Formal Access vs. Substantive Capability

In the AI Story

The distinction was not invented for AI; it emerged from the documented failure of every previous wave of technology democratization. The One Laptop Per Child initiative distributed millions of devices and produced, in contexts without supporting educational infrastructure, null or negative learning outcomes. Mobile banking was celebrated as financial democratization, but its transformative effects in Kenya (M-Pesa) depended on Safaricom's existing distribution network, Kenya's regulatory environment, and financial literacy levels — conditions unevenly present in countries where mobile money has been deployed with less success. Social media promised a level playing field of voice and produced algorithmic amplification of existing power. In each case, formal access was achieved. Substantive capability was not.

Applied to AI, the distinction yields a specific prediction: the distribution of tools like Claude Code to five billion connected humans will produce genuine benefits for those whose educational systems, institutional contexts, and professional networks provide the foundations that the tools amplify, and will produce formal participation without substantive capability for those whose foundations are weaker. The outputs will diverge along lines that track existing inequality, and the divergence will be larger than it would have been without the tool — because amplification is faithful and the foundations are unequal.

Segal's The Orange Pill acknowledges the distinction in passing — 'access requires connectivity, and connectivity requires infrastructure that billions of people do not have' — but the acknowledgment occupies a paragraph in a chapter whose emotional arc celebrates the rising floor. Toyama's framework insists that the acknowledgment is not a caveat but the central argument. The infrastructure gap, the educational gap, the institutional gap, the cultural-capital gap are not asterisks on the democratization thesis; they are the mechanism that determines whether democratization is real or rhetorical.

The distinction has specific policy implications. Measuring AI democratization by subscription counts, connectivity statistics, or tool-access metrics produces a flattering number that bears little relation to outcomes. Measuring by substantive capability — the capacity of users to produce output that meets professional standards, reaches relevant markets, and generates economic or developmental value — produces a much harder-to-construct but much more honest metric. The technology industry's reporting strongly favors the first. Toyama's framework argues that the second is the only one that matters.

Origin

The distinction has roots in Amartya Sen's capability approach to welfare economics — Sen's insistence that the evaluation of development must attend to what people can actually do and be, not merely to what resources they nominally possess. Toyama extends this framework to technology, arguing that the capability approach is the appropriate standard for evaluating technology democratization claims. The full articulation appears in Geek Heresy and across his subsequent essays on AI.

Key Ideas

Binary vs. continuous. Formal access is binary (have it or not). Substantive capability is continuous and multidimensional.

Distributed vs. developed. Formal access can be distributed cheaply and quickly. Substantive capability is developed slowly and expensively through structural investment.

The conflation. The democratization narrative routinely measures the first and claims the second, generating flattering metrics that correspond weakly to outcomes.

The capability approach. Sen's framework provides the rigorous standard for distinguishing formal from substantive democratization and for evaluating whether technology expansions actually expand human capability.

Predictive value. The distinction predicts, with empirical support across multiple technology transitions, that formal democratization without parallel investment in substantive capability reproduces and often widens existing inequality.