Reputation Systems in the AI World
Every art world has a mechanism for converting the abundance of production into a ranked hierarchy of reputation, directing the attention of audiences and the money of patrons toward work the system deems best. The French Salon. Radio airplay. Academic citations. Becker observed that reputation systems become more important as production becomes more abundant, because the ratio of what exists to what any individual can attend to grows. The system does not merely reflect quality; it constructs it. Work the system elevates is treated as good; work it ignores is treated as nonexistent. The AI world faces a reputation problem of unprecedented scale because the tools have made production so easy that volume overwhelms any existing mechanism for sorting it. Its current reputation systems — inherited from social media — are crude, misaligned with what most participants say they value, and reward qualities that may not be the qualities that matter.
In the AI Story
The mechanisms currently operating in the AI world include follower counts (measuring popularity rather than quality and susceptible to manipulation), viral metrics (measuring emotional resonance in the moment rather than lasting value), revenue figures (measuring market demand but not the quality of what is demanded), speed-of-production metrics (measuring efficiency but not whether the efficiently produced thing is worth producing), and the informal opinions of a small number of high-visibility commentators amplified by the same platforms whose metrics they compete within.
These mechanisms share a characteristic Becker identified in every art world's reputation systems: they reward the qualities the system is designed to measure, which may or may not be the qualities that matter. A reputation system measuring follower counts produces a world that optimizes for followers. A system measuring revenue produces a world that optimizes for revenue. None are inherently wrong; none measure what most participants, when asked directly, say they value most: the quality, depth, originality, and usefulness of the work.
The misalignment between what reputation systems measure and what participants say they value is chronic across art worlds. Jazz musicians in the 1950s valued improvisational daring; the commercial music industry valued smooth accessible performances. Academic researchers value original rigorous work; the citation system rewards publication volume. The misalignment does not paralyze the world but shapes it: participants learn to optimize for whatever the system rewards, which means the system produces more of whatever it measures and less of whatever it does not.
The discourse between triumphalists and elegists described in The Orange Pill is, in part, a conflict about reputation. Triumphalists have captured the existing system: their metrics are impressive, their posts go viral, their narratives align with platforms' preference for shareable stories. Elegists lack a reputation mechanism valuing what they value. The silent middle has no reputation mechanism at all, because ambivalence does not go viral.
Origin
Becker's analysis of reputation systems drew on studies of the French Salon (Harrison and Cynthia White's Canvases and Careers, 1965), the classical music concert system, academic citation practices, and popular music charts. Art Worlds (1982) synthesized these into a general framework.
Extensions to digital reputation systems appeared in scholarship on social media economies, platform capitalism, and the attention economy — literatures that have begun to intersect with analyses of AI-generated content and its discoverability.
Key Ideas
Reputation systems construct quality, not merely measure it. Work the system elevates is treated as good; work it ignores is treated as nonexistent.
They become more important as production grows. When audiences cannot survey everything, whatever sorts the abundance determines what is seen.
Current AI-world systems are inherited from social media. They measure engagement, speed, and volume — proxies that imperfectly track what most participants say they value.
Participants optimize for what is measured. The system's measurements become the criteria of success, regardless of whether they correspond to the values participants hold in principle.
Building better reputation systems is a convention-negotiation problem. It cannot be solved by any single platform; it requires coordinated adoption of new evaluative practices.
Debates & Critiques
Some argue that market-determined reputation is the only non-authoritarian alternative — that any attempt to define reputation by criteria other than what participants will pay for imposes gatekeeper preferences. Becker's framework acknowledges the risk but notes that market reputation is not value-neutral; it reflects the preferences of whoever has purchasing power. The question is not whether reputation systems embody values but which values they embody and whose interests they serve.
Appears in the Orange Pill Cycle
Further reading
- Howard Becker, Art Worlds, Chapters 5 and 7 (University of California Press, 1982)
- Harrison and Cynthia White, Canvases and Careers (University of Chicago Press, 1965)
- Joel Podolny, Status Signals (Princeton University Press, 2005)
- Shoshana Zuboff, The Age of Surveillance Capitalism (PublicAffairs, 2019)
- James English, The Economy of Prestige (Harvard University Press, 2005)