Martin Dimitrov on ‘Dictatorship and Information’ and how the CCP stays in power

Politics & Current Affairs

A scholar of communist and post-communist regimes talks to Christopher Marquis about his latest book.

A screenshot from

The “dictator’s dilemma” arises from the incapacity of authoritarian leaders to calibrate repression and concessions because they cannot or are not willing to access information about elite dissatisfaction or popular discontent.

I recently talked to Martin Dimitrov about this issue in the context of his new book, Dictatorship and Information: Authoritarian Regime Resilience in Communist Europe and China, published by Oxford University Press in 2023. Martin is a professor of political science at Tulane University. He is also an associate editor at the journal Problems of Post-Communism, and an associate editor at the Journal of Asian Studies.

We discussed the various mechanisms and channels of information the Chinese Communist Party leaders use that enable them to effectively retain power. Martin explained how state security, the Communist Party, and classified internal media publications can help single-party communist regimes solve the dictator’s dilemma, which in turn can prolong regime lifespan, although obviously it cannot ensure its indefinite survival.

Regarding protests, we discussed the recent COVID-zero protests and A4 movement in China, with Martin pointing out that COVID-zero protests are an exception, as usually protests in China do not result in accommodation, negotiation, and compromise.

Recent advances of technology have also affected information collection methods. For example, with regard to petitioning, I was surprised to learn that despite technology making it easier to make complaints about government organizations online, the rate of petitioning continues to go down in China, which indicates the decreasing level of responsiveness from the Chinese government.

Dictatorship and Information and our conversation included a huge range of other related topics.

Christopher Marquis: To start off, maybe just describe the core messages and arguments of your book. There’s a puzzle at the core of your book that you’re trying to resolve and understand. If you could also give us some details on that, that’d be great.

Martin Dimitrov: Yes, I’d be happy to. The book is engaged with an argument that is very prominent in political science and in the field of authoritarianism, and the argument is that autocrats cannot solve what is called the “dictator’s dilemma.” The dictator’s dilemma involves making decisions about deploying concessions in light of the available information about levels of discontent. According to this, because dictators don’t have this information, they miscalculate, they overuse repression, and they under allocate concessions. And as a result of these miscalculations, they are faced with surprise eruptions of popular discontent, and the empirical expectation from this theory is that autocracy will be short lived.

But what I’m focusing on in this book is a subtype of authoritarian regimes, namely, the single-party communist regimes which enjoy the longest average longevity among any authoritarian regime type. This longevity of the Commons regimes, I argue, suggests that they have found ways to mitigate or potentially solve the dictator’s dilemma.

I go through some of the hypotheses outlined in the recent literature on how the dictator’s dilemma can be solved. These arguments are that authoritarian regimes, like China, encourage competitive elections, they encourage protests, or they encourage media liberalization as ways of generating information about popular discontent. And what I argue is that those three channels, of course, provide that information, but they provide it to two sets of audiences. They provide it both to the autocrats, and they provide it to the general public, so the general public can use that information to facilitate collective action and to facilitate rebelling against the regime. From the autocrats’ point of view, I argue these three channels– competitive elections, protests, and liberalized media–are sub optimal.

I go into this very long discussion — nearly 500 pages — of the channels that autocrats themselves internally value, and that they foster over the lifespan of Communist regimes. And the message is that the dictator’s dilemma can be solved using state security, the Communist Party, and internal media, and that solving the dictator’s dilemma prolongs the lifespan of Communist regimes, but cannot ensure their indefinite survival.

Christopher Marquis: I would also ask you about protests specifically and am especially interested in your thoughts on the anti-COVID protests. The A4 movement resulted in this very quick turnaround of the situation. China opened up very quickly. What’s your sense about that and how does the regime process information like protests? Because it’s a fine line. If you give too much, then people will protest more, and it’s a negative spiral. But if you don’t respond, then that also could create a negative spiral. So it’s a very delicate balance.

Martin Dimitrov: Absolutely. It’s a delicate balance. One of the central arguments of the book is that the Chinese government, or the government of any authoritarian regime, for that matter, does not want protests, and I think this holds for the A4 revolution — the anti-COVID protests that engulfed China in December of last year. Now, once it has protest, the question is, what does the government do. Of course, there are two options. One is to repress the protesters; another one is to give concessions. In this case what the protesters wanted was an end to zero COVID. And of course, we cannot definitively prove that the protests and the end of zero COVID are linked. But one imagines they are because, as you yourself mentioned, zero COVID ended very soon after the protests. So, the government in this case responded to what the protesters wanted. But it didn’t need these protests to figure out the zero COVID had to end, it already knew that the zero-COVID policies were highly unpopular.

The protests gave it a pretext for ending the policy. It could show that it is responsive to popular demands or popular preferences. But what’s interesting for me is that even though zero COVID ended, according to reports that we have seen in the West, a number of the leaders of these protests were arrested afterwards. They were detained, at least, and we don’t know whether they were in prison because of how the judicial system works, but they were detained.

This, I feel, is further evidence that the Chinese government does not want protest, and the protests were blamed on hostile foreign forces. I do feel that although protests reveal information, the Chinese government and the governments of other communist regimes have more efficient and less dangerous ways in terms of regime stability for compiling that information.

Christopher Marquis: Maybe you can say a little bit more about social media. The government, of course, monitors those channels for informational purposes, but it doesn’t want to encourage or allow free debate on the media. Because there is the ability to censor and suppress social media at a pretty rapid rate. It seems this is useful to gather information too.

Martin Dimitrov: What the government has is its own source for media content, which are the internal media. The internal media is classified journalism. The journalists, who are monitoring the social media, are engaging in other types of reporting. Some old-fashioned journalistic reporting as well, and they are assessing popular discontent. They generate reports, but these reports don’t circulate publicly. These are internal reports that move up the chain, and some of them reach all the way up to the top of the leadership. Internal journalism is one of the three pillars of the information gathering system that autocrats foster. The other two are state security – the secret police apparatus. And in addition to state security/secret police, the Communist Party itself. The Communist party is a vast information gathering machine. It’s also the recipient of information that is gathered by the secret police, and by the internal media, by these classified journalistic reports.

Christopher Marquis: You also discuss the petitioning system, which, of course, dates back a thousand years. What is the role that plays in the information gathering and how has it evolved?

Martin Dimitrov: It is really more than 2,000 years, according to some Chinese scholars. The earliest examples of petitioning go back to the Han dynasty. I’m not a historian but certainly it has a very long history. Petitions are, of course, an important channel for information collection. These are crucial channels, because citizens’ petitions or complaints is a form of voluntary provision of information to the regime. The book spends a lot of time distinguishing between involuntary extraction and voluntary provision. It argues that involuntary extraction is prone to error because citizens are oftentimes aware that they are being monitored, and they may strategically dissimulate because they know that they’re being monitored.

When citizens complain, they reveal what they want, and oftentimes what they want is welfare. They want to be paid on time, they want the provision of various goods and services. Those are actual concerns that citizens have, and they don’t lie about them. An important condition that facilitates petitioning is high level of responsiveness, and when responsiveness is low, as it currently is in China, the rate of petitioning goes down, and the rate of protests goes up. What we’ve had in China for quite some time is declining levels of petitioning and increasing levels of protests. And you know where I stand on protesters. There are potential negative side effects, where the dissatisfaction of an individual or a small group of citizens becomes widely known, and it can facilitate further anti-regime mobilization. Petitioning is certainly something that the Chinese Communist Party promotes in contrast to protest, which it actively discourages.

Christopher Marquis: How does technology further influence the mix of feedback mechanisms? Obviously petitioning is much easier online. Probably you don’t have to go to the petition office, or even mail a letter. You can do it online. Then it’s also, I assume, much more difficult to organize some mass type of event. And I’d be interested to hear how advances in technology have influenced information collection mechanisms.

Martin Dimitrov: This is a question of central importance, Chris. There is this sense, and it’s a very widespread sense, that we should expect the number of petitions to go up because of the ease of online petitioning. Citizens, of course, still have the option in China today of writing a letter to the authorities, and some of them do. They also have the option of going in person to the government offices. Some of them do. But of course, most petitions today– over 50% in most provinces and some provinces as high as 80% of the petitions–are submitted online, either through an online form or through web chat or some other cell-phone enabled mechanism. However, despite the fact that most petitions are electronically submitted today, the rate of petitioning continues to go down. That is something that is not appreciated in the China literature. And it’s buried in one of the chapters of my book. This shows that citizens continue to feel that the rate of responsiveness is quite low. And what this indicates for me is that the citizens are rational individuals, and, given the low level of responsiveness, they are less likely to launch these petitions.

Now, the other issue is that indeed you’re quite right. In the second part of your question. which is that protesting has become more difficult because of the ubiquity of surveillance. I mean there are cameras everywhere. Citizens are monitored in terms of the physical movement around cities and rural areas, and of course, their online communications are also monitored by government.

I did an analysis of a large corpus of protests with one of my former PhD students, Zhu Zhang. We analyze 65,000 protests in China, which is the largest corpus of protests in the literature. This data is crowdsourced protest data and it extends from 2013 to 2016. The person who was allowing for this data to be submitted was detained, and eventually he was sentenced on a charge of provoking quarrels. The data access stopped in 2016. But we do already have the initial couple of years of the new administration.

What we found is that the police are present in the majority of these protests. The standard understanding of protests, which is based on literature from the nineties, is that protests, even if they’re not encouraged, are tolerated. And there is this government attitude of negotiation, accommodation, being kind essentially to protesters. We certainly didn’t find that in this massive corpus of protests that we analyzed. We found that cases of negotiation compromise are very rare. And we did find that in contrast to what literature argues, in most cases you do have the police presence there.

But what we didn’t know is the number of protests that never even materialized, because the police would prevent citizens from even getting to a physical locality. We don’t have data on that. We know this happens. A lot of protests never materialize because the police are monitoring people’s physical movements around the city and their WeChat communications. If a protest takes place, the police emerge. They appear very quickly there, and most of the time they don’t use brute force to disperse the protesters, but they’re actively working to disperse the protests. And in over 10% of the cases they do use brute force. But most of the time, they are just physically encouraging citizens to disperse. You are absolutely right. It’s harder to have a protest, and it’s harder for that protest to result in accommodation and negotiation and compromise. The zero-COVID protests are an exception that proves the rule, because that policy was repealed. But the protest organizers were detained.

Christopher Marquis: Can we shift gears a bit to consider the non-China cases and research in your book, which really helps with understanding generalizability of your model versus what is China specific. First, I want to ask you is around using Bulgaria as a comparison case. As you mentioned, the book is long – about 500 pages. So I must admit I focused much more on the theory and on the methods, and the China chapters. I focused less on Bulgaria. But I think it’s important to think about why you chose a comparison case to begin with. And then why, specifically, Bulgaria. Could you say a little bit about that, and your design and selection of cases, and also generalizability?

Martin Dimitrov: Of course. You and I and many others are interested in China. China is a country of huge consequence for the world today. But if we’re interested in the evolution of information gathering systems in a communist regime, we need to know how those systems have evolved in a regime that has completed its life course. So, a regime that emerged, developed, and then ceased to exist. And I feel that given that China remains a single-party communist regime, and will remain one for the foreseeable future, it needed to be compared with another regime that no longer exists in order for us to be able to appreciate the full course of these information collection systems. In terms of comparisons, I felt that I needed to match the case on certain characteristics of China. First, I needed a case that is a unitary state, like China. Not all communist regimes are unitary. The Soviet Union was a federal regime, as was Yugoslavia, as was Czechoslovakia, for instance, so those already excluded. Second, I also wanted a regime that did not have Soviet troops on its soil — we oftentimes think that all East European regimes have Soviet troops. Czechoslovakia did, as did Hungary and Poland. But Bulgaria did not. And of course, China didn’t have Soviet troops on its soil.

What the absence of Soviet troops on one’s soil means is that it just heightens the importance of handling the information problem on your own. The Soviet troops provide a shield. They provide a guarantee that if you cannot assess the level of discontent, you have the Soviets to protect you. Third, I was interested in countries that have sizeable and territorially concentrated ethnoreligious and ethnolinguistic minorities. As we know, China has that feature. Bulgaria had a 10% Turkish minority which was ethnically, linguistically and religiously different from the orthodox Christian ethnic majority. The fourth and final criterion is a decline in levels of repression. Both Bulgaria and China experience that. When we take into account all of these features that I was interested in, Bulgaria emerged as the top candidate using these criteria.

And of course, I expected that I would be writing a book about extraordinary differences between these two regimes, because of the differences in size and the differences in history. There are also differences in bureaucratic capacity which actually are in Bulgaria’s favor. The way that I developed this bureaucratic capacity index, China had limited bureaucratic capacity for reasons that I discussed in chapter 1. But what they had in common is that they emerge within five years of each other. For the information problem in these two regimes, despite their vast differences in history, levels of social economic development and state capacity, the solutions are quite similar. I highlight these parallels in chapters 3 and 4, which are about the origins of information gathering, and then, later on, I argue that the two regimes diverge because what we have in Eastern Europe is the emergence of a Socialist welfare state. In China, it’s much more limited, this welfare state.

The other developments that occur in China, like the Great Leap Forward and the Cultural Revolution. There is a divergence between China and Bulgaria here. But then they converge again in 1988, when they’re both experiencing massive levels of social discontent that eventually expresses itself as the 1989 protests in China and Eastern Europe. In Eastern Europe, we have eventual democratization of the regime. In China we have the survival of the Chinese Communist Party for now, 34 years after the 1989 protests. I felt that this comparison allowed me to highlight the other parallels. And then the differences between these two regimes.

Christopher Marquis: I really appreciated that and Chapter 9 on scope conditions where you look at the range of authoritarian regimes to further examine your theory. Your book really drove home to me the importance of comparative work. I also do work beyond China, and as I’ve been engaging with China scholars, I feel that there’s at times too much of an overweight on China, as unique and special place and less of interrogation of how and why – or why not – general principles can apply.

Martin Dimitrov: I fully agree with that. My sense is that we can’t know what is truly unique about a case until we have a comparison. The other thing is that oftentimes you have the same institution that exists both in China and in another country. But it operates in different ways. That’s also quite valuable to know, just because there is institutional isomorphism doesn’t mean that these institutions would function in identical ways.

In terms of these three main channels for the collection of information that are prioritized by regime insiders — the Communist Party, state security, and the internal media that produce reports that are not for public circulation — communist regimes have all three. But as we’re moving further away from communist regimes, we have fewer and fewer of these channels. The one channel that exists in all autocracies is the secret police. All autocracies have it, including the ones that are multi-party autocracies. But very few autocracies have internal journalism. This only exists in communist regimes. Finally, some may think that the type of party described in my book is everywhere. First of all, that’s not true, because there’re some autocracies (the military regimes) that don’t have parties. In Chapter 9, I analyze Argentina during the last military dictatorship. I also discuss Chile under (Augusto) Pinochet. There’re some cases where dictatorships don’t have parties, but typically they do. But those parties outside of the communist world are very different from the communist parties, because they are numerically smaller as a percentage of the population. And they don’t have the deep penetration that communist parties are able to ensure. So communist parties have both breadth and depth of penetration and non-communist parties don’t have either of those features, which means that their capacity to function as sophisticated collectors of information is limited by comparison with the parties that exist in the communist world. So, at the end of the day, the specificity of communist regimes is that they have these information collection channels. The single-party commerce regimes have communist parties, and those communist parties are extraordinary because of their functions both as information collectors and as institutions that coordinate the information gathering activities of other entities.

Christopher Marquis: I have a question on methods, your focus on archival ethnography. I know archival work and also ethnography, but I’ve not seen them used together. Can you let us know about your approach to the research? And one of the reasons why I would ask this is that studying information in autocracies does not sound like a topic where finding and gaining access to information is going to be easy. Could you reflect on your methods and approach to data?

Martin Dimitrov: To start with the end of your question. Yes, it is not easy to get information about how authoritarian regimes, and in particular, communist authoritarian regimes, have approached this problem of information collection. But I feel that with enough patience and willingness to explore new collections, one could find what one needs. It took me a long time. But in the end, I felt happy with what I had, and I felt that I had enough to provide answers to the questions that motivated the project.

In terms of what I did, I conducted about a hundred interviews, but they’re not an important source of information for me, because it was rare for me to find individuals who can knowledgeably comment on the totality of the information collection infrastructure. I also read about a hundred memoirs of various leaders of communist regimes. With the memoirs there is this dual problem of strategic forgetfulness and an invention of facts. So, as a source, they are written after the events. For me, they were not particularly valuable.

What I was looking for were contemporaneous sources that were not meant for public dissemination. And that’s what I found in the archives. So, these materials were produced for regime insiders, and when they were being generated, there was no expectation that they would ever be seen by non-insiders, that they would be seen by individuals who are not their intended recipients. So, what the government did is that it talked with remarkable honesty about the problems that it had in terms of evaluating discontent, because it was talking to itself. One set of regime insiders were collecting this information for another set of regime insiders.

This notion of archival ethnography is indeed something that political scientists don’t do. Standard ethnography involves immersion. The scholar gets immersed within a group of individuals, and the hope is that through this prolonged process of immersion, soaking and poking, there would be understanding. That understanding is, of course, the payoff. What I did is I immersed myself in regime-generated materials, usually in the archives where I read, although oftentimes I would collect materials, and then I would just process them for a long time in my office.

I did a multi-sited archival ethnography. I list the archival collections that I worked in there. There are 30 different archival collections located in China, Russia, Bulgaria, Germany, Taiwan, the United States, Cuba, and Argentina. So, I went to a lot of places looking for these materials, and some of the non-Chinese and non-Bulgarian collections are relevant for the scope conditions. What the documents gave me was a view onto how regime insiders themselves understood the information problem. What I talk about in this book is internal understandings and what I mean by these internal understandings is the understandings that emerge from the archival materials.

A lot of these understandings are about language — the language that was used. The terminology that was used by regime insiders is something that we might find at first perhaps hard to understand, but I feel it’s very revelatory. They talk about enemies. There’s this looking for enemies all the time, and we may feel that this is a paranoia, but it’s there both in the beginning periods and also throughout their existence. These regimes did feel that their capacity to stay in office was uncertain. They felt that they were being threatened. Some of these enemies are perhaps imaginary. But then there were others, actual opponents of the regime which were true enemies as far as the regime were concerned.

What these materials allowed me to do is to understand what regime insiders wanted to know, how they went about finding out, and once they had the information, what did they do with it. Cumulatively, I argue in the book that what archives allow us to do is to open the black box of authoritarian information collection, transfer, and use — and by doing that, to see what the dictators themselves wanted to see.

All documents have biases. I talk about the biases. And I talk about ways of overcoming some of those biases. And in terms of method, I also use quantitative analysis. I don’t just use qualitative methods. It doesn’t suggest that there’s only one path, that there’s only one way to use the materials that we collect through archival ethnography. They, of course, allow us to engage in process tracing, which I do in the book. But these archival materials also reveal quantifiable indicators of information gathering, transfer and use which I highlight in the book, in the various chapters.

Christopher Marquis: Super interesting. What’s your sense on the archives that you visited in China. Most of the work I’ve published on China uses quantitative data. But I’ve interviewed hundreds of people as well. I know that there have been increasing restrictions on some of the databases I use – such as only Chinese nationals at Chinese universities can uses them. I’ve read about various archives being restricted too. Do you think that if you were to try to do this project today, you’d be able to get access to those archives?

Martin Dimitrov: I would just have to go and see how it goes. It does appear that access to some materials has become more difficult, and to others perhaps impossible. But I have not attempted to go back since 2019 to try to access the collections that I worked with prior to 2019.

What I can say is that I got very lucky. I was able to collect a lot of materials. Maybe it was a question of timing, because this project began in 2007 and some of the China material was collected that early, in 2007, 2008 and 2009, and then it continued to through to 2019. So that’s a long period of time. A lot of changes occurred during that period, and then other changes have occurred since. But China is a large country, and sometimes these collections of Chinese materials are actually located outside China, in various repositories in greater China, but also in the U.S. I think the answer to that question is, we would just have to see what the situation is.

Click here for more from China Book Chats.