Technical Defense: Evidence Packs with Limits Stated

Why this matters

A design review becomes a debate between confidence levels: the loudest voice wins, the quiet risk loses, and the decision records who argued best rather than what fits. Evidence packs change the game: numbers, drill results and listed assumptions replace rhetoric, and reviewers attack the evidence instead of the presenter. The L72 lab prepares a pack showing the solution's limits and unproven assumptions; this lesson defines what a pack that survives review contains.

Concepts

The pack has four parts. Decisions: the ADRs behind the design with their trade-offs. Numbers: measurements (capacity, cost, budget burn, recovery times) quoted with dates and contexts, never naked figures. Drills: rehearsal results (failure injection, restore drills, handover audits) proving the mechanisms work. Limits: the unproven assumptions and known gaps, stated up front in plain language. A pack missing the fourth part is a sales pitch; reviewers rightly treat it as one, and the defense starts from distrust.

Present limits first. Opening with what was not proven (scale beyond rehearsal, threats outside the model, costs past the assumption date) disarms adversarial review and focuses questions where they help. Hidden limits discovered by reviewers decide the verdict against the design no matter how strong the rest; volunteered limits frame the design as honest engineering with bounded claims. The no-claim rule from the brief governs throughout: fit for this context, not one right answer.

Defense is dialogue, not performance. Reviewers probe the weakest stated limit; the defender answers with evidence or accepts the gap with an owned action. Changed minds are recorded (which evidence moved whom); unchanged disagreements are recorded too, with the test that would settle them. The review log joins the pack: future revisits start from what was already argued, not from zero.

Worked example

A fixture design defends with a full pack: three ADRs, dated numbers, two drill results, and four listed unproven assumptions led with up front. Reviewers probe the scale assumption; the defender quotes the rehearsal limit and accepts a load-test action with a date. The pack passes with the gap owned, not hidden; the review log records both.

Common wrong move

Defending by volume: more slides, more certainty, fewer limits. Reviewers read volume as fear and dig where it hurts; the limits surface anyway, framed as discoveries instead of disclosures. Lead with limits, defend with evidence, record everything.