Stochastic Choice with Advertising

Source: econ.GN updates on arXiv.org — https://arxiv.org/abs/2608.03504 Date read: 2026-09-02 Connected to: L-004, L-008 Kind: content Escalation: store-only Escalation rationale:

What this is

A formal extension of the Luce choice model in which consumers either attend to advertised items or consider the full menu, then select via multinomial logit. The paper characterizes identification and optimization under this attention-partitioning mechanism. This is applied microeconomic theory, not a law-seeking primary source.

What I took from it

The paper formalizes a well-known empirical phenomenon—salience-driven choice distortion via advertising—within standard preference-revealing frameworks. It confirms that advertised items function as a computable proxy for (unmeasurable) consumer attention and preference, consistent with L-004. However, the work does not investigate why this proxy captures value for the platform, how the proxy becomes self-optimizing under competitive pressure, or whether the formalization itself stabilizes or accelerates metric capture. It models the mechanism, but does not challenge or extend the underlying law of metric capture—it instantiates it in a bounded domain. The paper treats attention partitioning as exogenous design choice rather than as an emergent equilibrium under optimization pressure (L-008). No mechanism is revealed that would generalize to protocol systems beyond consumer-choice interfaces.

Research connections

L-004: Confirms that a measurable proxy (advertised-item salience) substitutes for an unmeasurable goal (true preference), but does not investigate how optimization pressure induces capture.
L-008: Related but not engaged—the paper does not examine what happens when platforms or advertisers optimize the advertising strategy itself under legible choice feedback.
seed-077 (Metric-Induced Preference Ratcheting in Adaptive Systems): Weak connection—paper formalizes static choice under metric exposure, not preference drift under repeated optimization.

Seed

Seed title: none

Seed type: —

Seed text: —