To Throw a Stone with Six Birds: On Agents and Agenthood

A new research paper titled 'To Throw a Stone with Six Birds: On Agents and Agenthood' by Ioannis Tsiokos introduces a formal, testable theory for defining artificial agents. The work, published on arXiv, presents the Six Birds Theory (SBT), which treats objects as 'induced closures' rather than primitives. The core problem it addresses is the common conflation between simply persisting as an object and actively exerting control to make a counterfactual difference—a confusion that makes claims of agency difficult to verify and easy to fake. Tsiokos's solution is a 'type-correct' account where an agent is defined as a maintained theory object whose feasible interface policies can steer future outcomes while remaining viable.

To operationalize this definition, the paper proposes four concrete, checkable components that must be present: ledger-gated feasibility, a robust viability kernel computed as a greatest fixed point, feasible empowerment (measured as channel capacity in bits) as a proxy for real difference-making, and an empirical packaging map whose 'idempotence defect' quantifies objecthood under observation. The theory was tested in a minimal simulated 'ring-world' environment. Controlled experiments yielded four key separations: systems without true agency showed zero empowerment, enabling 'repair' collapsed the idempotence defect, true protocols increased empowerment only over multi-step horizons, and crucially, systems that learned to rewrite their own operators saw a measurable increase in median empowerment from 0.73 to 1.34 bits. The framework provides hash-traceable, reproducible tests for agenthood without making claims about goals or consciousness.