High Modernism

In the AI Story

Scott's four conditions for high-modernist catastrophe — high-modernist ideology, an authoritarian state willing to act on it, a prostrate civil society unable to resist, and a failed feedback loop that prevents correction — map uncomfortably well onto contemporary AI deployment patterns. AI governance is being developed by technical experts armed with confidence in their frameworks. Implementation is often authoritarian in practice if not in name, with affected populations having limited voice in deployment decisions. Civil society is often unable to resist because the tools are deployed faster than democratic deliberation can catch up. And feedback loops are often weak or absent, with outcomes measured by technical metrics rather than by the populations affected.

The connection between high modernism and Toyama's framework runs through the concept of context. High-modernist schemes fail because they treat context as friction to be eliminated rather than as substance to be understood. The schemes impose abstract categories (the standard forest, the ideal city, the optimized farm) on contexts whose specific characteristics the categories cannot capture. When the schemes are implemented, they destroy the local knowledge that had been invisible to the planners, and the systems fail. The same pattern appears in AI deployment: the tools are designed for abstract users in abstract contexts, and their deployment in specific contexts produces outcomes that track the local conditions the designers could not see.

Scott's alternative to high modernism is not anti-planning but a particular kind of attention to local knowledge and local agency. He argues for planning that begins with and remains accountable to the specific practical knowledge — mētis — of the people who actually live in the contexts being planned. This is close to Toyama's prescription for AI: invest in the human and institutional capacity that the technology will amplify, with the participation of the communities who will live with the outcomes.

The high-modernist temptation in AI is particularly strong because AI systems appear to offer precisely what earlier high-modernist projects could not: the ability to aggregate and process information at scales that make central planning seem feasible. The appearance is deceptive. The information aggregated is always already simplified by the categories that make aggregation possible, and the categories smuggle in the assumptions of the people who designed them. The more sophisticated the aggregation, the more invisible the simplification, and the more likely the resulting plans are to repeat the failures Scott documented.

Origin

The concept was developed by James C. Scott in Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed (Yale, 1998). Scott's framework has become foundational in critical development studies and has been increasingly applied to technology and governance questions in the AI era.

Key Ideas

Confidence in centralized planning. High modernism is the belief that expert knowledge applied from above can redesign complex systems.

Legibility and simplification. The state can act only on what its categories make legible, and those categories systematically simplify — and often destroy — what they fail to capture.

Four conditions for catastrophe. The combination of ideology, authoritarian capacity, prostrate civil society, and failed feedback produces the predictable failure mode.

Metis as alternative. Local practical knowledge is the form of understanding that high modernism cannot capture and systematically destroys.

AI as current iteration. AI governance and deployment often reproduce high-modernist patterns at unprecedented scale, with the same structural features and predicted outcomes.