Next to each other is not because of each other

This research can describe who sells the technology, who buys it, what they paid, when the contract renews, which law applies and how mature each rollout is. Across eleven tables and 1,922 records, the person a system is used on appears in exactly one place: as someone who sued. Rollout maturity is even measured on a scale that runs from innovator to laggard — the buyer's vocabulary, end to end.

73 of the 87 vendor-published sources in this research carry the top evidence tier. That is defensible for a fact like which company won which contract, and it is not the same as independent confirmation. Stated here because a reader deserves to weigh it, and because the limits of a record are part of the record.

Add up every value in this dataset and you get about $102.8 billion. Roughly a third of that is not government money at all — it is companies announcing their own investments: a data-centre expansion in Saudi Arabia, a stake bought in another firm. And one $9 billion US defence contract is counted four separate times, once for each supplier on it. We are pointing this out about our own dataset because anyone quoting the total as government spending would be wrong.

The timeline shows laws and deployments together, and it is tempting to read cause into the order they appear. The data does not support that reading. The fields built to link one event to another were never filled in, and the deployment records carry no date at all. Thirty-two events are tagged with labels like “law follows capability”, but those tags point at nothing. Treat the timeline as two stories shown side by side, not as one causing the other.

A country reporting few suicides may have few suicides, or may not be counting them. The West Bank & Gaza record 0.65 deaths per 100,000 — which would be the lowest rate on earth by a wide margin, and much more plausibly measures a fragmented registration system in a region where the death is heavily stigmatised. Russia's falling rate runs alongside a rising share of deaths filed as “undetermined intent”. India's official figures are police reports; verbal-autopsy studies find substantially more. In at least 24 countries suicide or its attempt is a criminal matter, which suppresses both help-seeking and recording. WHO's own position is that most member states lack vital registration good enough for this purpose, and that roughly one suicide in six goes missing worldwide — one in three in lower-income countries. The register that documents this is not a footnote to the chart. It is the finding: a low number is sometimes a fact about a country, and sometimes a fact about its filing.

It is tempting to read prescription volume as a thermometer for how ill a population is. The record does not support that, in either direction. In England, antidepressant items rose 50% in nine years while hypnotic and anxiolytic items FELL 16% over exactly the same period, from the same prescribers under the same system. In the United States, antipsychotic use among adults rose from 1.9% to 3.0%, while among children and adolescents it fell, 1.3% to 1.1%. And where a national registry lets diagnosis be counted directly, Denmark's new schizophrenia diagnoses went slightly down, 1.8 to 1.6 per 10,000, across eighteen years in which antipsychotic prescribing rose almost everywhere it was measured. Prescribing moves for its own reasons — guidance, capacity, recognition, duration of treatment, the licensing of new drugs, deliberate deprescribing campaigns. Sometimes it tracks illness. Here it demonstrably moves in opposite directions at once.