Governance of the Surplus

In the [YOU] on AI Field Guide

The cycle’s central argument is that the orange pill enables clear-eyed action rather than paralysis or intoxication, and governance of the surplus is where clear-eyed action is most urgently required. The tools of creation are now broadly available. The motive for using them—the intrinsic satisfaction of building, the imagination-to-artifact ratio collapsing to the width of a conversation—is powerful and durable. What is underdeveloped is the opportunity: the social and institutional structures that channel individual creation toward collective value. Governance of the surplus is the construction of that opportunity layer.

The cycle introduces the concept through Shirky’s analysis but gives it teeth through the specific failure modes that AI-generated software introduces. The lolcat problem scales: when a billion people can build software through conversation, the volume of experimental-phase output overwhelms any discovery mechanism designed for the first surplus. And unlike a badly written blog post, a badly built software tool can cause real damage—mishandling user data, producing incorrect results in safety-critical contexts, or failing at edge cases the creator never tested because the code was generated by a system whose internal logic was opaque to the person directing it.

Origin

The concept developed through Shirky’s analysis of the first surplus’s governance evolution. Wikipedia’s early years were characterized by unsustainable intensity among its most prolific editors, many of whom burned out before the community developed norms that supported sustainable contribution. These norms did not emerge spontaneously; they were developed through years of experimentation, conflict, and deliberate institutional negotiation. Open-source licensing was hammered out through decades of philosophical argument. Stack Overflow aggregated individual answers into a searchable knowledge base through platform design choices that made quality visible.

Each of these institutions solved a version of the discovery problem through a combination of platform design, community norms, and governance structures. The second surplus requires analogous institutions designed for the specific characteristics of AI-enabled creation: complex artifacts produced by creators of widely varying technical sophistication, with failure modes invisible to surface-level evaluation and liability that is distributed across human creator, AI model, model developer, and hosting platform in ways that existing legal frameworks cannot cleanly accommodate.

Key Ideas

Tiered Quality Standards. No single quality threshold serves the full range of AI-enabled creation. A personal utility built for the creator’s own use needs no external standard. A tool shared within a community that understands its limitations needs modest standards of reliability and transparency. A tool handling sensitive data or operating in safety-critical contexts must meet standards approaching those of professional software development. Governance that applies a single threshold either suppresses the experimental surplus or fails to protect users from consequential failures. The institutional challenge is tiering that is sensitive to context without being so complex that it stifles creation.

Distributed Liability. The question of responsibility when AI-created software malfunctions cannot be resolved by extending existing frameworks, which assign responsibility based on control. In AI-enabled creation, control is distributed: the human creator directed the AI but may not understand the code; the AI generated the code but did not choose the purpose; the model developer enabled the creation but did not direct it; the platform distributed it but did not produce it. The governance frameworks that served concentrated professional production do not apply to distributed amateur creation, and the institutions that serve the second surplus must develop new liability architectures adequate to the distributed agency of the new production landscape.

Platform Governance as Power. The platforms that host, distribute, and facilitate discovery of AI-created software will exercise decisive influence over the surplus—determining what gets created, shared, discovered, and used. Platform governance is the most consequential and least visible form of power in any participatory ecosystem. The Shirky Principle predicts that existing platform operators will attempt to extend their current governance frameworks to the new production landscape, and that this extension will capture surplus value for platform operators while failing to serve the diverse community of creators the surplus encompasses.

Ascending Governance. Just as the ascending skill barrier relocates difficulty rather than eliminating it, ascending governance relocates regulatory activity from restricting creation to evaluating it. Governance that restricts who can create suppresses the surplus and forfeits its value. Governance that evaluates what has been created—providing quality assurance, liability frameworks, and platform accountability—channels the surplus toward collective value without suppressing the creative energy that produces it.