Rise of the Machines

Thomas Rid published Rise of the Machines in 2016, roughly three years after his earlier Cyber War Will Not Take Place had argued — against a chorus of alarm — that the strategic category of “cyber war” was overused and under-specified. Where that first book was a corrective, Rise of the Machines is an archaeology. Rid, then a professor in the Department of War Studies at King’s College London and later a chaired professor at the Johns Hopkins School of Advanced International Studies, set out to trace the long intellectual arc that leads from an anti-aircraft prediction problem in 1941 to the state-sponsored intrusions and cryptographic subcultures of the twenty-first century. The common thread he follows is cybernetics — a word now almost quaint in English, but once a serious ambition to unify biology, engineering, communication and command under a single theory of feedback and control.

The book arrived in a period of intense public argument over cyber conflict, mass surveillance after the Snowden disclosures, and the political reach of Silicon Valley. Earlier works by Fred Turner, George Dyson and Walter Isaacson had already shown that the digital present had a surprisingly countercultural past; what Rid adds is an insistence that the technology and the myth cannot be separated. For every working machine built in the name of cybernetics, he shows, there was a parallel story — about transcendence, about autonomy, about the fusion of human and machine — that did more cultural work than any particular device. The book is thus both intellectual history and reporting, written for a general reader but drawing on archival research, declassified cables, and interviews with figures who built or wrote about the systems in question.

The central argument is that cybernetics has been less a stable science than a recurring vision, and that this vision has been remarkably promiscuous — equally comfortable in a Pentagon briefing room, a Marin County commune, a cryptographer’s conference and a Russian intelligence directorate. Rid resists both utopia and panic. The machines, he writes, have not risen in the melodramatic sense; what has risen is a set of beliefs about them, periodically renewed, that cause their human users to misread what the devices actually do. The point is not to debunk cybernetics but to show how its promises recur — autonomous weapons, perfect intelligence, networked emancipation, cryptographic sovereignty — and how reality tends to overtake each promise obliquely, delivering something adjacent to what was wanted and often more consequential.

The book is organised chronologically rather than thematically, which gives its chapters a readable momentum. It opens with Norbert Wiener at MIT during the Second World War, working with Julian Bigelow on the problem of aiming an anti-aircraft gun at a maneuvering bomber. The AA predictor never became a fielded weapon, but the attempt — in which pilot, aircraft, radar, and gun had to be modelled as a single feedback loop — produced the conceptual apparatus Wiener would unveil in 1948 as Cybernetics: Or Control and Communication in the Animal and the Machine. From Wiener, Rid walks the reader through the Macy Conferences of the late 1940s and early 1950s, where Wiener, John von Neumann, Warren McCulloch, Margaret Mead and Gregory Bateson tried to build a common language across neuroscience, anthropology, mathematics and engineering. He then moves to SAGE, the Semi-Automatic Ground Environment, the vast continental air defence system that grew out of Project Whirlwind at MIT and taught the US military and its contractors how to build real-time networked computing.

From the military mainframe the book pivots to its counterculture inheritance. Rid devotes substantial attention to Stewart Brand and the Whole Earth Catalog, to the Bay Area communes that read Wiener and Buckminster Fuller together, and to the slow process by which engineers from defence laboratories became the priesthood of a new, decidedly anti-authoritarian technics. The term cyborg, coined in 1960 by Manfred Clynes and Nathan Kline in a paper on how humans might be modified for space travel, is one of the small set pieces Rid uses to show the migration: a word intended for suited astronauts becomes, within a generation, a label for a political identity. A long chapter on virtual reality follows the arc from early flight simulators through Ivan Sutherland’s head-mounted display to Jaron Lanier’s VPL Research, with the shift of commercial energy to immersive consumer experiences in the late 1980s. Rid is careful to show that the hype of that moment collapsed before it delivered, and that the next wave — networked rather than immersive — came from a very different direction.

That next wave is the cypherpunks, and this is where the book’s connective tissue to the modern cyber story becomes clearest. Rid tracks the emergence of public-key cryptography out of academic mathematics (Whitfield Diffie, Martin Hellman, the RSA trio) and the ideological project that grew around it — Timothy May’s crypto-anarchy manifesto, Eric Hughes and John Gilmore and the Cypherpunks mailing list, David Chaum’s digital cash experiments, and the Clipper chip dispute that defined the first crypto war of the 1990s. He treats these movements not as a prehistory of Bitcoin but as the moment when the cybernetic promise crossed into politics: cryptography offered the dream, long implicit in cybernetics, that an individual armed with the right mathematics could stand outside the state’s reach. The dream never arrived in the shape its advocates wanted, but it reshaped what both corporations and intelligence agencies thought was possible on a network.

The last stretch of the book is where Rid does his most original reporting. He gives one of the fullest accounts yet published of Moonlight Maze, the long intrusion into US government and research networks that was detected in 1996 and attributed, cautiously, to Russian state actors. Working with researchers who had kept artefacts of the investigation for twenty years, Rid shows that Moonlight Maze was not an anomaly but an early mature instance of the kind of persistent, patient, state-sponsored espionage that would later become routine. The thread continues through the Kosovo-era information operations, the slow emergence of a public literature on offensive cyber, and the Stuxnet disclosures of 2010 that transformed the political conversation. The book does not try to deliver a full account of post-2010 cyber conflict; Rid had done that elsewhere and would return to it. Here the point is to close the arc: a problem that began as how to help an anti-aircraft gun aim at a bomber ends with the question of how a state should write software to break a centrifuge.

The specific cases and names are what make the long middle of the book memorable. Readers meet Louis Couffignal and Stafford Beer, each a now-forgotten cybernetician who in his day was read by generals and prime ministers; Beer’s Project Cybersyn in Allende’s Chile, intended as a real-time management nervous system for the economy, is treated with the mixture of fondness and skepticism it deserves. The book is attentive to the women who shaped the field and have been undercredited — Margaret Mead’s role in the Macy Conferences, the female software corps of SAGE. It attends to visual culture — the covers of science-fiction paperbacks, the aesthetic of the Whole Earth Catalog, the gothic artwork of early crypto zines — because for Rid the imagery is part of the mechanism by which the ideology spreads. On the hardware side, SAGE’s AN/FSQ-7 computers, the vacuum-tube racks of Whirlwind, the Honeywell Kitchen Computer that no one bought, and the modems that stitched the first networks together are described crisply enough that a reader without a technical background can follow what each actually did.

Reception of the book was broadly favourable in the quarters that matter for a work of this kind: historians of technology, defence intellectuals, and the serious cyber-security press. Reviewers in The New York Times, The Wall Street Journal, Foreign Affairs and the Financial Times treated it as a durable synthesis rather than a quick entry in the cyber-doom literature. Specialists in science and technology studies pushed back in places — some arguing that Rid’s treatment of counterculture flattens the differences between its strands, others that his cybernetic frame can be stretched to cover too much. His account of Moonlight Maze was greeted as a genuine contribution and became a reference point for later work on Russian intelligence operations. The book also sharpened a debate about whether cybernetics as a word is worth reviving. Rid’s position is that the word has done its work and should be retired; others, including proponents of a renewed interest in control theory inside AI research, have taken the opposite line. Since publication, Rid’s own work has moved further into the archival study of information operations, culminating in his 2020 book Active Measures, which can be read as a direct continuation of Rise of the Machines on the political-warfare side.

For a reader coming to AI and warfare now, the book is less an introduction to any current system than a map of the inherited vocabulary. Terms that governments and defence contractors use every day — unmanned, autonomous, loop, loiter, feedback, swarm, network-centric — did not fall out of the sky in 2010. They have histories, and those histories carry assumptions that still shape procurement decisions and operational doctrine. Someone reading alongside Paul Scharre’s Army of None or Kenneth Payne’s I, Warbot will find that Rise of the Machines explains why those books take the categories they take as given. It pairs especially well with Fred Turner’s From Counterculture to Cyberculture, which covers the Bay Area side at greater depth, and with George Dyson’s Turing’s Cathedral, which gives the Princeton-side computing story. It is thinner on the Soviet cybernetics tradition, on China, and on the full span of the AI winters; a reader who wants that will need to look elsewhere.

Nearly a decade after publication, the parts of the book that have aged well are the historical reconstruction — Wiener, Macy, SAGE, Moonlight Maze — and the insistence that the mythologies around machines deserve as much scholarly attention as the machines themselves. The parts that read as more of their moment are the sections on virtual reality and the specific contours of the 2010s cyber debate, which has since been overtaken by generative models, large-scale information operations, and a more aggressive public posture by both American and allied cyber commands. The book did not foresee generative AI; it was not trying to. What it offers instead is a reminder that each wave of enthusiasm about intelligent machines rediscovers questions that were asked, and sometimes answered, decades earlier. In a field prone to amnesia, that reminder has not lost its force.