How Infrastructure Capture Repeats Across Every Technological Revolution
We were asking the wrong question again.
Every major infrastructure technology — railroads, electricity, telecom, the internet — follows the same pattern: a small group captures control, aggregates extraordinary power, and reshapes society before anyone thinks to ask the right questions. With AI, this concentration is more extreme and more intimate than ever before.
The Translation
AI-assisted summaryFamiliar terms
This analysis identifies a structural recurrence across every major infrastructural transformation in industrial civilization: a technology emerges with democratizing promise, a small cohort — often exhibiting Dark Triad personality traits — captures control of the new infrastructure, and that control is leveraged into asymmetric power that eventually subsumes political governance. The railroads are the canonical case, where operators recognized that controlling the logistics layer meant controlling the economy, leading to price-fixing, coercion, and Rent Extraction until antitrust intervention. The same arc repeated with electricity, telecommunications, the internet, and social media. Each cycle, the critical political question — how long until infrastructure controllers capture the state? — went unasked until the capture was well advanced.
The argument is that AI represents an intensification of this pattern, not a departure from it. The core AI labs constitute an unprecedentedly small group of identifiable individuals who are ingesting the Intimate Conversational Data of hundreds of millions of users under assumptions of anonymity, while simultaneously running behavioral experiments on those users — experiments they openly publish. Facebook's documented capacity to induce depression in user subgroups or influence voting behavior through algorithmic manipulation provides the empirical precedent. There is no rational basis for assuming equivalent manipulation is not occurring within AI systems, and considerable rational basis for assuming it is.
The framing rejects the utility question as a distraction. Whether AI is useful is not in dispute. The operative question is whether this civilization will once again fail to interrogate the power dynamics of a new infrastructure until the window for meaningful intervention has closed — a failure that, given the intimacy and scale of AI's data access, may prove irreversible in ways previous cycles were not.