Convention Collapse

In the AI Story

Under the old conventions, a backend engineer attempting frontend work was violating a convention — not a written rule but a shared understanding about who does what. The violation would have been met not with formal punishment but with informal resistance: skepticism from colleagues, pushback from reviewers, a general sense that the engineer was operating outside her competence. Claude Code did not just make it technically possible for the backend engineer to build a frontend feature. It dissolved the convention that made such cross-boundary work illegitimate.

The dissolution is not limited to the frontend/backend division. Many conventions that organized software production — the spec-to-implementation handoff, the role boundaries between designer and developer, the assumption that a product manager writes requirements and an engineer implements them — were responses to translation costs that AI has dramatically reduced. Each of these conventions is now under pressure, and each will dissolve, transform, or be defended through institutional inertia.

Convention collapse produces predictable emotional responses. The engineers in Trivandrum oscillated between excitement and terror, not because they had changed but because the conventions that organized their professional identities were changing. The productive vertigo Segal describes is the characteristic emotional signature of a convention set in flux — the water becoming visible precisely because it is churning.

Not all conventions that are vulnerable will actually collapse. Some are defended by institutional structures that outlast their rationale. The academic convention of peer review persists long past the technological changes that transformed what could be checked and by whom. The medical convention of the physician-pharmacist division persists despite information systems that could support many patterns of collaboration. Convention collapse is a possibility when constraints dissolve — whether it actually occurs depends on political, economic, and institutional factors beyond the technology itself.

Origin

Becker developed the concept through comparative studies of technology-driven transitions in art worlds: photography and painting, recorded music and performance, cinema and theater. The pattern was consistent: new technology destabilized not just production techniques but the conventions that had organized the cooperative work of the field.

The framework has been extended to organizational studies by scholars including Paul DiMaggio, Walter Powell, and others studying institutional change under technological pressure. Recent applications to AI-era software development follow the same analytical logic.

Key Ideas

Conventions are responses to constraints. When the constraint changes, the convention's rationale weakens.

New technologies destabilize conventions, not merely augment them. The more significant effect is often social reorganization rather than capability expansion.

The Trivandrum case is paradigmatic. The dissolution of the frontend/backend division in five days illustrates how quickly AI can restructure conventions when constraints collapse.

Collapse produces predictable emotional responses. Participants experience vertigo, exhilaration, terror — the characteristic signature of conventions becoming visible as they change.

Not all vulnerable conventions actually collapse. Institutional inertia, political interests, and incomplete constraint-dissolution can preserve conventions past their original rationale.

Debates & Critiques

Some argue that what looks like convention collapse is actually gradual evolution — conventions shift slowly, and dramatic stories about sudden dissolution exaggerate the pace of change. Becker's framework allows for both: conventions can persist through technological change if defended, and can dissolve rapidly if the underlying constraint disappears and no institutional structure maintains the old pattern.