Last week, I argued that Peru’s institutions were right to stop Pedro Castillo when he attempted to dissolve Congress and govern by decree. A democratically elected president does not acquire the right to destroy constitutional constraints simply because those constraints have become politically inconvenient.
One week later, the more difficult part of that argument has arrived.
Castillo is gone, but Peru’s crisis has not ended. Protests have spread across the country, particularly through poorer southern and Andean regions where Castillo received substantial support in the 2021 election. Roads have been blocked, airports targeted and government buildings attacked in some areas. Protesters are not united around one demand: some want Castillo released, others want President Dina Boluarte to resign, many demand immediate elections, and others want Congress dissolved or a constituent assembly created.
The state’s response has become increasingly severe. On Wednesday, the government declared a nationwide state of emergency, suspending or restricting certain constitutional protections and authorising the armed forces to assist police in restoring order. The following day, violence around the airport in Ayacucho became particularly deadly. Protesters entered the airport area and clashed with security forces; gunfire subsequently spread into surrounding streets. Later investigations and autopsy evidence would establish that ten people ultimately died from the violence in Ayacucho, several after being shot.
This creates an uncomfortable problem for anyone who defended Peru’s institutions last week.
The government is constitutionally legitimate.
That does not mean every action taken by the government is legitimate.
Those propositions need to remain separate.
There is a tendency during political crises to divide institutions into democratic and anti-democratic camps and then evaluate everything they do according to which side they occupy. Castillo attempted to dissolve Congress, so the institutions resisting him become “the democratic side”. Protesters demanding his return can then be described as opposing constitutional order, making force against them appear like defence of democracy.
Reality is more complicated.
A democratic state can defend constitutional order through undemocratic methods.
A protest movement can contain legitimate grievances while some participants engage in violence.
Security forces can possess a legitimate reason to intervene while using illegitimate levels of force.
Political analysis becomes less useful when acknowledging one of those facts is treated as denying another.
The state has a genuine obligation to preserve public order. Citizens cannot permanently seize airports, destroy public infrastructure or use violence against police simply because they believe the government is illegitimate. Airports are not symbolic spaces alone; disrupting them affects ordinary citizens, emergency services, commerce and regional connectivity. Road blockades similarly impose costs on people who may have no involvement in the political dispute.
The right to protest is not an unlimited right to impose any cost on everyone else.
But state power has limits for exactly the same reason.
Police and soldiers possess something protesters generally do not: the organised coercive capacity of the state. They have firearms, legal authority, detention powers and institutional protection. That asymmetry creates a higher standard of responsibility.
A protester throwing a rock and a soldier firing a rifle are not equivalent simply because both can be described as “violence”.
The expected lethality is different.
The training is different.
The institutional obligation is different.
This is why democratic systems generally treat lethal force as a last resort rather than another tool of crowd control. Peru’s own military regulations permit firearms in circumstances involving imminent threats to life or severe injury and state that lethal force should be a last resort. The dispute in places such as Ayacucho will therefore not simply be whether protesters behaved lawfully. It will also be whether the security forces’ response remained proportionate to the threats they actually faced.
That distinction between necessity and proportionality is fundamental to legitimate state coercion.
A government may have a legitimate objective and still use illegitimate means.
Preventing protesters from taking an airport may be legitimate.
Firing indiscriminately into surrounding streets would not become legitimate simply because the original objective was.
This is the same reason criminal law distinguishes self-defence from revenge. The existence of an initial threat does not provide unlimited authority afterwards.
Democratic government requires the state to remain constrained precisely when it possesses the strongest argument for using force.
That is harder than it sounds.
Security forces operating during riots face uncertainty, fear and rapidly changing threats. Officials may not know whether a person approaching them is peaceful or armed. Protesters can attack police. Crowds can become dangerous quickly. Judging every decision retrospectively from a safe distance risks ignoring the information available to officers at the moment.
But uncertainty cannot become blanket immunity.
If anything, the state’s monopoly over legitimate violence makes independent investigation more important because the state cannot be the final judge of whether its own force was justified.
This is where courts, prosecutors, ombudsmen, journalists and human-rights organisations become essential.
Their purpose is not automatically to assume the police were wrong.
It is to prevent the institution exercising coercion from being the only institution determining whether that coercion was lawful.
Checks and balances apply to batons and bullets as much as legislation.
The political geography of the protests makes this crisis even more serious. Castillo’s support was strongest in many poorer rural and Andean regions that have long felt distant from political and economic power concentrated in Lima. The current demonstrations therefore cannot be understood only as personal loyalty to a failed president.
Castillo has become a vessel for grievances larger than Castillo.
This is a recurring political mechanism. A leader can be personally incompetent or even anti-democratic while still representing a constituency with legitimate reasons for dissatisfaction. When the leader falls, those grievances do not disappear with them.
This is why I argued last week that the legitimacy of a constituency and the quality of its representative must be separated.
Castillo’s attempted self-coup was indefensible.
That does not mean rural Peruvians were wrong to believe the country’s political institutions have historically neglected them.
The distinction matters because governments frequently make a serious mistake after defeating a populist leader: they assume defeating the politician has defeated the political problem.
It has not.
The politician may have been a symptom.
Peru’s geography illustrates the underlying divide. Lima contains enormous political, administrative and economic power. Many communities in the Andes and southern regions experience weaker state services, poverty and a sense that national economic development has not translated into equal political inclusion.
Peru is one of the world’s major producers of copper and other minerals, yet communities located near extractive industries can still experience poverty and inadequate public services. That produces a politically explosive question: if enormous wealth is being removed from the ground beneath us, why does the state still appear absent from our lives?
Aggregate GDP cannot answer that.
National growth can coexist with local exclusion.
This is similar to the geographical politics visible in France, Australia and the United States. Globalisation and economic growth produce benefits unevenly across territory. Metropolitan regions often accumulate capital, professional employment and political influence, while peripheral regions can experience extraction or economic disruption without equivalent institutional investment.
Political alienation then becomes geographical.
Eventually the map of election results becomes a map of different experiences of the state.
Castillo’s 2021 election made that divide visible. A rural schoolteacher from Cajamarca defeated Keiko Fujimori by an extraordinarily narrow margin, drawing strong support from regions historically marginalised from Lima’s political establishment.
His presidency then failed spectacularly.
For those voters, the lesson could be interpreted in two completely different ways.
One interpretation is that Castillo proved outsider populism cannot substitute for competent government.
Another is that the political establishment never allowed their representative to govern and ultimately removed him.
The first may contain substantial truth.
So may parts of the second, given Congress’s repeated attempts to remove Castillo and Peru’s broader institutional dysfunction.
Political beliefs become durable when they contain enough truth to survive contradictory evidence.
This is why simply telling Castillo supporters that he attempted an unconstitutional seizure of power may not persuade them that Boluarte’s government deserves legitimacy.
They can accept the first proposition and still distrust the second.
Boluarte therefore faces a problem that constitutional succession alone cannot solve.
She was Castillo’s vice-president and became president according to the constitutional order after his removal. That gives her legal authority.
Yet many protesters view her as having aligned herself with the Congress they already distrusted.
This produces a gap between constitutional legality and sociological legitimacy: the state knows who the president is, but substantial parts of society do not accept that the political process producing her represents them.
That gap can be bridged through elections.
Boluarte has moved towards supporting early elections rather than completing the presidential term through 2026. Congress is debating how quickly elections can realistically be organised.
That seems increasingly necessary.
But early elections are not a magical solution.
Peru held elections in 2021.
They produced Castillo.
If the institutional structures generating political fragmentation remain unchanged, another election may simply redistribute the same instability among different personalities.
This is why calls for “new elections now” can be simultaneously necessary and insufficient.
Elections can restore a mandate.
They cannot automatically create stronger political parties, improve state capacity or resolve the executive-legislative conflict embedded in Peru’s constitutional system.
The country therefore faces two time horizons.
In the short term, it needs a political mechanism capable of restoring enough legitimacy to reduce violence.
In the longer term, it needs institutional reform capable of preventing the next election from producing another version of the same crisis.
Those objectives can conflict.
Institutional reform requires time, negotiation and political stability.
The public is demanding rapid political change precisely because trust has collapsed.
This is another reason constituent assemblies become attractive during crises. If citizens conclude the entire political system is malfunctioning, rewriting the constitution appears to offer a way of addressing everything simultaneously.
Chile should make Peru cautious about that assumption.
Constitutions matter enormously.
They do not solve every political problem.
A new constitutional document cannot automatically create competent parties, honest politicians or effective local government.
Institutions operate through organisations and incentives.
Writing better rules helps.
Expecting rules alone to manufacture political culture is dangerous.
Peru’s immediate challenge is even more basic: stopping the cycle in which every death produces more anger, more protest and therefore more coercion.
This is where repression can become self-defeating.
A government deploys security forces because protests threaten public order. People are killed. Their funerals become new political symbols. Families who were previously indifferent become angry. Protesters become more convinced that the government is illegitimate. The government sees larger demonstrations and deploys more security forces.
Each side’s reaction becomes evidence confirming the other’s narrative.
The government says the country faces increasingly dangerous disorder.
Protesters say the government is increasingly repressive.
Both observations can become true simultaneously.
That is an escalation spiral.
Breaking it requires political action, not simply more effective policing.
Dialogue sounds weak during disorder because it appears to reward protest.
But coercion alone cannot resolve a legitimacy crisis.
The government needs to distinguish between people engaging in violence and the much larger population expressing political demands.
If every demonstrator is treated as a threat, the state expands the coalition opposing it.
This is one of the oldest mistakes in counterinsurgency and protest management.
Indiscriminate coercion creates solidarity among groups that previously had different grievances.
Selective law enforcement does the opposite.
It isolates violent actors from peaceful constituencies.
That distinction is strategically important as well as morally important.
The protesters face a reciprocal responsibility. If demonstrations become associated with attacks on infrastructure, airports or individuals, the movement gives the government stronger justification for emergency measures and alienates citizens who might otherwise sympathise with demands for elections.
Non-violent mobilisation is not merely morally preferable.
It can be strategically superior because it reduces the state’s ability to frame opposition as a security threat.
Again, means affect legitimacy.
This is where both government and opposition need to resist the language of absolute moral ownership.
Boluarte cannot say: we represent constitutional democracy, therefore everyone resisting us threatens democracy.
Protesters cannot say: we represent the people, therefore every institution opposing us is illegitimate.
Peru contains millions of citizens who supported Castillo, millions who opposed him, people who despise Congress, people who fear Castillo’s return and people who simply want the political crisis to stop.
No faction owns “the people”.
Pluralism means accepting that legitimate political disagreement remains after a constitutional crisis.
The state has a higher burden because it possesses coercive power.
That burden is what distinguishes democratic policing from political repression.
Governments often describe restrictions as temporary necessities during emergencies. Sometimes they are. States genuinely need exceptional tools during extraordinary disorder.
The danger is that emergencies alter incentives.
Executive power expands.
Oversight weakens.
Security institutions acquire greater discretion.
Citizens become accustomed to restrictions.
Temporary measures can become easier to extend than remove.
This is why states of emergency need unusually strong oversight and clear limits.
An emergency should increase scrutiny, not suspend it.
If ordinary law is temporarily insufficient, the government should have to explain precisely why exceptional authority is necessary, what powers are being used and when those powers expire.
Otherwise emergency government becomes constitutional ambiguity in favour of the executive.
Peru has spent too many years suffering from ambiguous constitutional powers already.
The response should not be adding more ambiguity.
There is a broader lesson here about democratic legitimacy that complicates what I wrote last week.
Stopping an unconstitutional president is not the same thing as restoring democracy.
It is one necessary act.
Democracy also requires protecting the rights of people who supported that president after he is gone.
This is what liberal constitutionalism adds to majority rule.
Rights do not belong only to people making good political choices.
If they did, they would not be rights.
A Castillo supporter retains freedom of expression and peaceful assembly even if Castillo attempted to dissolve Congress.
A Boluarte supporter retains political rights even if protesters believe her government is illegitimate.
The state must protect both.
That is difficult when violence begins because governments naturally divide society into supporters of order and threats to order.
A mature constitutional system needs a third category: citizens who oppose the government but remain entitled to its protection.
That category is democracy’s real test.
Governments rarely struggle to protect supporters.
The legitimacy of the state becomes visible in how it treats opponents.
Peru’s institutions passed one test on 7 December. They refused to obey a president attempting to place himself above the constitution.
This week they are facing another.
Can those same institutions defend public order without treating constitutional restraint as something that applies only to presidents?
The rule of law cannot mean Castillo was constrained but soldiers are not.
It cannot mean Congress is protected but protesters are not.
It cannot mean the state may use whatever methods are necessary because the state happens to be defending the correct constitutional outcome.
Rules become meaningful only when they constrain the side with the strongest argument for breaking them.
That is why the deaths this week matter politically beyond the immediate tragedy.
Every unjustified use of force tells citizens that constitutional order protects institutions more effectively than people.
Every failure to investigate reinforces that belief.
And once enough citizens believe the constitution belongs to the political class rather than to them, defending the constitution becomes much harder.
Peru therefore needs investigations into the deaths, restraint from security forces, accountability for violence committed by protesters, a credible timetable towards elections and a much deeper conversation about why the country’s democratic institutions repeatedly fail to represent large parts of its population.
None of those measures requires defending Castillo’s attempted self-coup.
That is the point.
Democratic politics should be capable of saying Castillo was wrong and the state can be wrong afterwards.
Otherwise constitutionalism becomes tribal: rules for them, exceptions for us.
Last week, Peru demonstrated that a president is not above the constitution.
This week, the government needs to demonstrate that neither is the state.
