Regulation of Cross-Demographic Welfare Externalities

Determine whether the impact-space framework can constrain the welfare impacts that one demographic group imposes on another.

Background

The paper represents each alignment protocol by its deployment impact and defines externalities through changes in individual welfare induced by participation or by shifts in the learned model. It develops linear constraints for individual welfare floors, public-spirit trade-offs, and participation-externality bounds.

The discussion leaves unresolved whether the same framework can be extended from individual-level guarantees to explicit regulation of welfare transfers or harms between demographic groups. Such an extension would be relevant when groups differ systematically in preferences or when the designer seeks group-level distributive protections.

References

The information granularity resulting from the impact space analysis leads to a rich set of future questions: can we use this framework to constrain the welfare impacts of one demographic group on another?

— Algorithmic Impact Reveals the Hidden Social Choice Structure of Alignment  (2608.24046 - Wojtowicz et al., 25 Aug 2026) in Section 7, “Limitations and Discussion,” Discussion paragraph

Can every alignment mechanism (traditional RLHF, DPO, Max-Min, Nash) be characterized by the patterns of externalities that it creates over the population of individuals, and do we have the power to regulate these in a unified way?

— Algorithmic Impact Reveals the Hidden Social Choice Structure of Alignment  (2608.24046 - Wojtowicz et al., 25 Aug 2026) in Section 7, “Limitations and Discussion,” Discussion paragraph