A Prehistory of the Cloud
Source: https://mitpress.mit.edu/9780262529969/a-prehistory-of-the-cloud/ ↗
Hu is a former network engineer turned literature professor, and the book reflects both formations.
He traces how the metaphor of the "cloud" inherits older infrastructural imaginaries — railways, pneumatic tubes, Cold War bunkers, virtualized time-sharing systems — and argues that each inheritance carries political assumptions about centralization, sovereignty, and user passivity.
The analysis moves from nineteenth-century railroad networks through SAGE air defense to contemporary data centers, showing that the fantasy of dematerialized computing has deep roots and persistent consequences.
Hu is especially sharp on how "the cloud" naturalizes surveillance and obscures the labor and energy that sustain it.
For product people, the book reframes cloud infrastructure not as a neutral platform but as a historically loaded metaphor that shapes what we imagine is possible.
It is the kind of argument only someone who has both configured routers and read Foucault could make.
Central argument
Hu argues that 'the cloud' is not a novel technological paradigm but a metaphor that inherits the political and infrastructural logic of older systems — railroads, Cold War air defense networks, time-sharing mainframes — each of which encoded assumptions about centralization, state sovereignty, and passive users. The central thesis is that this genealogy is not merely historical curiosity: the fantasy of dematerialized, frictionless computing actively obscures the physical infrastructure, labor, and surveillance architectures that sustain it. By naturalizing these inherited assumptions, the cloud metaphor constrains what designers, engineers, and policymakers imagine as technically or politically possible.
Critique
Hu's archaeological method, moving from SAGE to AWS via literary and Foucauldian analysis, is analytically rich but tends to flatten the genuine discontinuities between historical systems whose political economies differ substantially. A serious objection is that demonstrating genealogical continuity in metaphor does not necessarily establish causal inheritance in actual technical or political practice — the fact that data centers and Cold War bunkers share an aesthetic of concealment does not prove they reproduce identical power relations. Practitioners may find the argument structurally overdetermined: if every infrastructural layer is read as encoding domination, the framework risks becoming unfalsifiable and offers limited traction for distinguishing better from worse design choices.
Why it matters for product
For a CPO, the book's sharpest implication is that decisions framed as purely technical — which cloud vendor to use, how to architect data storage, what telemetry to collect — are never neutral, because the infrastructure itself encodes assumptions about user agency and organizational control that shape product possibilities before any feature decision is made. Hu's argument about user passivity being built into time-sharing architectures is directly relevant to how product teams reason about data ownership, consent models, and the distribution of control between platform and end user. It reframes vendor lock-in debates from a commercial question into a structural one: what kind of sovereignty over user experience is actually available given the inherited logic of the platform you are building on.