The Trace-Agency Gap
A One-Sided Identification Frontier for Behavioral Evidence and Reflective Endorsement.
Explore the work →The trace is visible.
The truth takes work.
Eleven works on the gap between an observed record and the account we give of it.
11 works
The Coordination–Consensus Continuum
An observed conversation can be compatible with different underlying mechanisms. A causal framework for separating evidence of coordination from claims of consensus.
Explore the work →The Individuality Gap
Why observational uniqueness does not establish person-specific structure. Nearly-full uniqueness in a database is compatible with zero real individuality — the percentage is not a weak signal, it is not a signal. Open the three in-browser demonstrations, in English or Romanian.
Explore the work →Who Was the Click For?
A click can be addressed to the brand, the platform or the feed itself. What do standard funnels lose when they cannot distinguish the recipient?
Explore the work →The Story Comes Cheap
Brand authenticity as a cost inference, not a property consumers detect. If generative AI drives the cost of the carrying signal toward zero, which signals still separate an authentic sender from an inauthentic one? The paper derives that single condition — the signal floor — for identity goods.
Explore the work →What It Would Take to Test This
The costed protocol for the signal-floor account: what an honest test would actually require, priced out, so the claim can be falsified rather than merely argued. Supplements The Story Comes Cheap.
Explore the work →The Liar’s Dividend Without a Liar
Authentic work can carry a provenance penalty with no lie anywhere in the chain. Names the cost — the ambient provenance discount — the asset it depletes, and the only defence that survives: external corroboration, not what the image shows.
Explore the work →The Discount Comes Last
A euro withheld from someone who would have bought anyway is worth roughly ten of new revenue. An operator’s guide that routes promotions on customer and stock state together, and puts price reduction last behind four cheaper levers.
Explore the work →When Coordination Becomes Consensus
Follows one question — who is really speaking? — down the length of a listening pipeline. Collection, deduplication, classification, aggregation: each stage quietly discards some of the evidence that would separate coordinated amplification from genuine consensus, until a dashboard reports an agreement the raw trace never established.
Explore the work →The Silent Denominator
A vocal sample is not automatically the audience. Sharp bounds expose what the people who did not speak leave unresolved.
Explore the work →The Coordination–Consensus Continuum: A Conceptual Framework for the Attribution Problem in Social Listening
The conceptual starting point for the Coordination–Consensus Continuum and the attribution problem in social listening.
Explore the work →