This is an idea that I frequently talk about to students and colleagues that I stole from Tyler Cowen who may have stole it from someone else. It goes along with a lot of how I’ve been thinking about improving rationality through culture rather than education. In other words, it’s very difficult to increase a student’s raw ability to better solve problems, so a possibly more effective way for improving rationality is to incentivize students to adopt common rational decision-making heuristics through the norms and culture of a school. You may not be able to really help students be smarter, but you may be able to help them to be urged by their peers to think smarter.
One of these helpful heuristics is the “60/40 belief”. Instead of pushing students to tell me what they know, I push them to assign me likelihoods about their confidence in their beliefs. And personally, I never assign belief confidences above 60%. So all of my statements in the classroom are framed as 50–60% confidence — except of course about the actual math content that I teach, that’s another story. Now those are admittedly low confidence, but that’s okay and students need to see more of that in their authority figures.