Some of the most politically striking images from China this week contained almost no words.
Protesters in Shanghai, Beijing and other Chinese cities held up blank sheets of white paper. There was no printed manifesto, party logo or political programme. The emptiness was the message. In a country where criticism can disappear rapidly from the internet and public dissent carries substantial personal risk, protesters found a way to represent censorship by displaying what they believed they were not permitted to write.
The demonstrations followed a deadly apartment fire in Ürümqi, the capital of Xinjiang, on 24 November. Ten people were reported killed. The fire occurred after months of strict COVID controls in the city, and public anger quickly focused on whether pandemic restrictions had impeded residents’ escape or the emergency response. Authorities denied that locked exits prevented residents from escaping and rejected claims that COVID measures obstructed firefighters. Yet by that point, the specific details of the fire had become connected to a much larger reservoir of frustration with China’s zero-COVID system.
Over the following days, demonstrations appeared in cities including Shanghai, Beijing, Wuhan and Chengdu and on university campuses. Many protesters demanded an end to lockdowns and repeated testing. Some called for greater freedom of expression. In unusually confrontational moments, demonstrators in Shanghai were recorded calling for Xi Jinping and the Communist Party to step down.
Those slogans are extraordinary in the Chinese political context.
They should also be interpreted cautiously.
A protest containing several hundred people demanding Xi’s resignation does not establish that 1.4 billion Chinese citizens want regime change. People willing to demonstrate publicly under an authoritarian government are an unusually self-selecting group. Many Chinese citizens may oppose lockdowns while continuing to support the Communist Party. Others may dislike particular local officials rather than Xi. Some may support zero-COVID because they fear widespread illness among elderly relatives.
Visible protest tells us that dissatisfaction exists.
It does not automatically tell us its national distribution.
This is the same methodological problem I discussed during Iran’s protests. Authoritarian systems make public opinion unusually difficult to measure because both silence and protest are selected observations. Silence may represent support, indifference or fear. Protest may represent widespread hidden dissatisfaction or a highly motivated minority.
The government therefore faces the same information problem as outside observers.
That is where China’s protests become particularly interesting.
The conventional argument about authoritarian government focuses on accountability. Citizens cannot easily remove leaders through competitive national elections, opposition parties cannot openly contest power and independent media operate under severe constraints. Those are fundamental democratic criticisms.
But authoritarianism creates another problem even for a leader interested purely in effective government:
How does the leader know when the policy is failing?
This is an information problem rather than a moral one.
Every government needs feedback. A company needs to know whether customers dislike its product. A military commander needs accurate reports from the battlefield. A central bank needs reliable inflation data. A government needs information about whether policies are producing their intended results and how citizens are responding.
Democracies generate enormous quantities of this information almost accidentally. Opposition politicians have incentives to publicise government failures. Journalists search for scandals. Pollsters measure dissatisfaction. Citizens protest. Elections provide an extremely crude but powerful final signal.
The information is noisy and often partisan.
It is still information.
Authoritarian governments suppress many of these mechanisms because the same information that improves policy can threaten political control.
An independent newspaper revealing that a policy failed helps the government learn.
It also makes the government look incompetent.
An opposition politician identifying public anger provides useful feedback.
They also become a potential political rival.
A protest reveals dissatisfaction.
It can also demonstrate that collective action against the government is possible.
The authoritarian state therefore faces a structural trade-off between information and control.
The more completely it controls political information, the harder it becomes to know what people actually think.
China has historically attempted to manage this problem through an enormous bureaucracy, internal reporting mechanisms, local experimentation, online monitoring, surveys and other forms of feedback that do not require competitive national politics. The Chinese state is not simply blind because it is authoritarian. It possesses sophisticated mechanisms for collecting information.
But those mechanisms still operate inside a hierarchy.
Hierarchy creates incentives.
Suppose a local official is responsible for implementing zero-COVID. Beijing has made clear that preventing outbreaks is a political priority. The official knows that allowing infections to spread could damage their career. They therefore impose strict controls.
Now suppose residents become increasingly angry.
How strongly does the official communicate that anger upwards?
Reporting that the policy is becoming unpopular might be interpreted as admitting poor implementation. Reporting success is safer.
The next official in the hierarchy faces the same incentive.
By the time information reaches the top, bad news may have been filtered repeatedly.
This is sometimes called the dictator’s dilemma: authoritarian leaders can punish officials for failure, but the more frightening punishment becomes, the stronger officials’ incentives become to conceal failure.
Power can therefore reduce information.
The leader becomes stronger politically while becoming weaker epistemically.
This is not unique to governments. Companies experience it when employees fear contradicting chief executives. Militaries experience it when officers tell commanders what they want to hear. Academic institutions experience it when junior researchers are afraid to challenge famous professors.
Hierarchies produce information distortion whenever career incentives reward agreement.
Political authoritarianism magnifies the problem because the leader has fewer independent sources through which to verify what subordinates report.
China’s zero-COVID policy provides an unusually important case because it was initially extremely successful according to its stated objective. China avoided the enormous waves of deaths experienced in the United States, Europe, India and elsewhere during much of 2020 and 2021. Restrictions, testing, quarantine and border controls kept transmission comparatively low for long periods.
That success matters.
It is too easy, from the perspective of late 2022, to pretend the policy was obviously irrational from the beginning.
It was not.
The more interesting question is why a strategy that produced substantial benefits during earlier stages of the pandemic became increasingly difficult to change when circumstances changed.
Omicron is dramatically more transmissible than the original virus. Vaccines now exist. Treatments have improved. Much of the rest of the world has reopened. The economic and psychological costs of repeated lockdowns have accumulated.
The optimisation problem changed.
Policy needed to change with it.
This is the same dynamic-policy problem Hong Kong confronted in March. A strategy can be correct at time t₁ and incorrect at time t₂ without either judgement being contradictory.
The difficulty is that policies accumulate political meaning.
Xi Jinping’s government did not present zero-COVID merely as one temporary public-health instrument among several. China’s low death toll became associated with the superiority and competence of the country’s political system, particularly when compared with the enormous mortality experienced in several Western democracies.
Once policy becomes evidence of regime competence, abandoning the policy becomes politically harder.
Changing course risks looking like admitting that critics were right.
This is policy lock-in through political identity.
The stronger a government claims ownership of a successful policy, the greater the reputational cost when the policy eventually needs to change.
Democratic governments experience this too.
The difference is that opposition parties can help force policy change by making the political cost of maintaining an unpopular strategy larger than the cost of abandoning it.
In a highly centralised system, the same leader responsible for defending the policy is also responsible for deciding when it failed.
That creates a conflict of interest.
Updating becomes psychologically and politically expensive.
This is one reason institutions should make policy reversal easier.
A government should be capable of saying that a strategy worked under previous conditions and no longer works under current ones.
That is not inconsistency.
It is Bayesian updating.
Politics often rewards the opposite because leaders want to appear decisive and consistent.
Yet consistency becomes irrational when the world changes.
The protests may therefore provide information the Chinese political system struggled to transmit through ordinary channels: the social cost of zero-COVID has become much larger.
That does not mean protests alone caused any policy adjustments occurring now, and we should be cautious about assigning causality while events are unfolding. Chinese authorities had already been modifying some quarantine and testing rules before the demonstrations as economic pressures and the difficulty of containing Omicron increased.
But public mobilisation changes the information environment.
A complaint posted privately can be dismissed as individual dissatisfaction.
Thousands of people demonstrating across several cities provide evidence that dissatisfaction is coordinated enough to overcome substantial fear.
The decision to protest is itself costly.
That makes it informative.
Economists call this a costly signal.
If publicly criticising the government carries little cost, a protest tells us relatively little about the intensity of someone’s preference. If criticism risks arrest, university consequences or career damage, participation indicates much stronger motivation.
This is the paradox of repression.
Repression reduces the number of visible protests.
It can make the protests that survive repression more informative.
The government then faces another dilemma. If it represses demonstrations aggressively, it may restore public order quickly but lose additional information about the grievances underneath them. If it tolerates demonstrations, more citizens may join once they discover that protest is possible.
This is why authoritarian states are particularly sensitive to coordination.
One dissatisfied citizen is manageable.
One million dissatisfied citizens who believe everyone else is satisfied are also potentially manageable.
One million dissatisfied citizens who suddenly discover one another are much more dangerous.
The blank paper becomes important in this context because it is simultaneously protest and information.
It tells other citizens: someone else is dissatisfied too.
This is how informational cascades can begin.
Again, we should not assume one is occurring at a revolutionary scale in China. The state retains enormous coercive capacity, organisational strength and substantial public support. The demonstrations appear relatively small compared with China’s population, and security forces have already increased their presence in several cities.
There is little evidence as of tonight that the Communist Party is close to losing political control.
That is not the relevant threshold.
The more useful question is whether the protests alter policy.
Authoritarian responsiveness is often underestimated because we assume regimes either democratise or ignore citizens.
Reality contains a much larger middle category.
Authoritarian governments can respond to public pressure without allowing political competition.
They can change a policy.
Dismiss an official.
Compensate victims.
Increase welfare spending.
Relax restrictions.
The regime remains authoritarian while adapting enough to preserve legitimacy.
China has done this repeatedly.
This produces a form of responsive authoritarianism in which the government attempts to obtain enough information about public dissatisfaction to correct policies without creating institutions that could challenge the Communist Party’s monopoly on political power.
The arrangement can be effective.
Its weakness appears when the information needed for policy correction is politically dangerous enough that the system suppresses it.
Zero-COVID may have reached that point.
There is also an economic constraint. Lockdowns disrupt consumption, manufacturing, logistics and employment. China’s economy is already facing pressure from the property-sector crisis, weak consumer confidence and slowing global demand.
Public-health policy therefore interacts with economic legitimacy.
The Communist Party’s post-Mao political legitimacy has relied partly on extraordinary economic performance. Hundreds of millions of people experienced dramatic improvements in living standards over several decades.
That creates a powerful performance-based argument for the political system.
But performance legitimacy creates its own vulnerability.
If citizens tolerate limited political competition partly because the existing system reliably delivers rising prosperity and stability, prolonged economic stagnation changes the implicit bargain.
This does not mean slower growth automatically produces democratisation. Authoritarian governments can survive poor economic performance for long periods.
It means the political value of competence becomes smaller if competence is no longer experienced materially.
A young Chinese graduate struggling to find employment evaluates the system through different conditions from their parents, who may have experienced enormous upward mobility during China’s rapid industrialisation.
Generational reference points change political expectations.
This resembles Iran in a completely different ideological context. Revolutionary legitimacy depreciates when younger citizens did not experience the revolution. China’s legitimacy is less revolutionary today and more heavily associated with nationalism, stability and development.
Those sources of legitimacy also need renewal.
The protests therefore occur at an important moment for Xi personally. Only last month, the Communist Party’s 20th Congress gave him a third term as general secretary and installed a leadership team widely seen as dominated by loyalists.
Xi is now more politically powerful than any Chinese leader in decades.
That centralisation may increase his ability to implement decisions rapidly.
It may simultaneously make accurate disagreement harder.
This is the central paradox of personal power.
Removing internal rivals reduces political constraints.
Those rivals may also have been sources of alternative information.
A leadership team selected partly for loyalty may coordinate efficiently.
It may become less willing to tell the leader that a signature policy is failing.
Political scientists sometimes treat checks and balances purely as mechanisms preventing abuse.
They are also information-producing institutions.
Opposition creates criticism.
Independent media creates criticism.
Competing bureaucratic centres create criticism.
Courts create criticism.
Much of that criticism is irritating, partisan or wrong.
The system benefits from the fact that somebody has an incentive to produce it.
Authoritarian centralisation reduces noise.
It can also reduce signal.
This is why the white sheets of paper seem so symbolically appropriate.
The Chinese state possesses extraordinary quantities of data about citizens. Cameras, digital payments, mobile phones and online platforms create an information environment previous governments could barely imagine.
Yet the most important piece of political information this week may be something technologically primitive:
a blank piece of paper held above someone’s head.
The state can know where a citizen travelled, which COVID test they took and which QR code they scanned.
That does not mean it knows what they believe.
Surveillance and knowledge are not the same thing.
A government can possess enormous information about behaviour while remaining uncertain about preferences because citizens alter behaviour when they know they are being watched.
This is another form of the observer effect.
Control changes the data being observed.
The stronger the incentive to perform loyalty, the less informative performed loyalty becomes.
That may be the deepest problem authoritarian governments face.
They want compliance.
They also need to know whether compliance is genuine.
Methods that maximise the first can undermine the second.
Democracy solves part of this problem by making disagreement comparatively cheap. Citizens can criticise government without necessarily threatening the political system. Opposition therefore becomes routine information rather than an existential challenge.
The cost is constant noise, polarisation and political conflict.
Authoritarianism suppresses much of the noise.
The cost is uncertainty about what the silence means.
Neither system guarantees competent policy.
But the informational advantages of open disagreement become especially visible when circumstances change quickly and governments need to know whether an old policy remains sustainable.
China’s zero-COVID system is now confronting exactly that problem.
The protests should therefore not be romanticised into a revolution that has not happened.
Nor should they be dismissed because they may not threaten Communist Party rule.
Political significance exists between irrelevance and regime change.
If public pressure causes the government to reconsider how zero-COVID operates, then relatively small demonstrations will have altered policy inside one of the world’s most powerful authoritarian states.
That would tell us something important about political power.
Citizens do not need to overthrow a government before influencing it.
Governments respond to threats to legitimacy long before threats become existential.
The difficult question is what happens afterwards.
If Beijing relaxes restrictions, does it frame the change as a response to evolving scientific conditions rather than protest?
Probably.
Governments prefer to preserve the appearance of autonomous decision-making because openly conceding to protest can encourage future mobilisation.
That does not make the policy change meaningless.
It demonstrates how authoritarian responsiveness often works indirectly.
The government listens without admitting that it listened.
Citizens learn anyway.
That can alter expectations about future protest.
This is why even successful policy adjustment can create political risk for an authoritarian government.
Concession provides information too.
It tells citizens that collective pressure sometimes works.
The government therefore has to balance the benefit of correcting an unpopular policy against the cost of teaching citizens that mobilisation produces correction.
Democracies solve this more easily because policy responsiveness is supposed to be normal.
Governments are expected to lose elections or reverse policies when citizens reject them.
Authoritarian systems need responsiveness without appearing vulnerable.
That is a much harder equilibrium.
China’s white-paper protests may eventually disappear under police pressure.
Zero-COVID may be modified.
Xi may remain completely secure.
All of those things can happen simultaneously.
The important lesson is not that China’s political system is suddenly collapsing.
It is almost the opposite.
A government can appear extraordinarily powerful while still possessing vulnerabilities created by the very mechanisms through which it maintains power.
Censorship reduces visible dissent.
It also reduces information.
Centralisation reduces internal opposition.
It also reduces independent feedback.
Repression prevents coordination.
It also makes successful coordination more politically meaningful.
China’s government can remove words from the internet.
This week, protesters discovered that even an empty page can communicate.
Sometimes silence means nobody has anything to say.
Sometimes it means everyone is waiting to discover who will speak first.
