My Flagship Publication

Maths Casualties is a long-form investigative newsletter about what happens when math leaves the chalkboard and starts making decisions about real people.

It treats mathematics not as neutral truth, but as a power tool: a way institutions justify resource allocation, policing, triage, credit, insurance, medicine, education, war, and “risk.” The core question is always: who gets harmed when a model is treated like reality?

• Forensic case files on specific models, metrics, and “evidence-based” systems that produced measurable harm.
• Autopsies of assumptions: what got simplified away, what proxies stood in for people, what was “held constant,” what was never measured.
• Incentive analysis: how funding, publication, regulation, and corporate liability shape what gets modeled—and what gets ignored.
• Method critiques that non-specialists can still follow (without dumbing anything down): confounding, selection bias, external validity, distribution shift, measurement error, causal vs correlational claims.
• Accountability work: not vibes, not “AI ethics” platitudes—paper trails, technical claims, implementation details, and who benefited.

Skeptical, evidence-forward, and willing to name mechanisms (and the people who deploy them) without hiding behind academic politeness. Uncertainty is labeled clearly, but “uncertainty” is never used as a shield to avoid responsibility.

If a system claims “the math says”, Maths Casualties asks:

• Which math?
• On whose data?
• Under what assumptions?
• With what error bars?
• And who pays when it’s wrong?