A country that refuses to count cannot argue

On Saturday, October 17, some 550,000 people marched against the cost of living, according to the Interior Ministry. By nightfall, shops were being looted from Paris to Toulouse, and the ministry would count 2,397 arrests. By Monday, the government put the number of businesses damaged or looted at "at least 3,000." Prime Minister Sébastien Lecornu had already given his reading on X: ultra-left groups had "organized the violence," and "delinquents joined in to loot."

That is the most precise thing anyone in authority will say about who did it, and it will stay that way. Not because the police lack names. Because France has decided, by law and by habit, to describe the people it arrests by age, sex and passport, and nothing else.

Everyone else has decided too: plenty of people are certain they know where the looters came from. They may be right, or half right. I don't know, and neither do they. That is my point.

What France does count

The state counts something. In its 2025 report, published in July 2026, the Interior Ministry's statistical service, the SSMSI, found that foreign nationals, about 9 percent of the population, made up 37 percent of people implicated in thefts from vehicles, 36 percent for burglaries, 33 percent for violent thefts without a weapon and 22 percent for drug trafficking. For homicide it is 18 percent. The Justice Ministry reports that 17 percent of 2024 convictions concerned foreign nationals. On February 5, 2026, foreigners were 21,167 of the 86,979 people in French prisons, about 24 percent.

These numbers are real, and not small. But read the definitions. Mis en cause means suspected by the police in a solved case, not convicted; nationality can be self-declared, and the count includes tourists. The same report shows that 18-to-29-year-olds are overrepresented everywhere, and that unarmed violent theft is 29 times more frequent in the Paris urban area than outside towns. Youth and big cities do a lot of the work. The SSMSI calls its analyses "summary," and notes that since 2022 the foreign share has fallen for almost every indicator.

And nationality is the wrong instrument anyway. A French citizen born in Seine-Saint-Denis to Malian parents and one born in the Cantal to Auvergnat farmers are the same line in these tables. What people actually argue about, second generations, neighborhoods, families, is invisible by design.

The wall

This is deliberate. Article 6 of the 1978 data protection law, Informatique et libertés, forbids processing data that reveal "the alleged racial origin or the ethnic origin" of a person, with exceptions for research and statistics. In 2007, the Constitutional Council (decision 2007-557 DC) struck down a provision that would have allowed CNIL-authorized studies on "the diversity of origins." The ground was procedural, but the Council added that such studies "may rest on objective data" and "cannot, without violating Article 1 of the Constitution, rest on ethnic origin or race." What remains legal is country of birth and nationality, a person's and their parents', in voluntary, anonymous surveys.

The best of them, Trajectories and Origins (TeO2), surveyed about 27,000 people aged 18 to 59 for INED and INSEE in 2019 and 2020. It is excellent on housing, schooling, jobs and discrimination, and was not built to measure offending.

The one researcher who tried to look got the full treatment. Hugues Lagrange, a CNRS sociologist, matched school data on some 4,400 pupils around Mantes-la-Jolie with court records. In Le Déni des cultures (2010), he reported that teenagers raised in families from the Sahel were markedly overrepresented among those charged, at equal social background, and argued that family structure mattered: sibling numbers, spousal age gaps, isolated mothers. He also called for social mixing. The answer was not a better dataset. It was the word "culturalism." Some objections were serious, like the historian Pap Ndiaye's point that cultures are not fixed. That is a scientific dispute. It deserved data. It got anathema.

What the law punishes, and what it lets you say

The 1881 press law does not forbid stating facts about crime. It punishes defamation, and provocation to discrimination, hatred or violence against a group because of its origin. Éric Zemmour found the line. On March 6, 2010, on Canal+'s Salut les Terriens, in an exchange about identity checks, he said: "Why are people stopped 17 times? Why? Because most traffickers are Black and Arab, that's how it is, it's a fact." On February 18, 2011, the 17th chamber of the Paris criminal court acquitted him of racial defamation for that sentence, "shocking" perhaps but not "defamatory." In a second case over the same words, it convicted him of provocation to racial discrimination: a suspended €1,000 fine.

I'm not defending the sentence. Offered as a reason to stop people by appearance, it tells a whole group it deserves different treatment. But notice the oddity. Nobody in that courtroom could have checked whether it was true, because the figures that would confirm or refute it are not compiled. The court judged the provocation because the fact itself had been made unknowable. That is the French arrangement: you may not say it too bluntly, and no one may measure it.

The serious objection

The case against counting is not stupid. Ethnic categories can harden into identities, feed stigma, invite quotas; the CNIL has said it remains "reserved" about any ethno-racial classification. Fair enough.

Look across the Channel. Under Section 95 of the Criminal Justice Act 1991, Britain has published criminal justice statistics by ethnicity since 1992. In the 2024 edition, Black adults are 4 percent of the population of England and Wales and 12 percent of the prison population. And the report warns up front that "no causative links can be drawn," since income, geography and offense mix are not controlled. Britain counts and argues on the same page. Counting did not end the argument. It gave it a floor.

Stigma does not wait for statistics. It feeds on their absence: where numbers are missing, a whole group pays for the acts of a few, because no table shows how few.

The rabble above and the rabble below

So who gains from the fog? Not the Toulouse shopkeeper. Not the great majority of immigrants' children who study, work and get lumped in with whoever walked off with a television. Two groups gain. Below, the delinquents, of whatever origin, hidden in averages nobody may break down. Above, a governing class, left and right alike, that prefers not to know, because knowing would oblige it to act: on an immigration it neither controls nor chooses on merit, on neighborhoods it let slide. The rabble above lives comfortably with the rabble below, as long as nobody counts.

That is the real problem. I judge acts, not origins. Saturday's looters may turn out to be militants, teenagers, opportunists; in Toulouse, the prosecutor said most of the 107 people held had no prior record with the courts. Whatever the answer, I want it in a table, with definitions and margins of error, not at a café counter. A country that forbids itself to count forbids itself to debate. It does not abolish the question. It hands it to whoever shouts loudest.