Ads Research Lab
A Project Moosed research lab for ad design and the future of ads and marketing.

Papers

The lab's publication shelf. Papers are theory and framework work: defined terms, a stated model, numbered propositions that can be tested, and an honest account of what the evidence does and does not show. A paper is listed here once it enters review; it is published in full only after the Project Moosed review and audit route is complete.

Paper #1: Agents and Advertising

Status: In review  Theory paper, 2026. Author framing: Joel Cochran / Project Moosed.

Summary. A personal agent or AI browser acting for a person does not consume interruptive advertising the way a human eye does. It summarises, extracts, blocks, or ignores creative whose only payload is persuasion. Unless an ad gives the agent something valuable to take (a verifiable price, offer, availability, or term it can use for its person), its throughput through the agent is near zero.

The paper reviews what platforms, industry bodies, verification firms, and researchers have said on this topic so far, and grades those claims. It then sets out the gap the lab works in, and presents the dual-lane model: Human Ads and Agent Ads, side by side, from the same advertiser, site, and budget, with the agent lane measured in takes, presentations, and redemptions. The claim that the agent lane becomes the primary method of transactional marketing over time is stated in the paper as a prediction, graded as such, not as a finding.

The paper's own discipline is part of its point: its most important negative finding is that personal agents are not yet a large share of ad impressions on the measured evidence, so the direction is fast but the current base is small. Worked scenarios in the paper are illustrative, constructed for the paper, and labelled as such; documented real cases are cited separately as evidence.

Full text will be published here after review. The lab's working model, drawn from the paper, is live now on The Model page.

Paper #0: White Hat AI vs Black Hat AI

Precursor paper  Project Moosed.

The lab's frame for legitimate versus deceptive agent-facing practice comes from Project Moosed's earlier White Hat AI paper: build stronger, better systems in the open, and name the deceptive version plainly. Read it at projectmoosed.com.