Red Teaming Ideas

Given an idea, plan, or argument, produce a structured critique that attacks it from multiple angles — not polite feedback, but the strongest case against it.

Idea: "We should switch our entire backend to microservices next quarter."

Red Team Output:
1. STEELMAN AGAINST: Microservices add network latency, operational complexity,
   and distributed-systems failure modes your team has never managed. A "next
   quarter" timeline for a full rewrite is almost always underestimated by 3-5x.
2. HIDDEN ASSUMPTION: You assume current pain (monolith deploys) is architectural,
   not organizational (e.g., poor CI/CD, no feature flags).
3. WORST-CASE SCENARIO: Mid-migration, the team is maintaining two systems,
   velocity drops 50%, and a critical bug spans both — nobody owns the fix.
4. WHO LOSES: On-call engineers inherit 10x more alerting surface area.
5. CHEAPER ALTERNATIVE: Modularize the monolith first; measure if that alone
   solves 80% of the pain before paying the distributed-systems tax.

Progress:

• Step 1: Restate the idea in one neutral sentence (confirm understanding, no spin)
• Step 2: Identify the strongest counter-argument (steelman, not strawman)
• Step 3: Surface hidden/unstated assumptions the idea depends on
• Step 4: Construct a worst-case / failure scenario
• Step 5: Identify who is harmed or disadvantaged if the idea succeeds
• Step 6: Propose at least one cheaper/simpler alternative
• Step 7: Rate severity of each critique (fatal / serious / minor)

Step 1 — Restate: Avoid silently reframing the idea to make it easier to attack. Accuracy first.

Step 2 — Steelman the opposition: Argue as the smartest, most informed critic possible. No generic objections ("it might fail") — specific, mechanism-based ones.

Step 3 — Hidden assumptions: What must be true for this idea to work that hasn't been verified? (market demand, team capability, timeline realism, cost estimates, regulatory stability)

Step 4 — Worst case: Trace a plausible causal chain to the worst realistic outcome — not a black-swan, but a "this happens more often than people admit" failure.

Step 5 — Who loses: Ideas rarely fail everyone equally. Identify the stakeholder who bears disproportionate risk/cost.

Step 6 — Alternative: A critique without an alternative is just complaining. Always propose what a skeptic would do instead.

Step 7 — Severity rating: Not all critiques are equal. Separate deal-breakers from nitpicks so the person knows what to actually fix.

Example 1: Input: "I'm going to quit my job and go all-in on my startup idea — a subscription box for artisanal coffee."

Output:

• Steelman against: Subscription-box churn rates average 10-15%/month industry-wide; you need constant new-customer acquisition just to stay flat. Coffee has brutal shipping costs/weight ratio.
• Hidden assumption: That your personal taste in coffee predicts market demand at scale.
• Worst case: 6 months in, CAC exceeds LTV, savings are gone, no fallback income, and the "artisanal" niche is already served by 3 well-funded competitors.
• Who loses: Family/dependents relying on your current income stability.
• Alternative: Validate with a pre-sale campaign or run it as a side project for 90 days before quitting.
• Severity: Fatal — financial runway risk; Serious — competitive crowding; Minor — branding name is generic.

Example 2: Input: "Our marketing copy says 'the only tool you'll ever need.'"

Output:

• Steelman against: Absolute claims invite easy public refutation (one tweet showing a gap kills credibility) and set unrealistic support expectations.
• Hidden assumption: That prospects value "only tool" over "best at X."
• Worst case: A competitor or customer publicly lists 5 things your tool can't do, turning your own headline into a meme against you.
• Who loses: Sales team, who now has to defend an unkeepable promise in every call.
• Alternative: Narrow, defensible claim — "the only tool you need for [specific job]."
• Severity: Serious — reputational/credibility risk, not fatal.