23, on this view, documented a pattern of coordination between federal agencies (the FBI, CISA, the Surgeon General's office) and major social-media platforms in flagging and suppressing content during the 2020–22 period. The Supreme Court in Murthy v. Missouri (2024) declined to rule on the merits of these allegations, citing standing problems, but the underlying record is, on this view, evidence of soft-speech-suppression that escapes the standard erosion frameworks because it operates through private intermediaries rather than direct state action.

This position also emphasizes the institutional capture of universities, professional associations, major foundations, and parts of the federal civil service by ideological factions whose substantive policies and human-resources practices are, on this view, inconsistent with classical liberal-democratic norms. The literature here includes the work of Christopher Caldwell (The Age of Entitlement, 2020), Christopher Rufo (America's Cultural Revolution, 2023), and various essays in American Affairs and Compact. A more academic version of the argument is in Adrian Vermeule's Common Good Constitutionalism (2022), which argues that the modern administrative state has departed from constitutional first principles in ways more profound than any single executive's transgressions.

The chapter does not adjudicate this position. Like the others, it includes serious scholars and serious arguments. The empirical claims it makes — about the Twitter Files records, about university-policy patterns, about administrative-state scope — are themselves contested, with significant pushback in the rebuttal literature (including work by Yochai Benkler, Jonathan Rauch, and others arguing the Twitter Files materials have been substantially overstated). A serious civic education has to engage with this position, not dismiss it; and it has to engage with the rebuttals, not endorse the position uncritically.

Position 4: "It depends on which institutions you trust, and the underlying problem is institutional credibility"

A position that cuts across left-right but is most clearly articulated by institutionalist conservatives like Yuval Levin (A Time to Build, 2020; American Covenant, 2024), by some center-left commentators like Jonathan Rauch (The Constitution of Knowledge, 2021), and by parts of the heterodox tradition (Martin Gurri, The Revolt of the Public; Tyler Cowen). The argument runs as follows.

The deeper issue is not which administration is worse; it is that across the post-Cold War period, every institution that Americans depended on to mediate political disagreement — Congress, the courts, the press, the universities, the major churches, the public-health establishment, the intelligence community — has lost public credibility. Each loss is partially earned (institutions have made errors and been captured by partisan interests) and partially the product of partisan attacks that have not been adequately rebutted. The result is a society in which there is no shared mechanism for resolving contested empirical questions, no shared pool of institutions whose findings carry weight across the partisan divide, and consequently no way to resolve disputes short of contestation through the next election.

On this view, the question of "democratic erosion" is less productive than the question of "institutional re-building." Both parties have contributed to institutional decay through their hardball moves. Both parties' base voters now distrust the institutions associated with the other party's coalition. The way out, on this view, is not better partisan victories but slow, deliberate institutional rebuilding — investment in local civil society, in non-political voluntary associations, in the institutions that mediate disagreement, in re-establishing the credibility of expert bodies through more humble and transparent practice.

This position does not deny that some specific actors have been more aggressive than others. It argues that the focus on individual-actor accountability misses the structural problem and the structural fix.

How to engage these positions

Each of the four positions makes some claims that are empirically contestable and some that are normatively grounded. A serious reader of this chapter should be able to:

1. State the strongest version of each position in its own terms.
2. Identify the empirical claims each makes, and locate the data that bears on each claim.
3. Identify the value commitments each presupposes, and decide which value commitments you yourself hold.
4. Resist the temptation to consolidate all four positions into "the position my side holds." Each of the four positions has serious adherents who hold the position despite cross-pressures from their political coalitions. Liz Cheney is a Republican who holds Position 1. Some Federalist Society scholars hold Position 2. Yuval Levin and Jonathan Rauch hold something like Position 4 from different starting points. Position 3 is held by scholars across the conservative-libertarian spectrum but is not the position of every conservative. The four positions are not mapped one-to-one onto party affiliation, and the reader who treats them as if they were misses what is most useful about each.

The chapter does not adjudicate among the four positions. It is the reader's task to do so. The chapter's institutional commitment is to ensure that all four are presented in their strongest form, and that the reader has the analytical tools to engage them.

37.8 Comparative Perspective

What does the U.S. case look like alongside the comparative cases the literature uses? Five cases are most-cited in the post-2010 erosion literature; each is partially analogous to the U.S. case and partially disanalogous.

Hungary (Viktor Orbán, 2010–present). Orbán's Fidesz party won a two-thirds parliamentary majority in 2010 and used that majority to: rewrite the constitution; re-staff the Constitutional Court with party loyalists; restructure the electoral system to favor Fidesz; consolidate ownership of media in pro-government hands; restructure the judiciary to remove independent judges and reduce retirement ages to force out unsympathetic ones; restructure higher education (notably driving Central European University out of the country); and use government contracts and tax law to favor regime-aligned business interests. The pattern has been described by Lührmann and others as the most successful peacetime democratic erosion in a EU member state. By V-Dem's measures, Hungary moved from Liberal Democracy to Electoral Autocracy between 2010 and 2018. Freedom House classified Hungary as "Free" in 2010 and "Partly Free" by 2019. Hungary continues to hold competitive elections; Orbán has lost some local races and has faced organized opposition coalitions; however, the cumulative institutional changes have, on the comparative literature's assessment, tilted the playing field substantially.²¹

The U.S. case has partial analogies — sustained election denial, pressure on judicial independence, attacks on universities, executive-branch personnel changes — and partial disanalogies. The U.S. has not had a single party with a constitutional supermajority; the U.S. has not (yet) had a constitutional rewrite; the U.S. media is not state-controlled; the U.S. judiciary, despite being substantially reshaped by the 2017–20 confirmations, retains professional norms and includes judges of multiple ideological orientations whose decisions have substantively differed from administration positions. The U.S. case is not Hungary. Whether the trajectory is moving in the Hungarian direction is the contested question.

Poland (PiS, 2015–2023, then opposition victory). The Law and Justice (PiS) party, governing Poland from 2015 to 2023, engaged in a similar pattern of judicial restructuring, media consolidation, and institutional capture, though with less centralized power than Orbán. The European Union initiated Article 7 proceedings against Poland for rule-of-law concerns. In October 2023, an opposition coalition led by Donald Tusk won a parliamentary majority, and the new government has been engaged in a difficult, partial restoration of judicial independence and other institutions. Poland is a useful case because it shows that some erosion is reversible at the ballot box, even when significant institutional damage has been done.²²

Turkey (Erdoğan, 2003–present). Recep Tayyip Erdoğan and the AKP party, governing Turkey since 2003, have engaged in a longer and more comprehensive erosion. The 2016 coup attempt was followed by a state of emergency, mass purges of the judiciary and military, the restructuring of the constitution to give the president expanded powers (in a 2017 referendum), and the imprisonment of journalists, opposition politicians, and civil-society figures. Turkey moved from Electoral Democracy to Electoral Autocracy by V-Dem's measures around 2014–15 and has continued to move further away from liberal-democratic measures since. Turkey is a more advanced case than the United States; the question of relevance is whether the comparison illuminates a possible trajectory or is too distant in starting conditions to be informative.

Brazil (Bolsonaro, 2019–22). Jair Bolsonaro's presidency is partly relevant to the U.S. case: a populist-right president who declined to commit to accepting an election outcome (he disputed the integrity of Brazilian electronic voting machines without evidence), whose supporters stormed government buildings on January 8, 2023, in an event sometimes compared to January 6, 2021, and who has been the subject of subsequent legal proceedings. The Brazilian case is illuminating in part because the institutions held: Bolsonaro lost the 2022 election (narrowly to Luiz Inácio Lula da Silva), the transfer of power occurred, and prosecutions of Bolsonaro and his allies have proceeded under Brazilian law. The Brazilian case suggests that erosion can be resisted and that institutions can recover.

India (Narendra Modi, 2014–present). Modi's premiership has been associated with significant institutional changes: pressure on press freedom, the application of charges of "sedition" to opposition figures and journalists, the use of the Enforcement Directorate (a financial-crimes agency) against opposition politicians, restrictions on Muslim civil society, and changes to citizenship law. V-Dem reclassified India as "Electoral Autocracy" in 2021 — a controversial coding decision that the Indian government and many Indian commentators reject, and that other measurement frameworks have not adopted. India remains a more difficult case for the comparative literature because of its enormous size, federal complexity, and continuing electoral competitiveness. The Indian case may be less directly applicable to the United States than some of the others.

Israel (Netanyahu, 2022–25, judicial reform paused). In 2023, the Netanyahu government attempted a comprehensive judicial reform that would have curtailed the Supreme Court's authority to review legislative actions. Sustained street protests, including from military reservists and from major sections of the Israeli economy, forced the government to pause the reform. The October 7 attacks of 2023 and the subsequent war reshaped the political agenda. The Israeli case illustrates that civil society and street mobilization can constrain governments that pursue institutional change, and that exogenous events can substantially alter political trajectories.

The U.S. case is not identical to any of these cases. It is partially analogous to several. The strongest analogies are with Brazil (the January 6/January 8 parallel, the post-presidential legal proceedings) and with the early period of Polish PiS rule (judicial restructuring, executive-branch consolidation). The disanalogies include the U.S.'s federal structure (which puts substantial democratic operations outside federal-executive control), the size and diversity of the U.S. economy and civil society, the long-established traditions of judicial independence at the federal level, and the U.S.'s continuing two-party competition. The right reading of the comparative case material is not "the U.S. is becoming Hungary" or "the U.S. is robust because it is not Hungary." It is closer to: "the comparative literature describes a constellation of stress points whose presence, taken together, has predicted erosion in other contexts; the United States exhibits some but not all of those stress points; the institutional capacities to resist erosion in the U.S. are real but not unlimited; and the trajectory depends on choices that have not yet been made by elite and mass-public actors."

37.9 What Citizens Can Do

The Madisonian framework, restated for this context: the system relies on citizens supporting institutional norms even when their own party is in office. The Founders understood that ambition would always counteract ambition at the elite level only if mass-public actors held both parties accountable for institutional behavior. The most basic civic obligation in a system of separated powers is not partisan loyalty but institutional support — and that obligation falls hardest on the side currently holding power.

A short list, drawn from the comparative-democracy literature on what mass-public action has done in cases where erosion has been resisted:

Vote. Voting in elections (both general elections and primaries) is the most basic mechanism by which the public expresses approval or disapproval of institutional behavior. Turnout has consequences, including for the kinds of candidates who emerge from primary systems. Both parties are responsive (imperfectly, but meaningfully) to primary-election turnout patterns.

Hold your own party's leaders accountable. This is the harder ask. Cross-pressures are real; partisan loyalties are real; the cost of public disagreement with one's own coalition is real. But the comparative literature is clear: the strongest signal against erosion is not opposition-party criticism (which is expected and discounted) but same-party criticism (which is costly and meaningful). Liz Cheney, Adam Kinzinger, the seven Senate Republicans who voted to convict Trump in 2021, the Democratic state attorneys-general who blocked Biden-administration overreach in some cases, the federal judges of various ideological orientations who have ruled against administrations that appointed them — these are the operational examples. Most readers will not be in a position to play this role at the elite level. Most will, at some point, be in a position to play it at the local level: in a county-party meeting, a school-board race, a state-legislative primary, a workplace conversation.

Support civil society. Independent press, professional associations, religious institutions, civic organizations, philanthropies, neighborhood groups — the dense network of voluntary associations that Tocqueville observed in the 1830s and that Robert Putnam documented the decline of in Bowling Alone (2000). The comparative-democracy literature finds that countries with stronger civil society (more associational life, more independent media, more functioning local government) are more resistant to erosion. The implication for individual citizens is to invest time, attention, and money in civil society rather than only in partisan politics. Yuval Levin's A Time to Build (2020) makes this argument from a conservative starting point; Putnam's Bowling Alone makes it from a center-left starting point; the empirical convergence is striking.

Engage legal challenges. When administrations of either party take actions of contested legality, the path of legal challenge is open. Public-interest law firms across the spectrum (the ACLU, the Institute for Justice, the Foundation for Individual Rights and Expression, the Liberty Counsel, the Pacific Legal Foundation, Americans United for Separation of Church and State, and many others) bring lawsuits. State attorneys-general bring lawsuits. Private parties bring lawsuits. The federal courts have ruled against significant pieces of executive-branch action under both the Trump-1, Biden, and Trump-2 administrations. Litigation is slow and imperfect; it is also one of the constitutional system's primary mechanisms for resolving disputes about the limits of executive power.

Defend free expression, including the expression of those you disagree with. Free expression is the precondition for the institutional functioning of a democracy. The comparative literature finds that erosion is preceded and accompanied by the suppression of opposition voices. The defense of free expression is a non-partisan obligation — it requires, in different contexts, defending the speech of those whose substantive views one finds repellent, in order to defend the institutional capacity of the system to mediate disagreement at all.

Engage in peaceful protest. Protest is a constitutionally protected activity (First Amendment). Mass mobilization has historically been one of the most powerful ways civil society has constrained governments — the Civil Rights movement of the 1950s and 1960s; the anti-war movement of the 1960s and 1970s; the Tea Party in 2009–10; the Women's March of 2017; the racial-justice protests of 2020. The norm is that protest is non-violent, lawful, and oriented toward persuasion and political pressure, not toward intimidation or physical interference with the operation of institutions. The line between protected protest and unprotected illegal conduct is, in most cases, not difficult to identify in practice; the institutional and moral significance of staying on the right side of that line is substantial.

Recognize cross-coalition allies. This is the hardest ask. The defense of institutional norms, in nearly every comparable case in the comparative literature, has required coalitions that crossed partisan lines. The seven Republican senators who voted to convict in 2021 and the Democratic-led Senate that ratified that vote did, together, more than either could have done alone. The conservative legal-academic critics of Trump-1's election challenges (the Federalist Society members who opposed Texas v. Pennsylvania, the conservative judges who ruled against the Trump campaign in election cases) and the progressive litigators who sustained other challenges did, together, more than either could have done alone. Cross-coalition coordination on institutional questions is rare, costly for those who engage in it, and historically the best predictor of which cases of erosion have been resisted.

The list is incomplete. The point is that the work is not glamorous and is not new. It is the ordinary, slow, frustrating work of citizenship, performed by ordinary citizens in ordinary places. Theme 3 of this book — power flows to those who show up — applies here as elsewhere.

37.10 What This Chapter Does Not Resolve

This chapter does not predict the future of American democracy. It does not declare American democracy "in crisis" or "robust." It does not declare any one administration's actions to be the decisive turn. It does not declare any one analytical framework correct.

It does state institutional facts: V-Dem and Freedom House have measured a decline in the U.S. composite scores; the 2020 election was won by Biden, the 2024 by Trump, both fairly; January 6, 2021, occurred and is a category of event that does not have a precise equivalent in the modern Democratic Party's record; substantial pardons of January 6 defendants were issued in January 2025; the Trump-2 administration has taken executive-branch actions that legal scholars across the spectrum identify as institutionally significant; prior administrations of both parties also took actions that critics identified as institutionally significant; constitutional hardball is bipartisan in pattern and asymmetric in some specific aggregations of the data; the comparative-democracy literature flags certain warning signs that the U.S. case partially exhibits.

It then steel-mans four positions on what those facts mean. Position 1 holds that the asymmetric-erosion analysis is correct and the trajectory is concerning. Position 2 holds that the institutional system is responding correctly and the alarmism is overstated. Position 3 holds that the more empirically significant erosion is in the administrative state and in soft-speech-suppression. Position 4 holds that the deeper problem is institutional credibility across the board, which both parties have contributed to and neither has mechanisms to repair.

A serious reader of this chapter will leave with the analytical equipment to engage the question, not with a predetermined answer. That is the chapter's offer. The next chapter (Chapter 38, "The Future of American Democracy") takes up the proposed reforms — to the electoral system, to the courts, to the parties, to civil society — that have been proposed across the spectrum, and asks what each would require, what each would deliver, and what the trade-offs are. The chapters together (37 and 38) are the book's effort to give you, the citizen, the tools to think clearly about the most consequential institutional questions of your time.

The Founders made a bet. The bet was that ordinary people, given decent institutions and a serious civic education, are capable of self-government. The bet has not been resolved. It is being resolved in the choices being made now, by elected officials and by ordinary citizens, by Republicans and Democrats and independents, by those who are confident the system will hold and by those who are worried it will not. The chapter you have just read is one small part of the civic education side of that bet.

Chapter Summary

This chapter introduced democratic erosion as a distinct analytical concept (a gradual hollowing-out of institutions, distinct from democratic collapse), surveyed the V-Dem and Freedom House measurement frameworks and what they show about the United States, distinguished between formal constitutional rules and informal norms, applied the concept of constitutional hardball to a series of bipartisan examples spanning three decades, treated the 2020 election, January 6, 2021, the subsequent legal proceedings, and the Trump-2 administration's actions as institutional facts, steel-manned four distinct analytical positions on what those facts mean, drew on comparative cases (Hungary, Poland, Turkey, Brazil, India, Israel) to locate the U.S. case in cross-national perspective, and outlined the practices that the comparative-democracy literature associates with norm preservation. It does not adjudicate the contested question of whether American democracy is currently eroding, in what direction, or to what degree. That adjudication is the reader's task.

The next chapter takes up proposed reforms.