🥊 Critique & Decision
Attack your own ideas before others do — then decide with explicit criteria, not with intuition disguised as analysis. These are families 04 (Critique and Red Team) and 05 (Comparison and Decision).
By the end of the track, you’ll run a complete personal red team and turn difficult choices into matrices, rankings, and bets with exit conditions.
The track’s rule: an idea without a red team review doesn’t make it to the decision table.
Course roadmap
Detailed content
🥊 Red team and robustness
The 10 commands in family 04 — adversarial challenge, failure simulation, and strengthening under uncertainty.
The protocol requires you to refute or challenge the premise before agreeing—disagreement comes first, support comes later (if it holds up).
Models are complacent by default; /pushback turns agreement into something that must be earned.
refutation first · agreement earned · antidote to flattery.
Attacks the central thesis with the strongest possible objections and asks for an explicit defense of each.
A thesis that has never been challenged is a hypothesis with borrowed confidence.
strong objection · explicit defense · tested thesis.
Simulates a capable adversary trying to bring down a plan, system, or argument by exploiting real vulnerabilities.
It’s cheaper to find the failure in the simulator than in the market.
attack vector · vulnerability · mitigation.
Put the plan through extreme conditions (/stress) and map failure modes with detection and mitigation (/failure).
Every plan has a load limit; it’s better to learn it in a test than on a bad Tuesday.
extreme condition · failure mode · detection trigger.
Assumes the project has ALREADY failed and works backward: what were the likely causes?
Thinking of the future as the past frees the team to say the unthinkable — risks become legitimate.
prior retrospective · plausible cause · early mitigation.
Actively look for the case that overturns the general rule — the exception that refutes or limits it.
One well-documented exception is worth more than a hundred confirming examples.
negative case · rule delimitation · honest scope.
Identify cognitive biases in the reasoning (/bias) and areas the analysis hasn’t even considered (/blindspots).
Confirmation bias and survivorship bias silently distort even technically impeccable analyses.
confirmation bias · survivorship bias · unexamined area.
Redesigns the solution to work reasonably well across several scenarios—not perfectly in just one.
Robustness trades a fragile optimum for something good that can withstand the real world.
multi-scenario performance · hedge · margin of safety.
🎯 Comparison and decision
The 10 commands in family 05 — criteria, weights, rankings, matrices, and bets with a stopping point.
Set measurable criteria and their weights BEFORE evaluating the options.
The criterion chosen after the favorite option is selected magically adjusts so that option wins.
a priori criterion · relative weight · fixed yardstick.
Sorts options by criteria and explains each ranking — a ranking without justification is just numbered preference.
It’s the most commonly used decision command in day-to-day work; with parameters, it becomes a team standard.
ordering · positional rationale · criteria=impact,cost,risk.
Assigns scores by criterion on a defined scale, showing how the total score is composed.
A composite score reveals whether the win came from one dominant criterion or broad consistency.
anchor scale · score per criterion · transparent weighting.
Produces a side-by-side comparison by dimension — without forcing a premature verdict.
Separating comparison from decision keeps a first impression from closing the window for analysis.
comparable dimension · suspension of judgment.
Close the process: one recommendation with evidence, degree of reversibility, and cost of changing your mind.
Reversible decisions call for speed; irreversible ones call for depth. Knowing which is which changes everything.
two-way door · cost of reversal · clear commitment.
Build the full matrix: evaluate each alternative against every criterion, applying the weights.
The matrix exposes trade-offs hidden behind polished narratives.
weighted matrix · justified cell · sensitivity.
Eliminates dominated options and presents the frontier where no criterion can improve without worsening another.
Half the "options" on any list shouldn't be there—ruling them out saves entire meetings.
dominated option · Pareto frontier · efficiency.
/bet structures the bet: upside, downside, size, and triggers; /stop defines the objective condition for ending it without drama.
A bet without a stop-loss becomes a zombie project — kept alive by pride and costing a lot every month.
return asymmetry · kill criteria · graceful exit.