Methodology
The mintonomics pipeline.
Six stages. Sequential. Each one finishes before the next begins. Skipping a stage - typically simulation or adversarial testing - is the modal failure across the token industry.
Research
Before designing anything new, find five existing launches in the same category.
Catalog the model, the distribution, the outcome, the failure mode. The answer to "should we tokenise" is usually here. If every comparable project with a token is down 80%+, the default-to-no-token rule is already doing the work.
Spec
A token spec answers, in order: problem, stakeholders, utility, supply, distribution, demand.
Who is promising what to whom? What utility does the token actually provide? Where does demand come from, and is it real (fees, service payments) or implied (emissions, speculation)? The spec is the contract. Everything else is commentary.
Personas
Stakeholders become persona files: utility function, constraints, action space, decision rule.
The workspace ships eight standard personas: four cooperative (end-user, builder, capital provider, operator) and four adversarial (exploiter, mercenary LP, governance capturer, attacker). Every simulation runs against all eight. Project-specific personas layer on top.
Simulation
Agent-based, parameter sweep, observed outcomes. Not a price forecast.
Each persona becomes an agent. The protocol becomes an environment with state and rules. We sweep parameters across a defined range and observe what breaks. The output is a heatmap of outcomes - not a prediction, but a falsification boundary.
Adversarial testing
Re-run with adversarial personas dialled up. Most designs break here.
Can any persona extract disproportionate value? Can governance be captured under stated rules? Does the protocol survive a 50% loss of cooperative participants? Questions with a "yes" answer send the design back to stage 2.
Decision
Yes, no, or not-yet. Documented with rationale, signed off, and filed.
A recommendation file with: the decision, the reasoning, the key parameter values, the adversarial tests passed, and the revisit triggers if the answer is "no" or "not-yet". The decision is a living document - if conditions change, the file is updated and the change is logged.