Details

Bayesian Persuasion as a Means to Reducing Harm from Behavioural Ad Retargeting

Year: 2026

Term: Winter

Student Name: Owen Lucas

Supervisor: Zinovi Rabinovich

Abstract: The userbase of digital marketplaces, attention markets (i.e. social media platforms), and in general, large consumer-facing services that recommend items to users provide an undeniably useful service, given our bounded rationality and the data deluge of the modern age. The relationship producers and their advertisements have towards consumers and their attention have become increasingly liquid, in a broad sense. The leaps between the market crier, to newspapers, to billboards and TV show an increase in the distance ads can propagate, but as the marketplace itself transitioned into a global space, the challenge of ad delivery pivoted from a regional, demographic domain--- into a personal one, narrowing down the capture of attention by means of recommender systems (RecSys), and behavior- and data-mining. These practices enable the companies behind these ubiquitous platforms--- such as Meta and Alphabet--- to provide liquidity to advertisers bidding for attention by leveraging users' behavioural surplus; however, this practice leads to widespread negative social impact, and regulation often lags behind advances in technology, which necessitates an approach independent of regulation.