There are so many biases that affect individuals (hindsight, hyperbolic discounting, confirmation, etc) that to get correct answers, you have to look at and trust other accounts. You have to do studies, you have to trust those studies, and you have to track actual hard evidence.
But large enough systems become increasingly opaque, and it can be hard to test your web of trust empirically. This reduces huge organizations to cargo cult behavior, or worse, fundamentally unsound behavior because "it worked for me."
I don't know what to do about this besides what CFAR is trying, which is to get people better at weighing evidence and changing their damned minds. You have to constantly be asking yourself: "what would the world look like if this were true? What would the world like like if it wasn't?" It is hard work, but I think it is critical to our continued improvements as a civilization and a species.