The Epistemic Chain

Almost nothing you know arrived directly. It came through a chain of instruments and people, and now, increasingly, through models. Social epistemology has studied human chains for decades; the machine links are new.

In this experiment a fact about the world passes through five links. Each has a reliability and a cost of verification. You decide, for each, whether to check or to trust. At the end you learn whether you know the claim — and, if the chain failed, where responsibility for the failure falls under different accounts.

What to notice. Trusting everything is cheap and usually works, which is exactly what makes it dangerous. Verifying everything is impossible. The interesting positions are in between — and they distribute responsibility very differently depending on whether the unreliable link was a person or a model.