Probabilistic Thinking for Enhanced Decision-Making

Who created and checked this page

Correction history

Every accepted correction to this page is recorded with the exact change, so readers can see how the page improved over time.

1. 28 July 2026 · corrected by Emir Baycan

   Corrected two arithmetic/scale errors: fixed 'the top quartile of Good Judgment Project performers were selected as superforecasters' to 'roughly the top 2%,' since superforecaster designation was far more selective than a quartile (25%); and corrected 'one project abandoned per ten initiated is a 90 percent attrition rate' to the arithmetically correct 'about a 10 percent attrition rate.' The Recode 2018 analysis of Amazon's ~30 discontinued product lines and the informal, explicitly-labeled 'not full math' Bayesian-updating illustration were accurate and left in place.

   Before

   The article said the Good Judgment Project's top quartile became superforecasters, and that Astro Teller's 'one in ten' Google X project-abandonment rate is a 90% attrition rate.

   After

   Corrected the superforecaster selectivity to roughly the top 2% and fixed the attrition-rate arithmetic (1 in 10 is 10%, not 90%).

   Why: One figure overstated how selective superforecaster status was, and the other was a basic arithmetic error (confusing 1-in-10 with 90%). Corrections fix the math rather than adding new sources.

   View the full record →

Full contribution timeline

1. Correction 28 July 2026

   Emir Baycan: Corrected two arithmetic/scale errors: fixed 'the top quartile of Good Judgment Project performers were selected as superforecasters' to 'roughly the top 2%,' since superforecaster designation was far more selective than a quartile (25%); and corrected 'one project abandoned per ten initiated is a 90 percent attrition rate' to the arithmetically correct 'about a 10 percent attrition rate.' The Recode 2018 analysis of Amazon's ~30 discontinued product lines and the informal, explicitly-labeled 'not full math' Bayesian-updating illustration were accurate and left in place.

   Proof record →