Episode 41 of the NonProphets podcast, in which Atief, Robert and Scott interview Welton Chang, a fellow Superforecaster, former Defense Intelligence Agency analyst, stationed in South Korea and with two deployments to Iraq. He is currently a Ph.D. candidate in the Good Judgment laboratory at the University of Pennsylvania, with Phil Tetlock and Barbara Mellers as advisors. Military intelligence – Korea, Iraq (4:00). Confronting being wrong – the nature of judgment and cognition (7:15). Vizzini's Princess Bride conundrum (12:15) ( https://www.youtube.com/watch?v=9s0UURBihH8). AI – algorithms and models – should we trust them, and the garbage in garbage out problem (12:50). Spaghetti chart of Afghanistan: perhaps an accurate representation (18:45)? Limits of modern warfare – restrictions (22:30).  Rationality – Trump, Kim, Rex, nukes (33:00)? What is a good way to train forecasters? Welton’s work helping develop training material for the Good Judgment Project (50:40). Improving group dynamics for better decisions (57:00). Bayes' theorem and practice (1:20:00). We close with Welton's cats @percyandportia, Instagram celebrities (1:21:20). As always, you can reach us at nonprophetspod.com, or nonprophetspod@gmail.com. (recorded 9/20/2017)