A model is an explicit argument
An agent-based model represents individuals and their interactions directly. A colony-inspired model could assign each agent a location, a state, and a rule for responding to nearby events. Those choices would be hypotheses or teaching assumptions until supported by data.
AESRD proposes models whose purpose is written before their implementation. Adding biological detail is useful only when it helps answer the question or test the robustness of an answer.
Document the complete process
The ODD protocol—Overview, Design concepts, and Details—provides an established framework for describing simulation models. The programme would use it to organise model purpose, entities, scheduling, processes, and evaluation, with links to the actual implementation.
Documentation would also distinguish fixed inputs, estimated parameters, illustrative settings, and random draws. A screenshot of a simulation is not enough to reconstruct the process that produced it.
Proposed verification and experiments
Small deterministic cases would check whether movement rules, state transitions, and bookkeeping work as specified. Stochastic experiments would use recorded seeds and repeated runs, reporting distributions rather than a single selected trajectory. Sensitivity work would test changes in both numerical settings and structural assumptions.
Verification asks whether the code implements the model. Validation asks whether the model is useful for its stated real-world purpose. Passing the first does not automatically establish the second.
A reusable public foundation
The intended initial outputs are transparent teaching models, experiment manifests, and a checklist for future empirical projects. Public demonstrations would carry a persistent label identifying synthetic inputs and unvalidated biological assumptions.
Research use would require a documented dataset and a clear evaluation plan. Contributors would be encouraged to report cases where the model fails; a useful model has a visible boundary, not an unlimited claim.