Sarajevo 1914: Black Swan Questions
post by JohnBuridan · 2024-06-21T21:27:43.299Z · LW · GW · 0 commentsContents
A Disaster will come… eventually None No comments
The 110th Anniversary of Sarajevo is next Friday.
To commemorate I am thinking about the following questions.
- What does the sentence: The assassination of archduke Ferdinand caused WWI mean? Is it true?
- Was World War I inevitable?
- Given the invention of the tools that make things like machine guns were machine guns inevitable?
- Was the invention of machine guns in any sense good? Or wholly bad?
- Is it possible to prepare for events outside the training distribution?
- What were the greatest unexpected positive developments in history?
- For well-defined questions, is there a point in predicting things further than a year out?
- Is the statement, “If we don’t build powerful AI, then China will” the type of statement that can be true or false?
- Does predicting conflict with someone make conflict more likely?
- Does preparing for conflict between nations make conflict more likely?
- Does trusting someone in interpersonal affairs make them more trustworthy?
- If everything is ultimately doomed, does that effect what right action is today?
- Which artistic movement has caused the most damage in history? And what does that question mean, properly speaking?
- Which philosophical movement caused the most damage in history?
You probably have thoughts on some of these.
A Disaster will come… eventually
Philip Tetlock has a section in Superforecasting about how pundits make predictions that are not time-bound, and thus never proven wrong. A prediction without a deadline is useless.
Alternatively, in Antifragile Nassim Taleb espouses the view that knowing an eventuality is a powerful source of knowledge, because it means you can build a system robust for it. Scenario planning prepares one for many eventualities. If you can create a career or organization that adapts quickly to external forces, then you have done the only planning that matters.
However, neither prediction nor robustness planning are the whole story of what one needs to do to endure. Why? They rely upon two assumptions. 1) a stable signal and 2) a knowable distribution. In Tetlock’s work, even the best forecasters become close to 50% at 9 months to a year out. And in Taleb’s work there is always the danger of a black swan. You can plan for everything inside the distribution of known events, but nothing will prepare you for what is outside the distribution of what you have perceived. And there is a lot that exists outside that distribution. The combinatorial possibilities can explode in the face of scenario planners, especially when the stakes get high and things get weird!
But both these assumptions are not true exactly.
1) 30-year mortgages and pensions and actuaries and fixed-income investments exist. People plan around them. You can think it foolish, but there is an industry that lives years in the future. Maybe they are not forecasting anything that interesting. The same legally binding promises will exist in the future. This type of forecasting might not seem super, since it is so quotidian, but it is superb at conserving the future.
2) Similarly, is it really so impossible to prepare for things outside the distribution? Savings, capital, liquidity, intelligence, force, compassion, and personnel. There are very, very few problems that won’t bend - at least a little - when you have these raw resources in reserve.
----
0 comments
Comments sorted by top scores.