Sort, Don't Pick
Spend decision time sorting options into a good-enough bucket, then flip a coin
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 6
- Confidence
- 85%
Sort, Don't Pick reframes decision-making as a sorting problem, not a picking problem. First you split options into a bucket of things you'd genuinely be happy with and a bucket you wouldn't. To test a borderline item, use the Only-Option Test: if this were the single thing you could have, would you be happy? Yes means it belongs in the good bucket. Once options are in that bucket, Duke argues you should stop trying to rank them, because under uncertainty you cannot reliably detect the small differences in expected value, least of all how each will actually turn out. So you flip a coin among them. All meaningful decision-making time belongs to the sorting and to enlarging the good bucket with new candidates, not to splitting hairs among options already inside it.
Origin
Annie Duke articulates this in How to Decide as a way to combat choice paralysis, grounded in her poker and cognitive-science view that under uncertainty near-equal options can't be reliably separated.
Core principles
- 01You can't reliably split apart near-equal options under uncertainty
- 02Time spent picking between good options is time stolen from finding new good options
- 03The 'good enough' bucket always holds more than one option
- 04More information shrinks the bucket but never to a single option
How to run it
- 1
Enumerate the options
Lay out the choices actually available for the decision at hand.
- 2
Sort into like / don't-like buckets
Assign each option to a bucket of things you'd genuinely be happy with and a bucket of things you wouldn't.
- 3
Apply the Only-Option Test to borderline cases
Ask: if this were the only thing I could choose, would I be happy? A yes puts it in the good bucket; a no sorts it out.
Pro tip This test converts a vague comparison into a clean, single-item judgment your gut can answer.
- 4
Stop ranking inside the good bucket
Accept that under uncertainty you cannot reliably tell apart the expected value of items already in the good bucket.
Watch out The bucket always holds more than one option; pretending you can split them is where you lose time.
- 5
Flip a coin, breaking ties on optionality
Choose among good-bucket items at random, or lightly favor the one that's more quittable or can be done in parallel.
- 6
Reinvest time in growing the bucket
Redirect saved attention to finding new options worth adding, which is where real value lives.
In the wild
Faced with a long menu, instead of comparing two finalists dish by dish, you scan and drop each item into 'would enjoy' or 'wouldn't.' For anything on the fence you ask the Only-Option Test: if this were the only thing I could order, would I be happy? The 'yes' items form a bucket of several acceptable dishes. Rather than agonizing over which is marginally best, you simply pick one and return to the conversation.
→ A fast, low-regret choice and attention freed for higher-value things.
Common mistakes
Splitting hairs inside the good bucket
Trying to rank near-equal options pretends to a precision you don't have under uncertainty and drains time from finding new options.
Polling everyone for the 'right' answer
Other people have different tastes; canvassing them is a symptom of treating a sorting problem as a picking problem.
Is it for you?
Best for
High-frequency choices among several genuinely acceptable options where fidelity between them is unknowable.
Not ideal for
Decisions with large, forecastable differences in value between the options on the table.
From the transcript
“your your decision making time should be spent on the sorting but not on the picking”
“the only option test uh which is essentially to take anything out of that bucket so we can think about like a venue right if…”
From the episode
How To Make Better Decisions - Annie Duke - #233
Annie Duke