Dear friends,

This month, I have news about Unlocking Justice in the European press, proposed changes to the U.S. census, and a California law that lets system-involved individuals challenge racial bias with data.

The book crosses the Atlantic.

On September 25, the Dutch newspaper NRC published a feature by Karel Berkhout under a headline that translates roughly as "An American mathematician uses data to show how unjust the legal system can be." It opens with the Hitler photograph that hung in an officer's locker at my town's police station for nearly twenty years. It goes on to cover my research on calls to the police, bail decisions in New York, and federal sentencing, as well as the judges' counterattack against the bail study. Read the feature in Dutch.

The article also discusses the case of Dutch nurse Lucia de Berk, whose wrongful conviction rested in part on a flawed probability calculation. Statisticians helped expose its flaws, and her conviction was eventually overturned. To prevent such errors, NRC notes, every Dutch court now has forensic advisers, scientifically trained staff who help judges interpret expert evidence, statistics included. The program reached every court in the country by the end of 2020. American courts, take note.

In August, D, la Repubblica's weekend magazine, published my conversation with Giuliano Aluffi. The print title was "Se i conti non tornano," or "If the numbers don't add up." We talked about whose work appears in museums, who edits academic journals, and how data can expose discrimination in the criminal legal system. Read the interview in Italian.

Unlocking Justice also got a July recommendation from Arab News. The book is finding readers in places I only dared dream of.

Who gets counted.

Researchers at UCLA's Williams Institute documented roughly 360 federal data collections that removed questions or answer options about sexual orientation or gender identity between January 2025 and January 2026. Most involved gender identity. About 83% were processed as "non-substantive" changes, which do not require a public comment period. Read the report.

One affected survey is the National Crime Victimization Survey, which asks people about crimes they have experienced, including crimes they never reported to police. It removed the questions about respondents' gender identity, which makes it harder to measure violence against transgender people. See the survey questionnaire.

Then there is the census. On September 10, the Census Bureau proposed a rule with two parts. The first sets the rules for who counts and where. The second sets the rules for what the census asks.

Under the first part, the population count used to decide how many seats each state gets in the U.S. House of Representatives would exclude foreign citizens unless they also hold U.S. citizenship or lawful permanent resident status, commonly called green-card status. The exclusion would cover foreign citizens living here legally on temporary visas as well as those without legal immigration status.

To decide where, and whether, to count someone, the rule leans on government records. It says a person's usual residence "should be consistent with, and evidenced by, their tax records," and the bureau says it is considering linking tax returns, Homeland Security files, and other administrative records to determine legal status. It also invites comment on adding a legal-status question to the census form. Each of these shifts the count from people who are here to people who appear in government files, and the people missing from those files are not a random sample. Many low-income households file no tax return. During the first Trump administration, the bureau's own researchers estimated that a citizenship question would depress responses from households with noncitizens, and the Supreme Court blocked the question in 2019 after finding the stated reason for it was contrived. An undercount in a neighborhood costs everyone who lives there, because House seats and federal funding follow the count.

The same part of the rule would write into regulation the practice of counting incarcerated people at the prison or jail where they are held rather than at home. That choice moves political representation away from the communities people come from, which are disproportionately Black and Latino, and toward the mostly rural districts where prisons are built. More than a dozen states have stopped doing this when they draw their own legislative maps. The federal rule would lock the practice into the census itself.

The second part would prohibit questions about race and ethnicity on the census questionnaire. The proposal's own summary says the count should not be distorted by "questions about immaterial personal characteristics, such as race." Those questions would still be allowed on other surveys, including the American Community Survey, which collects information from a sample of households rather than counting everyone, and the bureau could continue drawing demographic information from existing records. One casualty would be a change that took years to win. In 2024, the federal government adopted a separate Middle Eastern or North African category in its race and ethnicity standards, after earlier standards had folded those origins into the "white" category. Agencies' deadline to implement the new standards has already been extended to September 28, 2029. If the ban is adopted, the checkbox would not appear on the 2030 census form at all.

These are still proposals. The general public comment deadline is October 13. Read the proposal and its instructions for submitting a comment. The docket number is USBC-2026-0628.

Challenging racial bias in California's courts.

In 1987, the Supreme Court decided the case of Warren McCleskey, a Black man in Georgia sentenced to death for killing a white police officer during a robbery. His lawyers presented a study of more than 2,000 Georgia murder cases showing that, after accounting for hundreds of other factors, defendants charged with killing white victims were more than four times as likely to be sentenced to death as those charged with killing Black victims. In McCleskey v. Kemp, the Court ruled five to four that the pattern was not enough to establish that McCleskey had been denied equal protection under the Constitution. He had to show that the prosecutors or jurors in his own case had intentionally discriminated against him. He could not, and Georgia executed him in 1991.

California's Racial Justice Act, passed in 2020, gives system-involved individuals another way to challenge racial bias. They can use evidence of racial differences in charges and sentences without having to prove intentional discrimination. Those differences do not automatically establish a violation. Courts must examine the evidence, compare people in similar circumstances, and consider whether there are explanations unrelated to race. The law also requires courts to consider how racial profiling and longstanding bias in policing and prosecution may have shaped the data.

This January, the final phase of the law's expansion to older cases took effect. It now permits challenges to felony convictions regardless of when the cases became final. Read the current law.

This connects directly to my work with collaborators on which California arrests lead to prosecution and how that varies by race. The paper is forthcoming in the Journal of Gender, Race & Justice, and I'll share more when it's out.

Reviews and a small request.

The statistics magazine Significance called the book "a wise and witty companion," and Harvard Magazine described it as "a narrative bridge between data-driven research and social justice efforts." In The American Mathematical Monthly, Kimberly Spayd wrote that "the writing and publishing of Unlocking Justice is a political act in the world we live in today: a proud invitation to work towards justice and equity in the criminal justice system using math and statistics, made accessible for a wide audience." I'm grateful for these reviews.

If you teach and are considering the book for a spring course, you can request a copy to review from Princeton University Press.

If you've enjoyed the book, please tell someone you think would like it. A rating on Amazon or Goodreads helps, too. Word of mouth is how this book finds its readers.

You can also find the book at Bookshop.org, Barnes & Noble, and Princeton University Press.

Perth in December.

In early December, I'll give the Dr Mandawuy Yunupingu Lecture at the Australian Mathematical Society's annual conference in Perth. I'll write more about the visit in a future newsletter.

Thank you, as always, for reading, and for caring about justice.

With gratitude,
Chad