In this thread below, I didn't really explore this comment, which I think is the most original thing I say in the thread. > Prediction becomes weird in a context where you have partial control, because you simultaneously want to be wrong while having people take you seriously.
Why disaster prediction is cybernetically weird
10 tweets · April 2020 · 19 likes · 0 retweets · read on Twitter
What is the weird thing here? It's a cybernetic thing. Control systems. Ha! It's related to the same reason why Predictive Processing language sounds confusing! Perceptual Control Theory is structurally analogous but phrased in a less confusing way. see slatestarcodex.com/2019/03/20/tra…
Planning to avoid disaster is very loopy! ➿ What you end up saying is "on the basis of something I think would happen if I didn't do what I'm going to do (to keep it from happening because I don't want it to happen) I'm going to do something that will prevent it from happening"
There's an important sense in which the desired outcome involves never finding out if original prediction was right! In a complicated system, you might, if you can run an adequate counterfactual simulation with new knowledge. But not in a complex system. twitter.com/helenbevan/sta…
Obviously ideally we get both: - minimal deaths - AND we'd get to find out whether our actions were indeed necessary for that (and if we took multiple actions for redundancy, which ones helped most) But the more you increase redundancy, the harder it is to tell what worked.
This is so on both governmental policy scales, and a single household. If we: - avoid public gatherings - wear masks at grocery store - wash groceries & deliveries on the way in - copper-tape handles - boost immunity w vitamins & nobody in house gets sick... what made the diff?
And yet if we do all of those things, and somebody does get sick anyway, it can still be really hard to know where the gap was—and the kicker is: *the more precautions we took, the harder it is to know where the virus got through anyway*
I guess this is actually sort of a U-curve: - no precautions means 🤷♀️, could have been anything (or many things) - some precautions means🤦♀️it was probably the things we didn't do - all known precautions means 🧐 wtf did we miss?
This is especially confusing when there are many actors. It might have been that your actions were redundant—if you hadn't got it, someone else might've. I think there's been a lot of that with covid, unconsciously: "it won't be so bad [because surely someone will do something]"
This apparently-inherent tradeoff seems like the sort of thing that Complexity researchers & Systems Thinkers might have a term for already—if so, I'd love to know what it is! @jim_rutt @NoraBateson @TaylorPearsonMe @cwodtke @RichDecibels @edelwax @snowded @Timber_22