The idea explores how agents with different moral goals could cooperate without direct communication or the need to simulate each other's decisions. Traditional models often rely on these mechanisms, which become impractical in large or infinite systems. This gap is particularly relevant for communities like Effective Altruism, where efficiently allocating resources across diverse priorities is crucial.

How Evidential Cooperation Could Work

One way to enable cooperation among such agents is through a framework where their actions provide evidence of each other's behavior. Instead of negotiating or simulating one another, agents could maximize a shared utility function that balances their individual moral goals. This approach builds on updateless decision theories, where agents act as if their choices are correlated with those of similar agents. For example, an agent focused on animal welfare might allocate resources to regions where their efforts have the highest impact, while another agent prioritizing global health does the same in different contexts—both benefiting from comparative advantages without direct coordination.

Potential Applications and Stakeholders

This framework could be valuable for:

• Researchers in moral philosophy and decision theory, offering a new way to model cooperation in complex systems.
• The Effective Altruism community, helping refine resource allocation across competing priorities like poverty reduction and existential risk mitigation.
• AI alignment researchers, providing insights into how AI systems with differing objectives might cooperate implicitly.

Stakeholders like academic researchers might be motivated by theoretical breakthroughs, while practical adopters could use the framework to improve real-world prioritization.

Execution and Comparisons

A starting point could involve formalizing the theoretical framework and testing it through simulations or thought experiments. Existing work on acausal trade and updateless decision theories provides a foundation, but this idea extends those concepts by focusing on diverse moral goals and evidence-based cooperation. Unlike traditional bargaining models, this approach doesn’t require explicit negotiation, making it scalable to large or infinite scenarios.

While the idea is theoretical, initial steps might include publishing papers or collaborating with organizations to explore applications. The emphasis would be on clarity—translating abstract concepts into actionable insights for non-experts.