When a democracy falls, do the people know?
A simple four-category test applied to acts over the first 90 days of Trump-II shows autocracy rising in the US due primarily to violations of the rule of law, which many people don't perceive.
Many political scientists are raising alarms about the status of US democracy.
Steven Levitsky, the famed scholar and co-author of 2019’s How Democracies Die, recently said on NPR that the US is now classified as some version of an authoritarian state.
The long-running Bright Line Watch study of the health of US democracy recently reported that the US has dropped precipitously on its democratic metrics. See also V-DEM Democracy Report (p. 46) and the POLITY scores from the Center for Systemic Peace, both of which have downgraded US regime status.
Multiple sources are tracking anti-democratic actions by the Trump administration in an attempt to catalog aberrant acts, including The Impact Map of federal funding and workforce losses, the United States Disappeared Tracker of people federal government has disappeared, the Trump Tyranny Tracker of significant events of concerns, and the Trump Administration Litigation Tracker of lawsuits against the administration, among others.
Growing dissatisfaction amid toleration
Among the general public, awareness, concern, and protest are also rising. There have been massive multi-city protests, such as those on April 4 and 19, organized by a new pro-democracy grassroots organization called 50501.
It’s hard to say how many Americans are engaged in general protests—and those that are more targeted (but not unrelated) like the so-called “Tesla Takedown” events. But the US doesn’t seem to have yet reached a critical mass of public protests that might force government reforms. Research suggests that this happens when about 3.5 percent of the population is active in the public protests, which would come to about 12 million Americans. The US is probably still about 10 million shy of this.
For the most part, Americans are going about their daily lives in much the same way they did in 2024–they go to work, go to school, get groceries at grocery stores, shop online and in stores, go out to restaurants, sporting events, and shows, and gather with families. Unless you are among those whose families have been directly impacted by immigration actions or federal government cuts, it is not likely you have personally experienced anything that feels like “regime change.”
One reason many Americans do not appreciate rising authoritarianism is that threats to the rule of law are not necessarily felt in a diffuse way.
But that’s because, as scholars have shown, when modern democracies fail, they aren’t dramatic about it. We don’t often see military coups or acts of violence that overthrow an executive, anymore; rather, when modern democracies fail, it tends to be a gradual shift. Like the unalarmed frog in a pot of water that slowly rises in temperature until it is boiling–modern democratic states fail so subtly that many only notice after it’s gone.
Tom Pepinsky, a Cornell professor who writes about authoritarianism and other topics, says that:
Every day life in the modern authoritarian regime is, in this sense, boring and tolerable. It is not outrageous.
Depending on the level of restrictions, there is so much similarity in quotidian life between some authoritarian states and democratic states that some people may not appreciate the difference. So why then does it matter? Why are political scientists so alarmed about regime change in the US?
Identifying authoritarian actions
To answer that question, we have to be more specific about what we mean by living in an authoritarian regime. Demystifying authoritarianism is a good way to tune one’s personal radar for anti-democratic government actions.
I find the following simple test to be helpful to filter the cacophony of rapid change in the US. We need tools that allow us to differentiate government actions that are typical policy changes that we might oppose because of our personal political preferences from government actions that represent democratic backsliding. I use what I’ve been calling a “4-category alarm test,” in which I assess whether an action violates one of these critical characteristics of a free democratic society:
Actions that restrict people’s ability to vote or register to vote
Actions that interfere with free and fair elections.
Actions that violate civil liberties.
Actions that undermine the rule of law.
An action that violates multiple categories may be more damaging than one that threatens only one. Actions that severely degrade one category may still be very damaging.
So, how is the Trump-II administration doing on these measures? To get a rough idea of what the first 90-or-so days of Trump-II have been like in terms of democratic threat, my research assistant and I gathered all the events from Olga Lautman’s Trump Tyranny Tracker and applied a natural language inference model, which estimates the extent to which each of the above four threats is present in each event. We examined all 2,714 events from the Tracker, which is a good representation of the events in the first 90 days of Trump-II.
We use the algorithm to assign a “threat score” to each event for each of the four threat types. The threat score is equivalent to the probability that the threat is present. For example, a story about ICE and DOGE seeking access to Medicare data to track immigrants receives the following scores in our algorithm:
In addition to being a privacy violation (e.g., civil liberties), using government data to entrap people is a violation of the laws about how the government is supposed to use and guard its citizens’ personal data.
We find that, on average, there are 36 threats per day described in Tyranny Trackers’ events. But not all of the events on Tyranny Tracker are threatening; 20 percent of events have zero threats across the four categories; 67 percent of events have 1 threat, the modal outcome.
When we apply the algorithm to all the actions in the Tracker, take the average of the scores and sum them across all four categories, we find the following distribution over the course of the Trump-II administration so far.

We labeled a few peaks in the distribution where the Tracker data shows the most destructive days in terms of democratic backsliding, such as when Trump proposed ending birthright citizenship, closed USAID without congressional authority, and when the Supreme Court required the administration to stop deporting people without due process. Overall, the graph has a disturbing positive slope.
Which of the categories is most often violated,? We aggregate the scores by threat type and present them in a boxplot:

The boxplot shows dots for each action and its threat score. The colored box shows the middle, or interquartile, range of the most common scores for each category. Threats to the rule of law are much more frequent than any other type.
When we plot each threat category over time, we find the following distribution.
![A graph showing four lines. The x-axis is time [0-90], days of Trump. The y-axis is the average threat score. The top line is for "rule of law" around 0.6 to 0.8, then civil liberties around 0.2, elections around 0.1, and voting around 0. A graph showing four lines. The x-axis is time [0-90], days of Trump. The y-axis is the average threat score. The top line is for "rule of law" around 0.6 to 0.8, then civil liberties around 0.2, elections around 0.1, and voting around 0.](https://substackcdn.com/image/fetch/w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf000507-1792-4481-8184-235a2f7bc104_3532x1795.png)
Again, we see that threats to the rule of law are the most frequent and score the highest on the threat algorithm, with civil liberties a distant second. Some actions score in the low range in threat to elections, but so far, we see very few threats to voting rights.
Lawbreaking has selective effects
One reason many Americans may not appreciate rising authoritarianism is that threats to the rule of law are not necessarily felt in a diffuse way—they primarily threaten the people involved in whatever legal action or inaction is at issue in those cases.
Whether people perceive threats to civil liberties, elections, and voting rights depends on how generalized those actions are. If public radio loses federal funding or citizens are asked to prove their citizenship at their next voting opportunity, these would be widespread challenges to democratic best practices; however, if voting rights are narrowly challenged in small communities or particular populations, as the US has historically done, the democratic threats would be concentrated and therefore harder for the general public to perceive.
So maybe the people who are watching for violations of the rule of law are the ones most loudly sounding the alarms about the rise of authoritarianism. To my eye, and many political scientists’, the rule of law violations are sufficient to downgrade the US regime status to some version of authoritarianism. Consolidating power by overpowering the other branches of government is a necessary step to corrupting elections, compromising voting rights, and restricting civil liberties—in other words, compromising the rule of law is the first step in creating a fully autocratic state.
As the threat scores continue to build—and I expect they will until something happens to stop the current trend—more and more people will be affected and clue in to what’s happening in America. Until then, just because many Americans are not yet perceiving democratic decline, it does not mean it’s not happening.
The threats to voting rights are locally focused and not making headlines. Is it possible that your Tracking data source is too focused at the apex of the fascist movement and missing what is happening in their ground game?
https://www.brennancenter.org/issues/ensure-every-american-can-vote/voter-suppression
This is a fantastic piece. I'm saving it to assign in classes and am sharing it widely. Thanks for the work and summarizing where we're at so cogently. (As I'm often wandering around using the term 'authoritarian' and then having to explain what I mean and why I'm so concerned - it's great to have this to share to help some of that explanatory load.)