[Paper Review] Diversity of preferences can increase collective welfare in sequential exploration problems

Pantelis P. Analytis, Hrvoje Stojić+2 more|arXiv (Cornell University)|Mar 28, 2017

Experimental Behavioral Economics Studies17 references3 citations

TL;DR

This paper demonstrates that intermediate preference diversity in sequential exploration settings—where agents follow popularity rankings—can significantly boost collective welfare by improving search efficiency and utility, even when search costs are high. The key finding is a non-monotonic relationship: moderate diversity (d ≈ 0.2) maximizes net utility by enabling better exploitation of objective quality through decentralized exploration, outweighing higher search costs.

ABSTRACT

In search engines, online marketplaces and other human-computer interfaces large collectives of individuals sequentially interact with numerous alternatives of varying quality. In these contexts, trial and error (exploration) is crucial for uncovering novel high-quality items or solutions, but entails a high cost for individual users. Self-interested decision makers, are often better off imitating the choices of individuals who have already incurred the costs of exploration. Although imitation makes sense at the individual level, it deprives the group of additional information that could have been gleaned by individual explorers. In this paper we show that in such problems, preference diversity can function as a welfare enhancing mechanism. It leads to a consistent increase in the quality of the consumed alternatives that outweighs the increased cost of search for the users.

Motivation & Objective

• To investigate whether preference diversity improves collective welfare in sequential exploration problems where agents follow public popularity signals.
• To analyze the trade-off between search costs and utility gains under varying levels of preference diversity.
• To determine whether diversity mitigates the inefficiencies of herding behavior in social learning environments.
• To evaluate how informational externalities from diverse explorers enhance later decision-making.
• To identify the optimal level of preference diversity that maximizes net utility in sequential search markets.

Proposed method

• Agents sequentially explore alternatives ordered by popularity, selecting the first with utility above a satisficing threshold T.
• Each alternative has an objective utility component (shared) and a subjective utility component (agent-specific), both normally distributed.
• Preference diversity is modeled via the variance of subjective utility (σ²_s = d), while objective utility variance is σ²_o = 1−d, with d ∈ [0,1].
• Search cost c is constant and linked to threshold T via optimal stopping rules from random search models (Chow et al., 1971).
• Simulations are run across 77 market configurations (d from 0 to 1 in steps of 0.1, c from 1/2² to 1/2⁸), each repeated 1000 times for stability.
• Net utility is computed as the utility of the selected alternative minus total search cost, with collective welfare measured as average net utility across agents.

Experimental results

Research questions

• RQ1 Does intermediate preference diversity lead to higher collective welfare in sequential exploration with popularity-based search?
• RQ2 How does preference diversity affect the balance between search costs and utility gains in social learning settings?
• RQ3 What mechanisms allow diverse preferences to improve collective outcomes despite higher individual search costs?
• RQ4 Is there an optimal level of preference diversity that maximizes net utility, and if so, what is it?
• RQ5 How do thick informational externalities from diverse explorers improve later agents’ decision quality?

Key findings

• The highest average net utility is achieved at intermediate preference diversity (d ≈ 0.2), not at d = 0 or d = 1.
• At high search costs (e.g., c = 1/2³), markets with d = 0.5 outperform those with no diversity (d = 0), despite higher search costs.
• Markets with d = 1 (complete diversity) incur the highest search costs and lowest net utility, as popularity signals become uninformative.
• At d = 0, agents herd on the first agent’s choice, leading to high variability in outcomes and suboptimal collective welfare.
• Diverse preferences lead to better search paths: alternatives with higher objective utility are more likely to be encountered earlier due to decentralized exploration.
• The mechanism enabling improved welfare is the generation of thick informational externalities—early diverse exploration improves the quality of the popularity signal for later agents.