Pros / Cons

When to use

Use this when someone faces a real "should I...?" choice with genuine tradeoffs and wants a decision they can defend: both sides argued properly, consequences sized, and a recommendation they are free to override.

Not for: decisions with an obvious right answer (a pros/cons theater wastes everyone's time), choices that are purely a matter of taste, or situations that are actually blocked on a fact nobody has gathered yet - there the move is "find X out first," which this method surfaces but does not replace. Not a substitute for domain expertise on high-stakes safety, legal, or medical calls.

Method

1. Restate the decision in one sentence, including the do-nothing option if it is viable - often it is the real baseline.
2. Steelman both sides. Decision point: write the strongest argument FOR and the strongest AGAINST as their best advocate would - if one side reads weaker, you have not steelmanned it, you have strawmanned it.
3. Size each point high / med / low impact. Ten trivial pros do not outweigh one high-impact con; never let the count decide.
4. Flag irreversibility. Decision point: any consequence that cannot be undone - data loss, burned relationship, signed contract, public statement - gets a "one-way door" marker and extra weight.
5. Name the decisive unknown, if one fact would settle it. Sometimes the honest recommendation is to go learn that fact cheaply first.
6. Recommend one option in two lines, and state its single biggest risk. "It depends" is banned unless you say exactly what it depends on.

Example

Input: "should we rewrite the billing service in Go or keep patching the Node one"

Output (abridged): Decision: Rewrite billing in Go now, or keep patching Node (do-nothing is viable).

For rewrite: (high) chronic memory leaks caused 3 billing incidents this quarter; (med) team already runs 4 Go services. Against: (high, one-way door) 2-3 months where bugs land in TWO codebases during migration; (med) the leak might be fixable in days if profiled - nobody has tried.

Decisive unknown: whether the leak is fixable cheaply - one profiling day answers it. Recommendation: Spend 2 days profiling before deciding. Biggest risk: the leak looks fixable, gets patched, and the architecture debt resurfaces in 6 months.

Pitfalls

• Strawmanning the loser. Writing a weak version of the side you already reject so the "analysis" just confirms your prior.
• Counting instead of weighing. Declaring the side with more bullets the winner regardless of impact.
• Missing the one-way door. Treating an irreversible cost as just another medium con.
• Cowardly ending. Closing with "it depends" and no recommendation, leaving the person exactly where they started.

Output format

**Decision:** <one sentence, include the do-nothing option if viable>

**For** <option>
- (high|med|low)[, one-way door] <steelmanned point>

**Against** <option>
- (high|med|low)[, one-way door] <steelmanned point>

**Decisive unknown:** <the one fact that would settle it, or "none">
**Recommendation:** <option> - <why, two lines>. Biggest risk: <one line>.