The Undersea Network
Ninety-nine percent of intercontinental internet traffic travels through submarine cables, yet almost no one outside the telecommunications industry writes seriously about them.
Starosielski combines ethnography, media theory, and geopolitical analysis to trace the routes these cables follow — routes that often repeat the paths laid by telegraph companies in the nineteenth century.
The book examines how landing sites are chosen, how local communities negotiate with global infrastructure, and how the physical vulnerability of a cable on an ocean floor shapes the politics of connectivity.
Her argument is that networks are not abstract topologies but material systems embedded in specific landscapes, economies, and power relations.
For product leaders, the book makes vivid the infrastructure dependency chain that sits beneath every cloud service and every global user base.
It is a reminder that distribution is always, in the end, a physical problem.
Central argument
Starosielski argues that submarine cable networks are not neutral or abstract infrastructures but deeply material systems shaped by historical path dependency, geopolitical power, and local negotiation. A core finding is that contemporary internet cable routes largely replicate the trajectories laid by nineteenth-century telegraph companies, meaning colonial-era decisions about where infrastructure lands continue to structure global connectivity today. She shows that the physical vulnerabilities of these cables — and the politics surrounding their landing sites — actively determine who has access to the internet and on whose terms.
Critique
The book's ethnographic and humanistic methodology, while rich in texture, leaves underexplored the engineering and economic constraints that genuinely drive cable routing decisions — factors like ocean floor topography, signal attenuation, and capital concentration among a small number of consortium operators. A technically oriented reader might argue that what Starosielski frames as geopolitical choice is sometimes closer to physical necessity or financial path dependency, and that distinguishing between these explanations matters if one wants to understand what is actually changeable. The analysis risks overstating human agency in infrastructure design while underweighting the degree to which the physics and economics of the medium constrain available options.
Why it matters for product
Product leaders who build for global user bases typically model latency, availability, and reach as cloud configuration problems — selecting regions in AWS or GCP as if geography were a software variable. Starosielski's argument reframes this: those region choices sit atop a physical layer whose routing was fixed decades ago and whose single points of failure are literal ocean floors, meaning resilience and distribution strategies have hard geopolitical and material ceilings that no vendor abstraction removes. More concretely, it is a prompt to audit which user populations a product cannot reliably serve not because of product decisions but because of where cables happen not to land — a dependency that belongs in risk frameworks, not just infrastructure runbooks.