Special thanks to Christopher Santos-Lang and Van Parunak for their feedback.

Paper: arxiv.org/abs/2504.21579

Workshop: coin-workshop.github.io/coine-2025-d... (13/13)

Thanks to my co-authors @christophsalge.bsky.social and @petelewis.bsky.social, the COINE organisers, and everyone who engaged with our work. (12/13)

We also derive insights for 𝗺𝘂𝗹𝘁𝗶-𝗮𝗴𝗲𝗻𝘁 𝘀𝘆𝘀𝘁𝗲𝗺 𝗱𝗲𝘀𝗶𝗴𝗻:

1. Perfect information isn’t always best for cooperation
2. Different kinds of noise affect outcomes differently, so designers should model them carefully (11/13)

We show that these “flaws” create cognitive conditions where institutions can bootstrap into existence. What looks irrational for the individual may be beneficial for the group, and ultimately, for the individual (groups are often more successful than individuals (10/13)

-𝗣𝗿𝗼𝗯𝗮𝗯𝗶𝗹𝗶𝘁𝘆 𝗱𝗶𝘀𝘁𝗼𝗿𝘁𝗶𝗼𝗻: Small probabilities are overestimated.
-𝗣𝗲𝗿𝗰𝗲𝗽𝘁𝘂𝗮𝗹 𝗵𝗲𝘁𝗲𝗿𝗼𝗴𝗲𝗻𝗲𝗶𝘁𝘆 (𝗻𝗼𝗶𝘀𝗲): Individuals perceive the world differently due to noise, bias, and unique experiences. (9/13)

This brings us to the role of 𝘤𝘰𝘨𝘯𝘪𝘵𝘪𝘷𝘦 𝘣𝘪𝘢𝘴𝘦𝘴 and 𝘱𝘦𝘳𝘤𝘦𝘱𝘵𝘶𝘢𝘭 𝘶𝘯𝘤𝘦𝘳𝘵𝘢𝘪𝘯𝘵𝘺, long viewed as flaws, but which might actually help institutions form.

We investigate:
-𝗟𝗼𝘀𝘀 𝗮𝘃𝗲𝗿𝘀𝗶𝗼𝗻: Individuals perceive losses as larger than wins. (8/13)

Standard game theory assumes agents have perfect information, e.g. exact knowledge of current participation.

But what if agents misperceive an institution as already existing? (7/13)

We explore this paradox using an 𝗲𝘃𝗼𝗹𝘂𝘁𝗶𝗼𝗻𝗮𝗿𝘆 𝗴𝗮𝗺𝗲-𝘁𝗵𝗲𝗼𝗿𝗲𝘁𝗶𝗰 𝗺𝗼𝗱𝗲𝗹: Benefits of joining and penalties of not joining grow as more people join. However, unless a threshold is reached, no one wants to join - even when it’s beneficial overall. (6/13)

However, a fundamental issue arises: individuals typically perceive participation as advantageous only after an institution is established, creating a paradox—how can institutions form if no one will join before a critical mass exists? (5/13)

We introduce the institution bootstrapping problem: Prior research suggests institutions emerge when universal participation yields greater collective benefits than non-cooperation. (4/13)

Institutions help communities manage common-pool resources and regulate individual greed, e.g. a village organises irrigation use and enforces rules collectively. (3/13)

We recently presented our paper: “Uncertainty, Bias and the Institution Bootstrapping Problem” at the International Workshop on Coordination, Organizations, Institutions, Norms and Ethics for Governance of Multi-Agent Systems (COINE) @ AAMAS 2025. (2/13)