Trust Stack
A layered model of the evidence an AI or search system can use to trust an entity, a source or a claim. Attributed to Kevin Indig, 2026.
What it actually means
Trust is not a single judgement. It is assembled from layers of different kinds of evidence: what you say, what your record shows, what independent parties say, and what can be verified against something official.
Stacking them makes the asymmetry visible. The layers are not equal, and the ones brands control are the ones that count for least.
Why the layer you control counts least
Self-description is free to produce, so it carries little weight. Registry filings are costly to fake. Independent coverage cannot be authored at will. Reviews come from people with no incentive to help.
Effort therefore inverts: the cheapest layer to produce is the one to stop investing in, and the expensive layers are where the trust actually accrues.