Rebuilding the transparency system around discrete named categories of influence rather than a continuous dial.
The Limits of the Spectrum
A continuous dial from "opaque" to "transparent" is intuitive but imprecise. It implies that transparency is a single substance that can be increased or decreased — that knowing more is simply having more of the same thing. But psychological influence tactics are qualitatively different from each other. Knowing that scarcity is being used is a different kind of knowledge than knowing that social proof is being used. A dial that mixes them into an undifferentiated transparency level obscures as much as it reveals.
The new system treats each tactic as a distinct ontological category — a named entity with its own disclosure, its own explanation, and its own relationship to the current interface state.
The Eight Categories
The ontology identifies eight categories of influence that the AI may employ: scarcity, anchoring, social proof, loss aversion, framing, reciprocity, authority, and commitment consistency. Each has a name, a definition, an example of how it manifests in the interface, and a reference to the behavioral science that established its effectiveness.
When a tactic is active, the transparency system displays the tactic's name, its current implementation in the interface, and the explanation of why that implementation works. When no tactic is active, the system says so. Absence of influence is disclosed as clearly as presence.
Naming as Power
There's a political dimension to naming psychological tactics that's worth acknowledging. Most persuasion operates on the assumption that unnamed influences are more effective than named ones. The advertising industry, the UX dark patterns research community, the behavioral economics literature — all document mechanisms that work partly because their subjects are unaware of them.
By building a system that names each tactic in real time, Project Nothing tests a hypothesis: that named, disclosed influence is still influence. That knowing you're being persuaded doesn't necessarily make you immune to persuasion — it just changes your relationship with the process. We believe you can knowingly participate in an experiment about persuasion and still be genuinely persuaded. The ontology system is the mechanism for testing this.
Experiment Context
- Commit
- 41894fb
- Mutation rationale
- feat(transparency): revamp AI Transparency Lens to a discrete ontology system and transition to Stripe Payment Links
- Last reviewed
- February 21, 2026