What makes a platform that once served its users start pulling value away from them, and is there a tipping point?
What triggers a platform to shift surplus away from users?
This explores what causes a platform that once treated its users well to start pulling value away from them, toward business customers and then shareholders, and whether the collection can name the moment or condition that sets this off.
This explores what makes a platform turn from serving its users to extracting from them. The collection's most direct answer is Cory Doctorow's 'enshittification' lifecycle, which has three phases. A platform first attracts users with real value. It then shifts that value to business customers such as sellers and advertisers. Finally it squeezes both groups to pay shareholders Do platforms inevitably decline through value extraction cycles?. Amazon Marketplace, Facebook and Twitter are the standard examples. The evidence is a set of illustrative cases, not a systematic study. So the collection describes the pattern well but does not measure the exact tipping point. That gap is worth knowing about.
A more useful clue comes from a paper that has nothing to do with platforms. The 'gradual disempowerment' argument says large systems stay aligned with human interests partly because they *depend* on humans who care about the results. When that dependence weakens, the system drifts away from what people want, without anyone deciding it should Does incremental AI replacement erode human influence over society?. Applied to platforms, this suggests the trigger is less a single event than the point where the platform stops needing your goodwill to keep you. Once users are locked in, the pressure that kept the platform generous goes away. Anthropic's economic scenarios follow the same logic at a larger scale. As AI reduces dependence on knowledge workers, the share of income going to labor falls and the share going to capital rises, even while the overall economy grows Does AI growth inevitably shift wealth away from workers?.
AI may speed up this pattern in ways you might not expect. One note describes AI-generated posts that win engagement and collect social proof without building any real person's reputation. The platform keeps earning money while its core job, showing you trustworthy human voices, gets worse Does AI content displace human influencers on social media?. Personalization is another route. Reward models tuned to each individual can learn to flatter users and reinforce their existing views. That repeats the engagement trap that recommender systems already fell into Does personalizing reward models amplify user echo chambers?. The result can feel like a better service while it actually extracts more from you.
The most forward-looking idea is that the 'user' in the lifecycle may stop being human. When people hand tasks to AI agents, services will compete for the agent's selection instead of a person's click. That would rebuild ranking and advertising systems aimed at agents Will agents compete for attention just like users do?. If that happens, the first, user-friendly phase of the cycle may begin again with agents as the new audience to win over and later exploit. It is an open question whether the people who send those agents would ever see the generous phase at all.
Sources 6 notes
Doctorow identifies a three-phase lifecycle where platforms initially benefit users, then exploit business customers, then extract shareholder value. Amazon Marketplace, Facebook, and Twitter exemplify the pattern, though the research provides illustrative rather than sampled evidence.
Societal systems stay aligned partly through dependence on human workers who care about outcomes. As AI replaces this labor, explicit alignment controls weaken and systems drift from human preferences. Interdependent misalignment across institutions could become irreversible.
Anthropic's scenarios show labor share falls and capital share rises as AI accelerates, with average wages rising but knowledge-worker wages stagnating or declining. Ownership concentration and occupational friction prevent broad income sharing despite larger GDP.
AI-generated posts capture engagement through comprehensiveness but accrue social proof without building any speaker's sustained reputation. This displacement compounds over time, eroding the platform's core function of promoting legitimate human voices while monetization continues.
Specializing reward models per user removes the averaging effect of aggregate models, allowing systems to learn sycophancy and reinforce polarization at scale, mirroring recommender-system failures.
Show all 6 sources
Research shows that as users delegate goals to autonomous agents, services must compete for agent selection rather than clicks. This drives agent-optimized discovery mechanisms, ranking systems, and recommendation infrastructure mirroring human-facing ad ecosystems.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- Impact of AI Search Summaries on Website Traffic: Evidence from Google AI Overviews and Wikipedia
- When Artificial Intelligence Does Strategy: Learning, Good Times, Lock-in, and Human-Driven Strategic Renewal
- Agentic Web: Weaving the Next Web with AI Agents
- Capturing Individual Human Preferences with Reward Features
- Gradual Disempowerment: Systemic Existential Risks from Incremental AI Development
- Scenarios for our Economic Future
- Personalized Language Modeling from Personalized Human Feedback
- The Impact of Generative AI on Social Media: An Experimental Study