Scott Alexander argues for the use of probabilities in decision-making even without well-defined models, using examples to show why this approach is necessary and beneficial in real-world situations. Longer summary

Scott Alexander argues against the idea that probabilities should not be used without well-defined models. He presents scenarios where probability judgments are necessary even in novel situations, such as alien contact or evaluating research proposals. The post discusses the importance of making probability estimates in real-world decisions, even when precise models are unavailable. Scott acknowledges some valid criticisms but ultimately argues that using probabilities allows for better decision-making, accuracy checking, and challenging overconfidence. He suggests that alternatives to probabilistic thinking fall short when faced with real-world choices. Shorter summary

Scott Alexander discusses the problem of overconfidence in probability estimates, arguing that extreme certainty is rarely justified, especially for complex future predictions. Longer summary

Scott Alexander discusses the problem of overconfidence in probability estimates, particularly when people claim to be extremely certain about complex future events. He explains how experiments show that people are often vastly overconfident, even when they claim 99.9999% certainty. Scott argues that extreme confidence is rarely justified, especially for predictions about technological progress or societal changes. He suggests that overconfidence contributes to intolerance and close-mindedness, and that studying history can help reduce overconfidence by showing how often confident predictions have been wrong in the past. Shorter summary