Data as Economic Asset
Coyle has framed the digital surplus problem by asking a deceptively simple question: is data more like oil, air, fish, or wine? Each analogy implies a different economics. Oil is rival and depletable — my use diminishes yours. Air is non-rival and non-excludable — a public good that markets systematically underprovide. Fish are rival but renewable — requiring management to prevent depletion. Wine improves with age but requires investment in production. The answer matters because it determines the appropriate regulatory and measurement framework. Her conclusion, developed in her Daedalus essay on socializing data, is that data exhibits characteristics of all four, which means that no single existing framework is adequate.
In the AI Story
The framework dissolves the confident analogies that dominate contemporary data policy. 'Data is the new oil' implies rival extraction and scarcity economics — the framework that justifies proprietary ownership and market pricing. But data is not depleted by use. 'Data is a public good' implies non-rivalry and non-excludability — the framework that justifies open access and public provision. But data can be excluded and data access generates private value. Neither analogy captures the full character of the asset.
Coyle's pluralism extends to data governance. Different data types require different institutional frameworks. Personal data resembles private property in some dimensions and collective resource in others. Aggregate data trained into AI models resembles a public good in its non-rivalry but a private asset in its extraction. The regulatory framework must be correspondingly plural — property rights for some dimensions, commons governance for others, public provision for still others.
The framework applies directly to AI governance. AI models are trained on data whose character varies across these analogies. The value captured by model providers depends on property-like excludability of trained weights. The value generated for users depends on public-good-like non-rivalry of access. The sustainability of the system depends on fish-like management of the training data commons. The quality of the output depends on wine-like patient investment in data curation. A governance framework built on any single analogy will systematically mis-regulate the dimensions the analogy cannot see.
For the AI-revolution reader, the data framework provides the analytical foundation for Coyle's advocacy of public digital infrastructure — a CERN for generative AI, data commons, interoperability mandates. These are not generic pro-regulation positions. They are specific responses to the character of data as an asset that resists governance through any single framework.
Origin
The framework appears in Coyle's Daedalus essay 'Socializing Data' (2022), in Cogs and Monsters (2021), and in her policy advocacy through the Bennett Institute's Programme on AI and Economic Progress. It synthesizes insights from the commons literature (Elinor Ostrom), the information goods literature (Hal Varian), and the data governance literature (Jonathan Zittrain).
Key Ideas
Four analogies. Data is simultaneously like oil (asset), air (public good), fish (renewable resource), and wine (patient-investment good).
Composite character. No single framework captures all dimensions; governance must be correspondingly plural.
AI training implications. Training data commons, model excludability, and access non-rivalry each require different institutional responses.
Public infrastructure case. The composite character justifies interventions — CERN-like public AI, data commons, interoperability — that pure property or pure public-good frameworks cannot.
Appears in the Orange Pill Cycle
Further reading
- Diane Coyle, 'Socializing Data', Daedalus 151(2), 2022
- Elinor Ostrom, Governing the Commons: The Evolution of Institutions for Collective Action (Cambridge University Press, 1990)
- Carl Shapiro and Hal Varian, Information Rules: A Strategic Guide to the Network Economy (Harvard Business Review Press, 1998)