3 ms·

Sure, but ideologically motivated reasoning will only provide "broken clock" predictions. What good forecasters would do with hard-to-predict events like the pa

Sure, but ideologically motivated reasoning will only provide "broken clock" predictions. What good forecasters would do with hard-to-predict events like the pandemic, the next recession or political revolution is to admit ignorance and revert to the base rate (e.g. once a century on average). Truly black swan events are things we don't even know we should try to predict. An example of ideologically motivated reasoning would be "The US is about to collapse because Biden/Trump is president". Unfortunately many if not most people think like this, that's one reason why so few are good at forecasting according to the book.

Fluorescence 3y ago
> will only provide "broken clock" predictions I think it's a bit more than a random prediction, it's more like a heuristic. Imagine the tree of all possibilities and all subsequent possibilities below those etc. It's just too vast to explore. The ideologue is not random a path through the tree but biased in a very human way with pessimistic assumptions about human motivations and behaviour that are sometimes true. A heuristic can be valuable without achieving more truth than fiction, just better than random. > ideologically motivated reasoning would be "The US is about to collapse because Biden/Trump is president" What I have in mind is the ideological forecaster would start with a set of assumptions and try to justify them against the odds e.g. they assume "Biden/Trump is a monster", they adopt the specific mindset of this monster and interpret the known events, which seem entirely innocuous on the surface, as contributing to a larger nefarious strategy and work through a chain of consequences to arrive at a compelling theory that hangs together. It's like exploring many moves ahead in chess without an evidence based goal function justifying continuing down a specific path. A reasonable individual might not presume others are monsters or pursue a line of reasoning so unjustified for so long. The heuristic works because sadly "X is a monster" and "innocuous actions are actually part of a nefarious plan" is sometimes true. I'm essentially defending conspiratorial thinking as a good thing in an adversarial situation so long as it's integrated sanely and not allowed to rule the day alone. If you don't have ideologues of different flavours in your intelligence unit, you might need to create them.
- martindbp 3y ago
  I think what you're describing is a sort of ensemble. I agree sometimes "conspiratorial" thinking is right, as in the invasion of Ukraine, or Hitler. Then the difficult part becomes how to weight the different world views in this particular time in history. But I agree at the very least you need to be able to simulate the thinking of ideologues in order to understand their thinking and actions. The invasion of Ukraine is a very interesting example where people underestimated how nefarious Putin really was, but if you read his writings on Ukraine, and do a bayesian update after Georgia and Crimea/Donbas the invasion shouldn't have been all that surprising. At least we should have assigned something like 10-15% chance.

Fluorescence 3y ago

> will only provide "broken clock" predictions I think it's a bit more than a random prediction, it's more like a heuristic. Imagine the tree of all possibilities and all subsequent possibilities below those etc. It's just too vast to explore. The ideologue is not random a path through the tree but biased in a very human way with pessimistic assumptions about human motivations and behaviour that are sometimes true. A heuristic can be valuable without achieving more truth than fiction, just better than random. > ideologically motivated reasoning would be "The US is about to collapse because Biden/Trump is president" What I have in mind is the ideological forecaster would start with a set of assumptions and try to justify them against the odds e.g. they assume "Biden/Trump is a monster", they adopt the specific mindset of this monster and interpret the known events, which seem entirely innocuous on the surface, as contributing to a larger nefarious strategy and work through a chain of consequences to arrive at a compelling theory that hangs together. It's like exploring many moves ahead in chess without an evidence based goal function justifying continuing down a specific path. A reasonable individual might not presume others are monsters or pursue a line of reasoning so unjustified for so long. The heuristic works because sadly "X is a monster" and "innocuous actions are actually part of a nefarious plan" is sometimes true. I'm essentially defending conspiratorial thinking as a good thing in an adversarial situation so long as it's integrated sanely and not allowed to rule the day alone. If you don't have ideologues of different flavours in your intelligence unit, you might need to create them.

martindbp 3y ago
I think what you're describing is a sort of ensemble. I agree sometimes "conspiratorial" thinking is right, as in the invasion of Ukraine, or Hitler. Then the difficult part becomes how to weight the different world views in this particular time in history. But I agree at the very least you need to be able to simulate the thinking of ideologues in order to understand their thinking and actions. The invasion of Ukraine is a very interesting example where people underestimated how nefarious Putin really was, but if you read his writings on Ukraine, and do a bayesian update after Georgia and Crimea/Donbas the invasion shouldn't have been all that surprising. At least we should have assigned something like 10-15% chance.