The history of surveillance.

In 1787 the English philosopher Jeremy Bentham sketched the Panopticon — a ring of cells around a central tower, lit so that a single unseen inspector could watch any inmate at any time, while no inmate could ever tell when he was being watched.

The genius was psychological, not optical: the guard need not actually watch; the prisoner need only believe he might be. Bentham meant it as a universal template — for prisons, factories, asylums, schools. In 1975 Michel Foucault revived it as the 'diagram' of modern power: watching that the watched person carries inside themselves.

Anthropologist James C. Scott called this 'legibility': the censuses, standardized names, and maps that let a state 'see' its subjects. The body itself became a record — Bertillon's mug shot in 1883, then fingerprinting across four continents.

The pattern is the tell: identity systems only ever accrete. Fingerprint, then photograph, then DNA, then face — each layer added, none discarded. As identification went electronic, Roger Clarke coined 'dataveillance' (1988) for the systematic monitoring of people through their data trails.

From Zamyatin's We (1921) to Orwell's Nineteen Eighty-Four (1949), the century made surveillance legible to the public. Orwell's telescreen was the Panopticon rendered electronic — a device that 'received and transmitted simultaneously' and could never be fully shut off.

Huxley's Brave New World offered the counter-model: control through pleasure, not fear. Together they set the terms of consent — teaching the public which surveillance to fear (the jackboot) and, by omission, which to accept (the convenience).

In the U.S., the FBI's COINTELPRO (1956–71) ran 2,370 covert actions to 'disrupt, discredit, or neutralize' dissidents. In parallel, the NSA's Project SHAMROCK copied nearly all international telegrams for 30 years, and MINARET watch-listed thousands of Americans by name.

Abroad, East Germany's Stasi built the most thoroughly informed police state in history for its size — ~91,000 officers, ~173,000–189,000 informants, and roughly 111 kilometres of files. Even when specific programs were shut down, the capability and the institutions remained.

When the Media burglary and Watergate cracked the secrecy open, the Senate's Church Committee (1975–76) exposed COINTELPRO, SHAMROCK, MINARET, and CIA abuses. SHAMROCK was terminated; the 1978 FISA required, for the first time, a court order for foreign-intelligence wiretaps on U.S. soil.

But note how quickly it re-tightened: Reagan's Executive Order 12333 (1981) expanded intelligence authorities, and the FISA Court would in time approve nearly every application. The reversal was real — the exception that proves the rule.

Within six weeks Congress passed the USA PATRIOT Act — 357–66 in the House, 98–1 in the Senate. Its Section 215 became the legal basis for bulk phone-metadata collection. DARPA proposed Total Information Awareness, to mine the records of the whole population; Congress defunded it in 2003, but exempted its tools and moved several components elsewhere.

They reappeared as 'Stellar Wind,' the NSA's warrantless wiretapping, revealed by the NYT in 2005. The permanent architecture kept accreting — DHS (2002), the DNI (2004), fusion centers, the No-Fly List, and in 2008 Section 702, the future legal home of PRISM.

The first story: a secret court order forcing Verizon to hand over all call records, every day. Then a cascade — PRISM (stored communications from nine internet firms), XKeyscore ('nearly everything a typical user does on the internet'), MUSCULAR (tapping Google's and Yahoo's private fiber), and BULLRUN (undermining encryption itself).

It also exposed a lie: three months earlier, DNI James Clapper had told the Senate the NSA did not collect data on millions of Americans — 'not wittingly.' The reckoning was partial: the USA FREEDOM Act (2015) ended bulk telephone collection, but Section 702 and upstream collection remained.

Shoshana Zuboff dates it to Google around 2001, when the company learned to monetize the 'data exhaust' of search and sell 'prediction products.' The machinery beneath is the real-time ad economy: one 2022 study found it exposes a typical American's activity and location 747 times per day.

Most consequentially, this private apparatus became a back door into the state's. Because Carpenter (2018) requires a warrant for phone-location data, agencies simply bought it instead — the 'data-broker loophole.' ICE, CBP, and the FBI purchased location data from Venntel, Babel Street, and Fog Data Science.

Its emblem is Clearview AI — 3+ billion faces scraped from the web, sold as a search engine to police. Around the citizen, a private sensor mesh has been built and rented to police: Flock's ~119,000 plate readers, Axon's acquisition of Fusus (cameras, gunshot detectors, drones into real-time crime centers).

Yet the vendors' claims collapse under audit: Chicago found only 9.1% of ShotSpotter alerts led to evidence of a gun crime; a study of 23,631 predictive-policing forecasts found a 0.1–0.6% hit rate, concentrated on Black and Latino neighborhoods. The authoritarian frontier is Xinjiang, where the IJOP platform fuses checkpoint, camera, and utility data to flag people for detention.

In 'The Vulnerable World Hypothesis' (2019), Nick Bostrom argues that if technology will eventually yield a cheap weapon of mass death, civilization's only general defense is total, pre-emptive monitoring — in some scenarios, in place before the technology is invented.

He sketches it concretely: every person wears a 'freedom tag' with cameras and microphones, uploading encrypted audio-video machine-interpreted in real time, with flagged feeds sent to 'patriot monitoring stations' staffed by 'freedom officers.' At ~$140 per person per year, he estimates, the whole world could be monitored for under 1% of GDP. Every capability this history traces is a component of the system he describes.