Why do all the best sellers end up offering the exact same bribe instead of competing to outbid each other?

This explores why a ceiling on seller subsidies makes the best sellers look identical instead of letting each one signal its exact quality. The clearest source in the collection is a formal search-market model How do seller subsidies sort quality in search markets?. Sellers offer subsidies to get a searcher to inspect them, and the equilibrium sorts into three bands. Low-quality sellers are never inspected. Intermediate sellers separate: each better seller offers a slightly higher subsidy, so the amount reveals quality. High-quality sellers all offer the maximum and pool at the cap.

The reason is simple once you see what separation needs. For a signal to tell types apart, each type needs a different signal it can afford and that a worse type would not want to copy. In the middle band that works, because a slightly better seller gains a bit more from being inspected and so will pay a bit more. As quality rises, though, the subsidy needed to stand out keeps rising, and eventually it would go past the cap. Every seller who would have offered more than the ceiling is forced down to the ceiling. The cap does not erase their quality differences. It cuts off the part of the signal range where those differences would have shown. Pooling at the top is what happens when the signal hits a hard limit, not a sign that the market failed.

What the reader may not expect is that the model presents this as a good outcome. It reports that the equilibrium reveals as much information as possible without running an auction. The searcher does not need to rank the top sellers precisely. It only needs to know which ones are worth inspecting, and all the pooled sellers clear that bar. A cap that stopped sellers from outbidding each other without limit would cost very little useful information. Perfect separation would mean ever-larger payments that mostly move money around without improving the searcher's choices.

The collection has little on this question beyond that one model. The other retrieved notes are about AI pricing agents colluding, protocol-driven market structure, and preference aggregation. They share some vocabulary ("pooling", "markets") but do not explain the mechanism. If you are reading this because AI agents will be the ones doing the searching, the useful next step is the collusion work. It shows how agents that face each other repeatedly drift toward coordinated outcomes Does agent heterogeneity prevent LLM pricing collusion?. That suggests a related question the collection does not answer yet: whether sellers who all sit at the same cap could start coordinating there.