Category Archives: civic theory

rule of law means more than obeying laws: a richer vision to guide post-Trump reconstruction

The Trump Administration flouts the rule of law by denying its obligation to obey statutes and court rulings. On April 22, two TIME magazine reporters drew Trump’s attention to a portrait of John Adams that he had “put in” the White House. They quoted Adams to the effect that a republic is a government of laws, not men. Trump had never heard of this quote and said, “I wouldn’t agree with it 100%. We are a government where men are involved in the process of law, and ideally, you’re going to have honest men like me.”

This is the present crisis. However, rule of law means more than obeying explicit laws, and it had been weakening for many decades. Here I will present Trump’s current administration as the most recent stage in a disintegrative process that began in the 1960s.

Law should take the form of rules that are general, durable, transparent, coherent, chosen in legitimate processes, consistently applied, and anchored to principles. The principles that motivate laws may be good or bad, which is why rule of law is insufficient for justice. (We also need good laws). However, rule of law permits people to plan, it provides important forms of fairness, it frustrates outright corruption, and it makes government accountable. When rule of law prevails, but the actual laws are unsatisfactory, we can work to change them. When there is no rule of law, we have little recourse.

Generality, durability, transparency, legitimacy of process, coherence, consistent application, and principle are relative terms. It is impossible, for example, for laws to be perfectly general. They should not be so durable that they persist when circumstances change. Instead of exemplifying any single principle, laws may balance conflicting principles along with practical constraints.

Nevertheless, rule of law is a guiding ideal for republican government. More importantly, a good political system creates incentives for the players to promote rule of law. In contrast, a corrupt system rewards biased enforcement, ad hoc exceptions, back-room deals, short-term arrangements, impunity, and other violations of rule of law.

You can tell that 21st century America neglects rule of law from our dependence on executive orders instead of laws, regulatory rulings instead of statutes, and budget deals instead of legislation. As I’ve noted before, the federal government still addresses carbon emissions under the Clean Air Act of 1970 and social media under the Telecommunications Act of 1996. This is because Congress has been incapable of passing major statutes, liberal or conservative.

Trump lacks any compunction about governing by decree (often on the social media platform that he owns) and has signed fewer statutes than any modern predecessor in his first 100 days. His attitude is unprecedented, yet he represents the third of three stages of decline.

Theodore Lowi’s great book The End of Liberalism: The Second Republic of the United States (first edition, 1969) already described the first two stages.

The first stage was exemplified by some of John F. Kennedy’s speeches. JFK was neither original nor very influential, but he expressed the prevailing midcentury modernist view of US politics. Kennedy declared that Americans had reached consensus on the grand questions. Both national parties were ostensibly committed to Keynesian economics, Social Security, desegregation, and the Cold War. However, said Kennedy, issues had become complex, and therefore governance should be delegated to non-ideological agencies with lots of expert staff who could manage all the particular issues that would arise.

As the New Frontier turned into the Great Society, the executive branch vastly expanded, but Congress stopped passing landmark statutes, and power shifted to appropriations committees and budget negotiators, rulemakers in the executive branch, Senate confirmation hearings that determined who could serve as regulators and judges, and courts, not only in the judiciary but also within the executive branch. Donohue & McCabe (2021) write, “as of March 2017, more than 1,900 administrative law judges (ALJs) were serving in at least 27 adjudicatory bodies, with their specific roles and responsibilities reflecting those of the agencies and departments in which they were located.”

Meanwhile, the 1960s had exploded the Kennedy-era consensus about basic issues. Social movements of left and right mobilized, competing to change society through the expanded federal government. From the 1960s through the Biden Administration, urgent debates roiled civil society, but the mechanisms of government remained negotiation and regulation rather than lawmaking.

For Lowi, the Occupational Safety and Health Act (OSHA) of 1970 exemplified this shift. Congress did write and pass OSHA, but “it did not attempt by law to identify a single specific evil that the regulatory agency was to seek to minimize or eliminate.” Instead, Congress vaguely endorsed the idea that, “so far as is possible every working man and woman in the nation [shall have] safe and healthful working conditions.” Congress gave the Department of Labor the power to issue actual regulations, subject to constant revision and negotiation, some of it before the Occupational Safety and Health Review Commission, which is a tribunal in the executive branch. This is not rule of law.

One result is that social movements have usually broken like waves on the shoals of the administrative state, leaving lots of small and inconsistent regulatory actions to reflect their ideals. The women’s movement, the gay liberation movement, and the Movement for Black Lives made discernible impressions on executive branch policies without enacting major laws. A side-effect is that social movements now benefit more from expertise inside the Beltway than from grassroots mobilization.

The third stage is Trump’s. Until he won office, a system that had neglected rule of law was nevertheless, in my opinion, usually used for benign purposes, at least for domestic policies outside of some aspects of criminal law. But this system was waiting to be hijacked by someone without principles. This is what we observe right now.

As Trump’s popularity plummets, the odds of a post-Trump reconstructive period are rising. We should not be thinking about how to restore the processes of 2022 (or 1990) but how to revive rule of law, properly understood.

For me, the three main strategies would be:

  1. expand the capacity of Congress to legislate;
  2. restrict the discretion of the president and executive branch; and
  3. codify the procedures of the administrative agencies and the rights of the civil service so that these become appropriate and coherent.

These strategies must be accomplished together, because, for example, to restrict administrative agencies without enabling Congress to legislate will just hamper government.

More specifically, I would favor: substantially more funding and staffing for congressional offices and committees; state-level electoral reforms, such as ranked-choice voting, which may encourage members of Congress to legislate instead of grandstanding; court rulings or (if necessary) a constitutional amendment clarifying the president’s obligation to execute statutes and making that obligation enforceable; substantial reforms of administrative law and the civil service; a general shift to taxing-and-spending instead of regulation to accomplish progressive goals; and legal repercussions for the Trump appointees who are currently violating laws.


See also: beyond Chevron; 16 colliding forces that create our moment; on the Deep State, the administrative state, and the civil service; and on government versus governance, or the rule of law versus pragmatism (2012).

remarks on partisanship

(Columbus) These are my notes for a talk today at Ohio State University on the assigned topic of “civic partisanship”:

If “partisanship” means active membership in a political party, then it is desirable. Parties can be worthy components of our civil society, especially when parties are internally diverse and meaningfully organized at the local and state levels. However, participation in party organizations is very rare today. Parties have become labels for entrepreneurial politicians, and sometimes for voters, rather than organizations that do anything at the local, state, or national level.

For example, in my state of Massachusetts, which is heavily Democratic, the state Democratic Party has 6 employees and an annual budget of $7 million in an election year. In her uncompetitive reelection race the same year, one candidate, Elizabeth Warren, spent $28 million. Most people who volunteer in politics will work for candidates, not a party.

If a political party is not an organization but a label that people attribute to themselves and others, then partisanship is at least somewhat problematic. It discourages deliberation and cooperation with people who hold a different party label.

First, partisanship can encourage hostility. Affective polarization means disliking people of the opposite party. It has risen symmetrically for Democrats and Republicans to alarming levels. In 2020, the ANES asked people to rate the parties on a 10-point scale. Giving the other party a zero score suggests that one is affectively polarized. By that standard, just eight percent of Democrats and five percent of Republicans were affectively polarized in 1978. These rates rose steadily in both parties during the 2000s, reaching 48% of Republicans and 39% of Democrats in 2020. That year, more than half of Republicans over the age of 60 rated Democrats at zero.

Note that youth were not especially polarized compared to older people. About 38% of people under 30 who identified as Republicans or Democrats rated the opposite party at zero. So if affective polarization is a problem, it is not especially a youth problem, and it’s hard to see K12 or college education as the main solution.

Second, even if it’s not hostile, partisanship can replace independent thinking about issues. There is no a priori reason that beliefs about Ukraine, vaccination, tariffs, abortion, and immigration should cluster in two ways, labeled Democratic and Republican. People should combine these issues in diverse ways. But there is evidence that many people put their beliefs together in packages based on cues from party elites.

That would be OK if it represented a valid division of labor. We could imagine that responsible and accountable professional leaders think carefully about issues and propose combinations of beliefs to busy citizens, who benefit from the professionals’ guidance. However, the influence of party leaders is problematic if they are self-interested and unaccountable, which is how I would broadly characterize elites today. Besides, in a two-party system, the choice of elites is badly constrained.

Third, when partisan labels are associated with left and right, this encourages a mental model in which everyone can be located on a spectrum. The very idea of polarization presumes that the left and right are located far apart or are moving further apart. This is a metaphor, not a fact, and it is misleading. People actually hold many beliefs that are more or less connected to each other with reasons. For example, I support Ukraine because I oppose authoritarianism. That is a pair of connected beliefs. Combinations of beliefs and reasons form networks. In any given group, individuals’ networks prove diverse when you map them. I have found that people who identify with the same party and leaders have unique networks.

When we think of people as polarized, we erase their individual thinking, which, in turn, discourages deliberation. There can be a vicious cycle in which we describe people as polarized, ignoring their uniqueness, which encourages them to become less individual and critical.

These points imply that we should teach students to appreciate actual parties but to be suspicious of partisan labels as heuristics. I also want to raise one other issue related to partisanship and civic education.

We have a civic religion in the USA, whose scripture is the Constitution. It is often used as the outline for studying government and politics. But the US Constitution does not mention parties, and its authors saw parties as grave threats to republican government. In a curriculum shaped by the text of the Constitution, political parties belong under the First Amendment as associations. This is misleading because they are integral to the political system.

What’s more, parties may be fatal to a constitution that establishes a presidential republic. In 1990, Juan Linz observed that every presidential republic except the USA had failed because the president sooner or later came into conflict with the legislature and was either defeated or became an authoritarian. One explanation of the survival of the US Constitution is that our two parties long encompassed opposing factions, notably white supremacist southern Democrats and progressive northern Democrats. Therefore, presidents were able to govern like prime ministers, assembling majority coalitions in Congress. That option ended during the Clinton Administration, when the parties sorted. Arguably, the Linzian nightmare has since played out.

94% of the time that the government has been shut down because of a conflict between the president and Congress has occurred since 1995. Three out of four presidential impeachments have taken place since then, but they have had no consequences for the president. Presidents of both parties have governed via executive order. Most recently, Trump has signed fewer laws but issued more consequential executive orders than any predecessor in the first 100 days.

If Linz’ theory is playing out, then it is political miseducation to teach students that the Constitution is an excellent design that, among other things, allows parties to flourish as voluntary associations. Perhaps the Constitution is, after all, a suicide pact. Students should at least be able to wrestle with that possibility. That would mean, not so much criticizing partisan attitudes or habits of thinking, but critically assessing a constitutional order that cannot handle parties as we know them.


See also: affective partisanship and young people; People are not Points in Space; the Constitution is crumbling; the relevance of American civil religion to K-12 education; and putting the constitution in its place

building a democracy helpdesk

This one-minute video introduces a project that Tufts engineering faculty and students and I have begun, with Better Together America and some pro bono advice from the Harvard Law School Transactional Law Clinics. In essence, we are trying to improve Americans’ know-how for launching and sustaining organizations, on the theory that civic organizations preserve democracy.

In the short length of this video, I don’t quite make the case that declining membership is a cause of declining trust in institutions. One piece of evidence (not in the video) is a statistical model that uses American National Election Study 2020 data to predict whether people will agree with this sentence: “Much of what people hear in schools and the media are lies designed to keep people from learning the real truth about those in power.”

In my model, education, age, and ideology are unrelated to how people answer this question. Women and white people are slightly more trusting. However, dwarfing those relationships is the role of civic engagement. People who say that they have recently worked with others to deal with an issue facing their community are far less likely to believe that schools or the media routinely lie.

Other measures of civic engagement (such as volunteering) also predict trust; and civic engagement predicts other liberal and democratic dispositions in addition to trust.

I believe that one of the obstacles to broader civic engagement is a lack of nuts-and-bolts knowledge. Therefore, helping people to form and sustain groups will strengthen civil society, which alone can save democracy.

See also: to restore trust in schools and media, engage people in civic life; tools people need to preserve and strengthen democracy, 16 colliding forces that create our moment, etc.

important findings about the persuasive power of facts

There is a huge body of research that suggests that people are not very susceptible to good arguments. Apparently, we believe things for unexamined reasons, cherry-pick evidence to support our intuitive beliefs, and minimize the significance of inconvenient evidence.

These findings contribute to a general skepticism about people’s capacity for democracy, and I fear that this skepticism is self-reinforcing. If we presume that humans cannot reason well, why would we try to build institutions that promote reasoning? Only half jokingly, I sometimes say that the theme of current social science is: people are stupid and they hate each other.

But I also argue that at least some of this research employs methods that are biased against discovering rational thought. In particular, if you ask random samples of people disconnected survey questions that interest you (not them) and then use techniques such as factor analysis to find latent patterns, you will, indeed, often discover that people are stupid and hate each other. More prosaically, you will develop scales for latent variables like knowledge or tolerance that yield poor scores. But such methods may overlook the idiosyncratic ways that reasons influence individuals on the topics that matter to them.

Of all people, those who believe in false conspiracy theories are generally seen as the least susceptible to good reasons; and previous efforts to convince them have often failed. However, in a 2024 Science article, Thomas H. Costello, Gordon Pennycook, David G. Rand report results of an intervention that substantially reduced people’s commitment to conspiracy theories, not only in the short term, but also two months later.

In this study, holders of conspiracy theories wrote about why they held their beliefs, and then an AI bot held a conversation with them in which it supplied reliable information directly relevant to the specific factual premises of each respondent. For instance, if a person believed that 9/11 was an “inside job” because Building 7 collapsed even though no plane hit it (see Wood and Douglas 2013), the AI might provide engineering information about Building 7. Many people were persuaded.

These results are consistent with a study of conversations with canvassers who succeeded in persuading many voters “by listening for individual voters’ … moral values and then tailoring their appeals to those moral values” (Kalla, Levine, A. S., & Broockman 2022). The two studies differ in that one used people and the other, an AI bot; and one emphasized facts while the other focused on values. But both results point to a model in which each person holds various beliefs that are more-or-less connected to other beliefs as reasons, forming a network. Beliefs may be normative or empirical–they function very similarly. Discourse involves stating one’s beliefs and their connections to other beliefs that serve as premises or implications.

People actually do a lot of this and are relatively good at assessing the rigor of such conversations when they observe them (Mercier and Sperber 2017). However, many of our methods are biased against discovering such reasoning (Levine 2024a and Levine 2024b), leaving us with the mistaken impression that we are a bunch of idiots incapable of self-governance.


Sources: Costello, T. H., Pennycook, G., & Rand, D. G. (2024). Durably reducing conspiracy beliefs through dialogues with AI. Science385(6714); Wood MJ, Douglas KM. “What about building 7?” A social psychological study of online discussion of 9/11 conspiracy theories. Front Psychol. 2013 Jul 8;4:409; Kalla, J. L., Levine, A. S., & Broockman, D. E. (2022). Personalizing moral reframing in interpersonal conversation: A field experiment. The Journal of Politics84(2), 1239-1243; Mercier, H. & Sperber D, The Enigma of Reason (Harvard University Press 2017; Levine, P. (2024a). People are not Points in Space: Network Models of Beliefs and Discussions. Critical Review, 1–27 (2024a), and Levine, P. (2024v). Mapping ideologies as networks of ideas. Journal of Political Ideologies29(3), 464-491.

how markets “think” about politics

As I write, US stocks are plunging. I have no idea what will be happening by the time you read this post. However, stepping back from the moment, what does it mean that Wall Street indexes rose after Trump won the 2024 election but fell last week? Or that Ukrainian government bond prices rose from October 2024 until last week and then fell rapidly?

One view is that markets have wisdom–or at least predictive value–because they aggregate information from many people. Investors think critically and rigorously because their money is on the line. The recent trends make sense on their face and confirm that markets are rational.

A different view is that capitalism involves a class struggle, and capital markets rise when the upper classes expect their interests to prevail. This model has no trouble explaining why business leaders, including registered Democrats, would tell Steven Rattner that they like Trump. They were not predicting prosperity for all but expecting to profit for themselves.

I would endorse a third model. Friedrich Hayek had a genuine insight: individuals have limited cognitive capacity and diverse motives. Therefore, individuals cannot reliably assess whole societies, let alone predict the future of anything large-scale. However, says Hayek, within our own domains of experience and expertise, we can reasonably predict specific prices. After a tough spring, farmers will expect the price of wheat to rise.

Prices allow us to plan efficiently. Many people do not seek to maximize wealth but to accomplish something else, such as holding onto a valued job or retiring soon. Nevertheless, the result of all their private planning is a market that is–in certain respects–efficient.

However, markets also create opportunities to profit by correctly predicting the large-scale situation. In turn, such predictions require assessing the present. For example, to guess how the US economy will fare over the next four years, it’s necessary to evaluate Donald Trump as a leader. One can buy bonds and other securities partly on the basis of such predictions. In this way, an accurate evaluation of Trump could pay off financially.

But Hayek’s defense of markets would not encourage us to trust the aggregate results of such thinking. Just because many people trade securities, it does not follow that their overall understanding of the present or their predictions for the society as a whole are reliable.

On the contrary, each participant in a market who tries to predict how a whole economy or country will perform is subject to the same cognitive limitations that–according to Hayek–beset us as voters and policymakers.

Markets do respond intelligibly to news. Wall Street indices fall every time Trump announces tariffs and rise whenever he seems to back off. But these changes are not predictive. In fact, we can easily predict market shifts as soon as we know what Trump says. The market adds little new information.

It’s true that putting money on the line gives an individual a motivation to think rigorously and critically. But motivations do not solve cognitive limitations. The businessmen who confided in Rattner said that they didn’t like “woke stuff” under Biden. Such feelings should not directly influence their market behavior under Trump. Nevertheless, their hostility to “woke stuff” could affect their stock trades by influencing their moods or by leading them to consume news and information that is tilted in favor of Trump. As cognitively limited creatures, we must rely on limited sources and a priori models–also known as ideologies.

In recent months, CEOs reported rising confidence in the economy, while consumers’ confidence slipped. A closer look at consumer confidence reveals that it fell by 28 points among Democrats but rose by 32.8 points among Republicans between January and February. So we can compare three changing predictions: those of corporate bosses, Democrats as consumers, and Republicans as consumers. Why do the Democrats diverge from the CEOs and the average Republicans?

  • The CEOs tend to have different values from the Democratic consumers. If everyone agreed that Trump’s tax policies will boost corporate profits but hurt the environment, CEOs would be more positive than representative Democrats.
  • The CEOs have different information from Democratic consumers. They are awash in data about their own balance sheets, plus business-oriented news. Democratic consumers are seeing negative assessments of Trump.
  • The CEO’s and the Democrats probably hold different mental models of such fundamental issues as the role of government and businesses in our society. Everyone holds such models, without which we cannot absorb new information.
  • Partisan identity is working as a powerful heuristic. Americans are using the party of the incumbent president to predict the economy. This may be unwise, but human beings must use heuristics, and a party label does convey relevant information if you combine it with a model of the society.
  • Some people act performatively. I would probably answer almost any survey question about Trump in a way that made him look bad, even if I didn’t completely believe the literal truth of my response. Some may even buy financial instruments to make a point–witness the popularity of Trump’s cryptocurrency.
  • Finally, the information that people absorb may reflect political agendas. Rupert Murdoch, Jeff Bezos, and other media barons want to affect public opinion, although their impact is uneven because news consumers are sorted ideologically.

This is not a simple model, but it does have a simple core. It is methodologically individualist, presuming that the decision-makers are human beings rather than classes or other abstractions. Regardless of their interests and social positions, these individuals are cognitively constrained and not primarily concerned with assessing the whole society. When they do make general assessments and predictions, these decisions reflect their mental models (which, in turn, often reflect their social positions), limited information, and concrete issues that are salient for them at the time. As a result, markets respond intelligibly to widely reported breaking news but have little predictive value.

See also: The truth in Hayek; making our models explicit; social education as learning to improve models; how intuitions relate to reasons: a social approach; etc.