Aula de cursos

Goal: turn the proposal into observations, source checks or tests that can alter the conclusion.

Three different sorts of claim

A documentary claim requires a source; an operational hypothesis needs observation or experiment; a normative judgment requires explicit rules, priorities and accountable human deliberation. Treating all three as matters a chatbot can settle alone is a category mistake.

A minimal experiment

1. State one refutable claim and its baseline.
2. Specify the comparison period and eligible population.
3. Choose a main metric and adverse side effects to monitor.
4. Document how missing data and confounders will be handled.
5. Assign a responsible reviewer, a deadline and stop conditions.

For a queue simulator, test arrivals per time unit, the number of servers at each station, the order mix and priority rules. Calculate mean waiting time, a declared high percentile and completed orders. Use a hand-computable example to verify the implementation. If tools cannot generate an app, require a transparent spreadsheet or pseudocode instead.

Convert this recommendation into a verification plan. Separate statements to check in sources, hypotheses requiring a pilot, and values a human must decide. Define units, baseline, success criteria and evidence gaps. Provide a small independent test and a condition for rejecting the proposed solution.

Deliverable: a traceable decision sheet, not a promise of guaranteed improvement. A simulation without calibrated real-world data illustrates possibilities; it does not establish a forecast.