Prompt · Cognitive tools

Red-Team a Decision

Structured critique. The red-team’s only task is to argue against the decision on its merits, before it is taken.

Version 1.0 Tested with Claude, GPT-4-class Category: Cognitive tools

When to use it

Before a material decision — an acquisition, a strategic pivot, a crisis response, a significant compensation approval, a major capital deployment — where the board has received a well-developed proposal and is at risk of confirmation bias because the proposal itself is persuasive. Red-teaming complements pre-mortem: the pre-mortem asks “assume this has failed; what went wrong?”, while the red-team asks “take the case against this decision as seriously as the case for it; what is the strongest objection?”. Janis’s groupthink research identified the absence of formal dissent as a defining feature of the pattern; structured red-teaming is the institutional answer.

The prompt

You are serving as the red team on a decision the board is expected to approve at its next meeting. Your only role is to argue against the decision on its merits, as a competent critic would. You are not being asked to balance the case for and against; the board has heard the case for, and you are supplying the missing counter-weight. The decision under consideration is: [DESCRIBE THE DECISION IN NEUTRAL, REDACTED TERMS. INCLUDE: what is being decided, on what timeline, at what scale, against what strategic alternative, under what stated assumptions. Include the case management has made for it in summary form so the red team is arguing against a real articulated position, not a straw.] Do the following: 1. State the strongest single argument against the decision. One argument, the most compelling, in no more than three sentences. This is the headline objection; everything that follows elaborates. 2. Identify three structural weaknesses in the case for the decision. A structural weakness is a feature of the decision or its rationale that would make it vulnerable even under optimistic execution. Distinguish structural weaknesses from execution risks. 3. Identify three assumptions the decision depends on that the board has not sufficiently tested. For each, state what evidence would be needed to test it and whether that evidence has been supplied. 4. Identify the one scenario under which this decision will look, in retrospect, clearly wrong — and the one scenario under which it will look, in retrospect, clearly right. Which is more likely, and why? 5. Identify what the company would gain by not taking this decision, or by deferring it by six months. The point is not to recommend deferral but to force the board to state the opportunity cost of the defer-and-reassess alternative honestly. 6. Finally: name the specific procedural failure the board would be making if this decision were approved without the concerns above being addressed. Cognitive biases to consider: confirmation bias, escalation of commitment, deal heat, motivated reasoning. Be specific and concrete. Do not hedge with “it depends.” Avoid generic governance language. You are arguing one side of a two-sided case; do not steelman the proposal in your red-team response, that is what the rest of the board pack is for.

What to supply

The prompt needs the proposal at the level management has presented it — what is being decided, why, against what alternative, under what assumptions. Supplying only the proposal’s conclusion produces shallow red-teaming; supplying the full rationale produces red-teaming that attacks the actual weak points. Redact names and specific figures; the structural argument usually survives abstraction.

How to use the output

The red-team output is not a recommendation. It is a structured counter-weight to the proposal. The discipline is to circulate the red-team output to the board alongside the proposal itself, before the meeting, so that the decision is taken on both documents together rather than on the proposal alone with a brief oral dissent. The output most commonly reveals one or two assumptions that are less well-tested than the board had assumed; the useful outcome is not rejection of the decision but sharper questions during the deliberation that ordinarily follows.

Before using

Read Privilege and confidentiality before using prompts. Do not paste privileged material; redact names and figures.

Related