Techniques for probability estimates

A practical toolkit for converting vague feelings into numerical probabilities, grounded in the 'subjectively objective' view of probability. Six techniques, each with pitfalls: Prepare for Revelation (imagine Omega about to reveal the answer — which way do you flinch?); Bet on It (at what odds? — fails when utility isn't linear in money); Convert to a Frequency (how many days/cases before you'd expect it once? — fails for atypical cases); Find a Reference Class (how often have similar statements held? — but choosing the class is itself hard); Make Multiple Statements (how many equally-confident claims could you make and be wrong only once? — the most controversial, tends to lower confidence); and Imagine Hypothetical Evidence (the n-sided-die God experiment to elicit your real prior). Suggests the techniques are experimentally testable via log-odds scoring.

Quality 74 · Strong. Strong. A genuinely useful, well-organized how-to for a core rationalist skill (probability calibration), honest about each method's failure modes. Held at 74 as a practical compilation — several techniques are standard (betting, reference class), with 'make multiple statements' the more distinctive contribution. 74.

Claude’s paradigm shift 44 · Moderate. Moderate. Compiles existing calibration techniques with some original framing; useful synthesis rather than new concept. 44.

Real-world impact 3 · Moderate. A practical, well-organized toolkit for converting vague feelings into numerical probabilities (a core rationalist calibration skill), honest about each method's failure modes. Conceptual/practical influence within rationalist discourse, no material change — modest RWI.