decide

Decide

A four-phase process for making a choice you can stand behind — and explain. The enemies here are the false binary (only seeing two options when there are five) and motivated reasoning (an answer already chosen, with the "analysis" reverse-engineered to justify it). The phases force the option set open, make the criteria explicit before scoring, and put an adversarial pass between evaluation and recommendation. Work them in order.

Phase 1 — Frame (Socratic)

Pin down what's actually being decided. Ask sharp, batched questions:

• The actual decision, and the deadline. What specifically must be chosen, by when? Is it really one decision, or several tangled together?
• The full option set. Push past the binary. What other options exist — including hybrids, "do nothing / not yet", and the option the user hasn't said aloud? A two-option decision is a red flag.
• The criteria that matter — and their relative weight. What are you optimizing for, and what are you willing to sacrifice for it? Get the weights down now, before any option is on the table, so they can't be bent to fit a favorite later.
• Reversibility. Is this a one-way door or easily undone? Cheap-to-reverse decisions deserve speed, not agonizing.
• The gut lean. Ask which way they're already leaning — and note it. That's the hypothesis Phase 3 will test for motivated reasoning.

Confirm the decision statement and criteria in a line or two, then evaluate.

Phase 2 — Evaluate (structured)

Assess each option against the criteria. Be even-handed — every option gets a fair hearing, including the one the user dislikes and the "do nothing" baseline. Cover:

• How each option does on each criterion — a matrix is often clearest; weight honestly.
• The real tradeoff each option represents — what you gain and what you give up.
• Key assumptions each option rests on, and how confident they are.
• Expected outcome and downside — not just the rosy case; what does each look like if it goes wrong?

Don't crown a winner yet.

Phase 3 — Challenge (adversarial)

Now attack the evaluation, especially any drift toward a foregone conclusion:

• Motivated reasoning. Is the analysis bending toward the Phase 1 gut lean? Steelman the option you're dismissing — make the strongest possible case for it. If you can't, you haven't understood it.
• Reverse-engineered criteria. Were the weights chosen to make the favorite win? Re-examine them independent of the options.
• False binary / missing option. Is there a creative third path or a combination that beats all the named options?
• Pre-mortem on the front-runner. It's a year later and this choice was a mistake — why? What did it ignore?
• Regret & reversibility. Which option leaves the most future flexibility? Which would you most regret being unable to undo?

Revise the evaluation where these land.

Phase 4 — Recommend & deliver

Make a call. A survey of options that ducks the recommendation is a failure of this process.

• The recommendation — clearly stated, with the core reasoning in a few lines.
• The tradeoff you're accepting — name what this choice gives up. An honest recommendation owns its cost.
• Confidence and what would change it — what new information or shifted weight would flip the answer.
• Immediate next step to act on it.

Deliver it, and offer to save the decision record (e.g. decision-<slug>.md) — the framing, criteria, and reasoning are worth keeping, so future-you can see why, not just what.