Assess generalizability to richer institutional environments

Determine how the experimentally estimated effects of opponent information and artificial opponents carry over from the stylized online all-pay auction to richer organizational and institutional environments containing relationships, reputations, and exit options.

Background

The experiment uses a stylized online auction designed to isolate escalation incentives. Although this setting captures important features of escalation, it omits relationships, reputational concerns, and practical exit options that characterize organizational and market interactions. The authors explicitly leave open whether the observed effect sizes will persist in such richer environments.

References

The setting is a stylised online auction rather than an organisational process; its all-pay structure reproduces real escalation dynamics but strips away the relationships, reputations, and exit options of practice, so how the effect sizes carry to richer environments is open.

— People escalate against a competitor labelled human and hold back against one labelled an optimising machine  (2609.21439 - Ferraz et al., 18 Sep 2026) in Discussion, paragraph beginning “Three scope conditions bound these results”