Sadly, we're only able to remember minority mistakes.

The mistakes of majorities are shadowy, elusive, as hard to remember as last night's dreams. Come to think of it, were they mistakes at all? No, on second thought, it was very complicated at the time, and perhaps we were right in principle.

(Seriously, though: we've memory-holed how confidently wrong people were on this topic, because it's a case where *most of us* were confidently wrong. We should adjust priors so a lot more of the probability mass is on "I don't know" and "I could well be surprised.")

How do you handle this problem: You are interviewed and asked about the future of–let's say the humanities or literature–and you outline scenarios but are eventually reported in public such that ''the probability mass on 'I don't know''' is mostly erased (because people want to know the future)?

Thank you @christophheilig.bsky.social for making this fabulous workshop on the relationships between narrative and AI real!

I haven’t known, however, that the green button is the solution for: "Seek the Lord while he may be found"