Conventions (Becker)
Becker's central insight about how creative worlds actually function: conventions are not rules in the sense laws are rules. Nobody writes them down. Nobody enforces them through explicit sanctions. They are maintained through practice — through the accumulated weight of thousands of instances in which people did things one way rather than another and found the one way worked. A convention persists because it solves a recurring problem. It disappears when the problem it solves ceases to exist. Software development runs on conventions as thoroughly as jazz: the sprint, the code review, the pull request, the frontend-backend division. Each solved a coordination problem and, in solving it, constrained what could be created and by whom. AI tools do not merely add capability — they destabilize the conventions themselves by dissolving the constraints those conventions were designed to address.
In the AI Story
Conventions are invisible in the precise sense that water is invisible to a fish: not because they are hidden but because they are everywhere, so thoroughly integrated into practice that separating them from the work itself requires deliberate analytical attention. A working musician in 1950s Chicago did not think of herself as following conventions. She thought of herself as playing music. The chord changes, the turn-taking protocols for solos — none of these registered as conventions.
Software conventions are historical products, not natural facts. The sprint emerged from Agile methodology, itself a 2001 reaction against waterfall development conventions. The frontend-backend division stabilized when web applications grew complex enough that a single developer could not competently handle both. Each convention solved a real problem — and each constrained what could be created and by whom.
What happened in Trivandrum in February 2026 was not primarily a demonstration of AI capability. It was a convention collapse. The division into frontend and backend, the spec-to-implementation handoff, the role boundaries — dissolved because the translation cost they addressed no longer existed. Under old conventions, a backend engineer building a frontend feature was violating a shared understanding about who does what.
The discourse between triumphalists and elegists that Segal describes is, in Becker's vocabulary, a contest between competing convention sets. The triumphalist convention organizes around speed, individual capability, and output volume. The elegist convention organizes around craft, embodied expertise, and the slow accumulation of depth. Both solve real problems. Neither is natural or inevitable.
Origin
Becker developed the concept of conventions across his studies of jazz, photography, and academic life. It received its fullest treatment in Art Worlds (1982), where conventions are presented as the social infrastructure without which cooperative creative production is impossible.
The framework drew on Becker's earlier work on labeling theory in Outsiders (1963), which showed that deviance is produced by social conventions about what counts as deviant rather than by properties of acts themselves.
Key Ideas
Conventions solve coordination problems. A jazz musician can sit in with a band she has never met because she shares the conventions of the genre: chord progressions, turn-taking, rhythm section freedom. Without conventions, the session is chaos.
Technologies destabilize conventions by eliminating their underlying constraints. Photography did not merely add capability to portrait painting — it dissolved the conventions that addressed the absence of other means of realistic likeness.
Conventions feel like facts to those inside them. The convention that frontend and backend are separate specialties felt natural to developers because the translation cost between human intention and machine execution made it rational.
New worlds experience convention instability. Traditional art worlds have conventions refined over decades. The AI world is establishing conventions in real time, under pressure, by participants who are simultaneously using tools and arguing about whether the tools should exist.
Conventions are social, not technical. Which conventions stabilize depends on which participants adopt them, which institutions reinforce them, and whose interests they serve — not on technological optimality.
Debates & Critiques
Some critics treat Becker's emphasis on conventions as relativism — a denial that some conventions are better than others. Becker's actual position was more modest: conventions are social products whose consequences can be evaluated, but the evaluation itself proceeds through further conventions. The point is not that anything goes but that making conventions visible is the precondition for deciding whether to keep them.
Appears in the Orange Pill Cycle
Further reading
- Howard Becker, Art Worlds, Chapter 2 (University of California Press, 1982)
- Howard Becker, Outsiders: Studies in the Sociology of Deviance (Free Press, 1963)
- Kent Beck et al., Manifesto for Agile Software Development (2001)
- Andrew Abbott, The System of Professions (University of Chicago Press, 1988)
- Etienne Wenger, Communities of Practice (Cambridge University Press, 1998)