NikosTsompanides
UserA practical framework for making better decisions under uncertainty.
Categories
Indexed Skills (5)
bet-evaluate
Size a decision that has been framed — calibrate confidence and say what supports it, enumerate the stakes and blast radius, classify how reversible it is, surface the alternatives including one with more upside than the obvious choice, and recommend whether to proceed now, run an experiment, or gather more evidence first. Use when someone asks how risky something is, how confident they should be, weighs options against each other, or has a framed decision that needs a verdict on how much evidence to require before committing. Not for framing the question itself — that is bet-frame.
bet-experiment
Design the cheapest test that would meaningfully reduce the uncertainty actually deciding a question, with explicit success and failure criteria set before it runs — or say plainly when testing would cost more than the uncertainty is worth. Also produces the strongest counterargument against the current direction and the evidence that would falsify it. Use when someone asks how to find out, wants to de-risk before committing, needs a spike, prototype, pilot, benchmark, or customer validation, or wants their reasoning attacked rather than supported.
bet-frame
Frame a consequential decision as a bet before committing to it — sharpen a vague question into an answerable one, separate what is known from what is merely assumed, and name the unknowns that actually matter. Use when someone is weighing a decision, says they are torn or unsure, asks whether to do something consequential, or presents a solution whose underlying problem has not been established. Domain-neutral — engineering, hiring, product, strategy, personal. This is the entry point; bet-evaluate sizes the bet afterwards.
bet-review
Review a decision whose outcome is now known, judging the reasoning on the information available when it was made rather than on how it turned out. States explicitly when a good decision produced a bad outcome or a bad decision got lucky, and extracts the beliefs worth carrying forward. Use for retrospectives, post-mortems, looking back on a call that has resolved, or when someone asks whether they got something wrong after a result came in. Not for revising a belief mid-flight while the outcome is still open — that is bet-update.
bet-update
Commit to a decision on the evidence available — recording confidence, what remains unknown, and what specifically would change your mind — and later, revise that belief when evidence actually arrives. Use when someone is ready to decide, asks what to do now, wants a decision written down, or brings new information bearing on a decision already made. Not for judging a decision after its outcome is known, which needs a posture that resists hindsight — that is bet-review.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.