How Platforms Turn People into Signals

Platforms do not evaluate people. They evaluate signals derived from people — rows of structured attributes, keyword-matched descriptions, and embeddings of fragments of past work.

This is not an incidental detail. It is the operating assumption of the infrastructure. To be legible to a platform is to be compressed. To be compressed is to be ranked. To be ranked is to be filtered.

Three consequences deserve attention:

1. Signal shapes opportunity. The information a platform can read determines what it can offer.
2. Silence is costly. Experiences that do not fit the platform's categories are often treated as absent rather than incommensurable.
3. Optimisation drifts. People learn to write themselves in the way the platform can read, even at the cost of accuracy.

The question, then, is not whether platforms reduce people to signals. They do. The question is which reductions we should build against.