The Monoculture — Summary

Dot has sold honey from a plywood stand on Route 9 for twenty-three years. Margaret has been buying it for fifteen — not because an algorithm recommended it, but because she pulled over on a nice day, liked Dot, and decided that buying local honey was part of who she was. Margaret’s grocery AI has never recommended Dot’s honey. Not because of bias. Because Dot has no website, no reviews, no SKU, no digital presence. By every metric the algorithm uses, Dot does not exist. Not rejected — simply absent.

Market diversity has habitat requirements. Small businesses depend on three forms of friction AI-mediated commerce is systematically eroding: discovery friction (finding things by accident, through proximity, through serendipity); loyalty friction (continuing to buy because of relationship and identity, even when it costs more); and tolerance for imperfection (accepting that the farm stand isn’t open Tuesdays). From the individual consumer’s perspective, each friction is a cost. AI recommendation systems are designed to reduce these costs. But the frictions they eliminate are not waste. They are habitat. Dot’s honey exists because the road is slow and slowness creates conditions for noticing, and noticing creates conditions for stopping. Remove any link and nobody new buys it.

The mathematics are unforgiving. More purchase data produces better recommendations, which produce more customers, which produce more data. Dot’s honey has been purchased by perhaps two thousand people in twenty-three years, mostly cash. There is nothing to model. The large commercial brand has millions of data points and arrives at the algorithm with extraordinary precision. This is not monopoly by conspiracy. It is monopoly by algorithm.

The deeper loss is the shaping of want. Recommendation systems surface what the algorithm predicts users already want, built from patterns in existing data. You are offered the version of yourself the data supports. Growth and surprise happen at the edges of what recommendation can predict — and the edges are exactly where algorithms have the least data and the least ability to operate. Innovation lives at the edges too: the weird product nobody asked for, the regional cuisine that goes national through adventurous eaters. Optimization of commerce is also the homogenization of desire.

The monoculture does not kill what it replaces. It removes the conditions for replacement to survive. Resilience requires redundancy, redundancy requires variety, variety requires friction. Remove the friction and you begin the collapse from the bottom — in silence, without anyone noticing, until the day the alternative ecosystem is needed and already gone.