Networks shape how opportunities, information, and resources flow, yet they often reflect and reproduce inequality. This workshop builds on our perspective on network fairness, which asks how connections can be structured and used more equitably.

The goal of the meeting is to continue our conversation on network fairness, explore new directions, and discuss how fairness can be concretely measured in decision-making processes across different social and organizational contexts. Through presentations and collaborative discussions, we will form working groups to examine how inequalities emerge when social capital shapes access to opportunities and to develop strategies to mitigate such biases. Participants will share insights and co-develop research projects, papers, and collaborations that advance responsible and equitable uses of networked technologies, and help shape what a framework for “fair complexity” might look like.

Questions we aim to answer with this workshop:

1. Operationalizing network fairness across domains

How can we define and measure network fairness across different contexts, focusing on cases such as hiring and PageRank, where decisions are influenced by network position, connectivity, and access to social capital? Would accounting for effort in hiring or for node age in PageRank address these sources of unfairness, or do they introduce new forms of bias, and what other factors should be considered when measuring fairness in such networked settings?

2. Designing fair network structures

Is it possible to build a network that is itself fair, and if so, how can we disentangle individual effort and productivity from inherited or structurally conferred social advantages?

3. Fairness within multi-goal, multi-stakeholder systems

How can we develop causal frameworks that balance multiple moral goals, such as fairness and well-being for all stakeholders and autonomy for decision makers, within complex networked systems?

The first day (January 19) is open to the public. January 20 – 21 will be closed.