France has spent much of this week experiencing its most serious urban unrest in years. On Tuesday morning, police stopped 17-year-old Nahel Merzouk while he was driving in Nanterre, a western suburb of Paris. Video subsequently circulated showing two officers beside the vehicle before one fired as Nahel attempted to drive away. Nahel, who was of North African descent, died from the gunshot wound. The officer who fired was subsequently placed under investigation for intentional homicide.
The video mattered because it gave millions of people unusually direct evidence through which to evaluate the initial police encounter. Cases involving police use of force normally depend heavily on competing testimony, incomplete footage and institutional investigations. Here, citizens could watch part of the encounter themselves. That does not eliminate the need for a judicial investigation, because a short video cannot establish every legally relevant fact, but it dramatically reduced the state’s ability to control the initial narrative through official description alone.
What followed moved rapidly beyond the shooting itself. Protests spread from Nanterre to cities across France, while several nights produced extensive rioting, looting, arson and attacks on public buildings. Schools, police stations, municipal buildings and businesses have been damaged. Thousands of people have been arrested, and the government has deployed tens of thousands of police officers in an attempt to restore order.
The first analytical mistake would be treating all of this violence as an understandable extension of anger over Nahel’s death. Explanation is not justification. Burning a shop owned by someone who had nothing to do with the shooting does not become legitimate protest. Attacking a school does not create police accountability. Destroying public infrastructure frequently imposes the greatest costs on the same lower-income communities already experiencing weaker public services.
The second mistake would be concluding that because rioting is indefensible, the political grievances surrounding the shooting therefore become irrelevant. A society does not experience several nights of nationwide unrest every time one police officer is accused of unlawful violence. Individual incidents become politically explosive when they connect to grievances that already possess enough depth for thousands of people to interpret the event as evidence of something larger.
This distinction between trigger and cause matters. A match can start a fire, but the match does not explain why the building was combustible. Nahel’s shooting was the immediate trigger. The scale of the reaction requires examining policing, inequality, geography, immigration, discrimination and the relationship between the French state and communities concentrated in many of the country’s poorer suburbs.
France’s banlieues are particularly important to this story. The word technically refers simply to suburbs, many of which are prosperous, but in French political discussion it frequently refers to lower-income peripheral neighbourhoods containing substantial populations descended from immigrants from North and West Africa. Many of these communities contain modern housing, schools, businesses and ordinary middle-class families, so reducing them to zones of poverty or violence would be inaccurate. Yet some also experience significantly higher unemployment, concentrated deprivation and difficult relationships with police.
France’s republican model makes discussion of these inequalities unusually complicated. The French state traditionally emphasises universal citizenship rather than organising political identity officially around race or ethnicity. In principle, a citizen is French rather than a member of a legally recognised racial community. This universalism has an attractive logic: the state should treat individuals according to citizenship rather than ancestry.
The difficulty is that refusing to classify citizens by race does not automatically make racial disparities disappear. If people of North African or sub-Saharan African descent experience discrimination in employment, housing or policing, a political system needs some mechanism for identifying the pattern. Formal colour-blindness can protect equality, but it can also make inequality harder to measure when social outcomes remain correlated with origin.
This creates a fundamental disagreement over what integration means. One interpretation says the state should minimise ethnic categories because constantly classifying citizens according to ancestry reinforces communal division. Another argues that ignoring those categories makes it difficult to identify discrimination operating through them. Both positions contain legitimate concerns, and France has never fully resolved the tension.
Policing sits directly inside that debate because police represent the state at the point where state authority becomes physical. Most citizens encounter government through schools, hospitals, taxation or public transport. A police encounter is different because an officer possesses legal authority to stop, search, detain and, under particular circumstances, use force.
That asymmetry requires unusually strong accountability. The police need enough authority to protect citizens and themselves in dangerous situations, but the legitimacy of that authority depends on citizens believing force will be investigated seriously when its legality is questioned. A state monopoly on violence is legitimate only because the state claims to constrain its own violence through law.
France changed its rules governing police firearms in 2017, broadening circumstances under which officers can shoot at vehicles when drivers refuse to comply and pose specified dangers. Subsequent debate has focused on whether the reform contributed to increased fatal shootings during traffic stops. Later academic research would associate the law with a substantial rise in fatal police shootings involving moving vehicles. That does not determine whether the officer who shot Nahel acted lawfully in this individual case, which remains for investigators and courts to establish. It does make the legal framework itself a legitimate subject of political scrutiny.
This is where individual and institutional accountability need to remain separate. If the officer acted unlawfully, individual prosecution matters. But if the legal framework, training or policing culture systematically increases the probability of unnecessary lethal encounters, punishing one officer cannot solve the larger problem. Conversely, identifying institutional problems cannot erase individual responsibility where it exists.
The riots themselves create an equivalent distinction. Some participants may be responding to experiences of discrimination or alienation. Others may be opportunistically looting shops or participating because collective disorder temporarily reduces the expected cost of criminal behaviour. A crowd does not possess one motivation simply because everyone occupies the same street.
This is why describing the unrest entirely as either “an uprising against racism” or “criminal violence” is analytically weak. Both mechanisms can operate simultaneously. Political grievance can create the initial mobilisation, and opportunistic behaviour can expand once policing capacity becomes overwhelmed. The composition of unrest changes as it spreads.
Social media accelerates this process. Videos of Nahel’s shooting circulated almost immediately, allowing outrage to spread nationally without depending on traditional media. The same platforms can then distribute footage of riots, police actions and calls for mobilisation in real time. Information that once travelled through newspapers and television now moves between participants faster than state institutions can construct a response.
This increases accountability because citizens can document state behaviour directly. It also increases coordination capacity for disorder. The same smartphone can record potential police misconduct and help groups identify where police are absent. Technology does not possess one political effect; it reduces the transaction costs of information for everyone.
The government’s challenge is therefore more difficult than simply deploying enough police to stop riots. In the immediate term, restoring public order is necessary. Residents have a right not to have their cars burned, shops destroyed or neighbourhoods turned into battlegrounds. The state would fail in another direction if fear of appearing repressive prevented it from protecting citizens.
But restoring order cannot become the endpoint because order and legitimacy are different public goods. A sufficiently large police deployment can eventually suppress most street violence. It cannot by itself persuade communities that policing is fair or that social mobility is accessible. If the underlying relationship between institutions and citizens remains unchanged, the state may restore quiet without reducing the probability of another explosion after the next controversial incident.
France has seen this before. The 2005 riots, triggered after two teenagers died while hiding from police in Clichy-sous-Bois, spread across many disadvantaged suburbs and led the government to declare a state of emergency. Nearly two decades later, another incident involving young men and police has again produced national unrest. The recurrence suggests that treating each episode as an isolated public-order emergency is insufficient.
Yet the comparison should not become deterministic. France in 2023 is not France in 2005, and enormous numbers of residents in disadvantaged suburbs participate peacefully in French political, economic and cultural life. The fact that riots recur does not mean these communities exist permanently outside the Republic. It means particular institutional problems remain capable of generating intense distrust.
Economic inequality is part of the story but cannot explain everything. France possesses a large welfare state, extensive public healthcare and substantial redistribution compared with many countries. Residents of poorer suburbs are not simply abandoned by a state that provides nothing. This makes the French case more interesting because redistribution alone does not automatically produce social integration.
People care about income, but they also care about status, dignity, security and whether institutions treat them fairly. A citizen can receive public services while believing police view them suspiciously or employers discriminate against their name or address. Material redistribution cannot fully compensate for perceived unequal citizenship.
This is a limitation in purely economic interpretations of political stability. Raising household income matters enormously, but social legitimacy also depends on procedural fairness. People evaluate not only what outcomes they receive but how institutions interact with them.
The police face their own legitimacy problem because officers work in environments where violence and refusal to comply can create genuine danger. Treating every use of force as evidence of racism or brutality can make officers believe political institutions do not understand the risks they face. Low morale and distrust can then weaken recruitment and policing quality.
A sustainable reform therefore cannot be constructed around the assumption that either police or residents constitute a single morally coherent side. Police need protection and legal clarity. Citizens need accountability and equal treatment. Those objectives are compatible if institutions are designed well.
Independent investigation is central because neither police unions nor protesters should determine the legal facts of Nahel’s shooting. Courts exist precisely to convert evidence into judgement through procedures that are slower than social media but more capable of examining competing claims.
The speed difference creates another modern political problem. Public opinion forms within hours. Judicial investigations take months or years. By the time courts establish facts, millions of people may already have incorporated one interpretation into their political identity.
Institutions therefore need to become better at communicating uncertainty. Governments should release verified information quickly without pretending investigations are complete. Police organisations should correct inaccurate initial claims when evidence contradicts them. Transparency can reduce the informational vacuum in which conspiracy and partisan narratives develop.
The political right in France is likely to emphasise law, order, immigration and parental responsibility after the riots. The left will emphasise policing, discrimination and social inequality. Each side will identify genuine components of the crisis and risk treating its preferred component as the entire explanation.
Marine Le Pen’s National Rally may benefit politically if voters interpret the unrest primarily through insecurity and immigration. That creates an additional incentive for the government to restore order rapidly, but political competition should not determine how evidence about policing is evaluated. A democracy weakens when governments avoid necessary reform because they fear admitting a problem will strengthen their opponents.
Likewise, the left should not minimise criminal violence because acknowledging it appears to support right-wing narratives. A shopkeeper whose business is destroyed does not experience the loss as an abstract expression of structural inequality. Political analysis should be capable of recognising structural grievances without denying individual agency.
This is where two-level responsibility becomes useful. Individuals remain responsible for actions they choose: an officer for an unlawful shooting if established, a rioter for arson or assault, a looter for theft. Institutions remain responsible for designing environments that change the probability of those actions occurring and determine whether they are detected and punished.
The two levels reinforce rather than replace one another.
France therefore needs both criminal accountability and institutional reform where evidence supports it. It needs police capable of controlling violent disorder and independent institutions capable of investigating police violence. It needs economic investment in disadvantaged neighbourhoods and expectations that residents obey the same laws as everyone else.
None of those principles requires pretending every grievance is justified or every institution is discriminatory.
The larger question is whether French universalism can become substantive enough that citizens with immigrant backgrounds experience the Republic not merely as a legal identity but as an institution that treats them consistently as members of it. That requires both integration and equality before law.
Integration cannot mean that citizens descended from immigrants must continually prove their Frenchness. Equality cannot mean that social differences become impossible to discuss because the Republic officially refuses racial categories. A universal national identity works best when citizens believe the universalism is reflected in institutional behaviour.
The riots also demonstrate why state legitimacy is a stock rather than a flow. Governments accumulate trust through thousands of ordinary interactions before a crisis. When trust is high, citizens may interpret one disputed police shooting as an individual failure. When trust is low, the same event becomes evidence confirming an existing belief about the entire institution.
This resembles banking. A bank run does not begin merely because one piece of bad news appears; it begins because enough depositors already doubt whether everyone else’s confidence will survive the news. Political institutions operate similarly. One event can trigger systemic distrust when legitimacy reserves are already weak.
Rebuilding those reserves takes much longer than restoring order on the streets.
France will eventually stop burning cars. Police deployments, arrests and exhaustion will reduce the immediate unrest. The more important measure of success will be what happens after the television cameras leave: whether the investigation into Nahel’s death is credible, whether policing practices are examined honestly, whether disadvantaged communities experience greater opportunity and whether political leaders avoid converting the crisis into a permanent conflict between “France” and parts of its own population.
The language used now matters. If every rioter becomes representative of immigrant-origin citizens, millions of peaceful French people are collectively associated with violence they did not commit. If every police officer becomes representative of alleged brutality, thousands of officers are collectively condemned for actions they did not perform.
Collective blame is politically efficient because it creates simple camps.
It is institutionally destructive because democratic citizenship depends on individual responsibility.
Nahel’s death deserves investigation because the state must account for lethal force used in its name. The riots deserve condemnation because political grievance does not provide a licence to destroy other people’s lives and property. France should be capable of maintaining both propositions without treating either as a concession to the opposing political camp.
That may be the central test emerging from this week. A mature political system should not have to choose between order and justice. Order without justice stores resentment for the next crisis. Justice without order leaves citizens exposed to coercion from whoever can mobilise violence.
The state is responsible for providing both.
The shooting in Nanterre was the spark.
Understanding why so much caught fire requires looking at everything that was already dry.
