Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Fairness-aware design of nudging policies under stochasticity and prejudices

Lisa Piccinin, Camilla Quaresmini, Edoardo Vitale, Mara Tanelli, Valentina Breschi

TL;DR

Problem: Adoption diffusion is hindered by structural inequalities and prejudices, and incentive policies can unintentionally amplify injustice. Approach: Extend the Generalized Linear Threshold framework with thresholds drawn from a Beta distribution to capture stochastic adoption, and design a fair Model Predictive Control (MPC) that optimizes incentives under equality and equity objectives. Findings: Simulations on real mobility-habit data show that injustice reduces adoption, equality smooths incentive distribution, and equity reduces disparities in outcomes, demonstrating the effectiveness of fairness-aware policy design. Significance: The work shows that fairness-aware nudging can achieve effective diffusion while mitigating social inequalities, informing policy on balancing equity and equality in resource allocation.

Abstract

We present an injustice-aware innovation-diffusion model extending the Generalized Linear Threshold framework by assigning agents activation thresholds drawn from a Beta distribution to capture the stochastic nature of adoption shaped by inequalities. Because incentive policies themselves can inadvertently amplify these inequalities, building on this model, we design a fair Model Predictive Control (MPC) scheme that incorporates equality and equity objectives for allocating incentives. Simulations using real mobility-habit data show that injustice reduces overall adoption, while equality smooths incentive distribution and equity reduces disparities in the final outcomes. Thus, incorporating fairness ensures effective diffusion without exacerbating existing social inequalities.

Fairness-aware design of nudging policies under stochasticity and prejudices

TL;DR

Problem: Adoption diffusion is hindered by structural inequalities and prejudices, and incentive policies can unintentionally amplify injustice. Approach: Extend the Generalized Linear Threshold framework with thresholds drawn from a Beta distribution to capture stochastic adoption, and design a fair Model Predictive Control (MPC) that optimizes incentives under equality and equity objectives. Findings: Simulations on real mobility-habit data show that injustice reduces adoption, equality smooths incentive distribution, and equity reduces disparities in outcomes, demonstrating the effectiveness of fairness-aware policy design. Significance: The work shows that fairness-aware nudging can achieve effective diffusion while mitigating social inequalities, informing policy on balancing equity and equality in resource allocation.

Abstract

We present an injustice-aware innovation-diffusion model extending the Generalized Linear Threshold framework by assigning agents activation thresholds drawn from a Beta distribution to capture the stochastic nature of adoption shaped by inequalities. Because incentive policies themselves can inadvertently amplify these inequalities, building on this model, we design a fair Model Predictive Control (MPC) scheme that incorporates equality and equity objectives for allocating incentives. Simulations using real mobility-habit data show that injustice reduces overall adoption, while equality smooths incentive distribution and equity reduces disparities in the final outcomes. Thus, incorporating fairness ensures effective diffusion without exacerbating existing social inequalities.
Paper Structure (1 section, 2 equations, 2 figures)

This paper contains 1 section, 2 equations, 2 figures.

Table of Contents

  1. Conclusions

Figures (2)

  • Figure 2: Adoption rate vs average policy over $N_{\text{MC}}$ Monte Carlo simulations for the different deficit scenarios (ND, RD, CD, and CRD). On the left panel, we always report the mean values (solid lines) and standard deviations (shaded areas) of the adoption rate over time.
  • Figure 3: Nudging policy received by each agent at each time instant over one of the $N_{\mathrm{MC}}=10$ Monte Carlo simulations. Each row corresponds to an agent, each column to a time step, while the color itself indicates the policy magnitude. The lower the nudging policy, the more the color turns towards blue.