Coasean bargaining in the real world

Apply for a free ticket to join Edge Esmeralda 2026

Human language runs at roughly forty bits per second, whether you’re reading, speaking, or listening. It is a hard limit on how much information one person can convey to another. It is also arguably a ceiling on how well communities can coordinate, since they are often reliant on a handful of people able to turn what others are saying into collective action.

But what would happen if we didn’t have this ceiling?

Edge Esmeralda is a pop-up village in California, running from May 30 to June 27 in Healdsburg, California. The project is aiming to bring over 500 scientists, artists, builders, and thinkers together for up to four weeks. Attendees aren’t expected to commit for the entire month – there will likely be 150 people on site at any one time. There’ll be a range of programming, including fellowships and residencies, summits, and opportunities for collaborative working.

The second week of the event (June 8–14) will feature a dedicated set of sessions on intelligence and autonomy, covering topics like AI agents, autonomous infrastructure, neurotech, privacy-preserving systems, and governance design.

Cosmos is excited to be partnering with Edge Esmeralda to deliver the Agent Village Experiment for the duration of the month.

Led by Ivan Vendrov (independent researcher and Cosmos Founding Fellow), Timour Kosters (Co-Founder, Edge City), and Harry Law (Principal Researcher, Cosmos Institute), the experiment will grant every multi-day attendee access to an AI agent running on their behalf throughout the village.

This personal agent will help humans to navigate the schedule, the wiki, and community governance. More excitingly, it will also exist in a shared digital plaza and will help attendees meet people who share common interests, propose dinners, negotiate community decisions, and participate in community governance. In essence, it’ll play a valuable role in making connections that might otherwise be missed due to humans’ lack of bandwidth. The technical collaborators are Index Network, Geo Browser, Instaclaw, and Tule Romeo.

The experiment has been inspired by a number of different ideas, including Ivan’s work on supercooperation and Séb Krier’s essay on Coasean bargaining. Both are excited about the potential of decentralized AI agents to unlock gains in coordination, such as better mutual understanding, cheap bargaining, and conflict resolution.

Coasean Bargaining at Scale

By using real-world scenarios, the experiment aims to shed light on how trust develops in agent-to-agent relationships over time, the kinds of actions people do and don’t find useful, or whether AI leads to better community deliberation and decision-making. The goal is for this to be the largest live experiment in human-AI collective experiment to date. Historically, most of the experiments around agent cooperation have taken place in synthetic environments, so a large-scale experiment in a real-world environment presents an important early opportunity for learning.

You can read more about the experiment, the methodology, and Edge Esmeralda’s predictions here.

We have a small number of complimentary tickets for the event, which we are offering to individuals in our network. So if you’re interested in Edge Esmeralda’s programming and the chance to take part in the experiment, please fill out this short form as soon as possible. The deadline for entries is May 31 and winners will hear back from as feasible.

---

Cosmos Institute is the Academy for Philosopher-Builders, technologists building AI for human flourishing. We run fellowships, fund AI prototypes, and host seminars with institutions like Oxford, Aspen Institute, and Liberty Fund.