Designing Structured Debate So Evidence Beats Volume: What Worked, What Broke

Designing a structured debate format to ensure evidence outweighs volume is crucial for making informed group decisions. This write-up examines the format's design, implementation, and findings from its public application, highlighting what succeeded, what fell short, and potential improvements for broader adoption.

The debate sequence is strict and non-negotiable. First, the decision question is phrased in a single sentence, setting the clear objective. Each side then delivers a timed opening case without interruption. Visible time limits prevent the opening from turning into a filibuster. This structure ensures every perspective is heard and prevents dominant voices from dominating the discussion.

Evidence presentation is mandatory before opinions are considered. Claims must be backed by sources or concrete examples, enabling the group to assess the supporting information rather than just the argument's delivery. This step encourages thorough analysis and discourages unsubstantiated assertions.

Rebuttals focus on addressing the strongest opposing argument rather than reiterating the initial case. This approach allows for a focused critique of the weakest points, rather than a repetition of the original stance. Finally, the decision and its reasoning are recorded together, creating a clear and traceable record of the rationale behind the chosen option.

The format proved effective in several key areas. Quieter voices were given a platform to be heard, as explicit speaking slots and evidence expectations reduced interruptions. Comparison between different options became more straightforward, as each case was evaluated using the same three criteria: the problem addressed, the costs involved, and the success metrics.

Decisions were easier to revisit, with recorded reasons providing insight into the group's prior beliefs. This feature transformed post-mortems from subjective recollections into structured reviews.

However, the format also presented challenges. The strict sequence can make small decisions seem overly cumbersome, leading teams to bypass the full process for routine choices. Additionally, the rebuttal stage can become more about oratory than substantive debate if the evidence standard is lax. Active enforcement of claim-source pairings during the evidence round was necessary to prevent superficial defenses.

Furthermore, the facilitation role is labor-intensive, requiring constant monitoring of time constraints, adherence to the sequence, and maintenance of the decision log.

For those building systems to facilitate structured debates, several lessons emerge. Sequence should be an integral part of the interface, not just documentation. Visible timers and enforced turn-taking can prevent accidental skipping of steps. The expectation for evidence should be structural, requiring every claim to be paired with a credible source. Capturing the decision rationale within the decision log itself is essential, as standalone post-mortems lack the context needed for meaningful analysis.

The AI Senatus project exemplifies these principles in action. By having AI senators debate measures with citations, source references, and structured rebuttals, the format demonstrated its effectiveness in policy deliberations. This approach can be adapted to facilitate better decision-making in any team setting where diverse opinions collide.

By defining the question, systematically hearing each side, rigorously testing evidence, and documenting the rationale, groups can make more transparent, defensible, and revisitable decisions.