I just recently read this article, and I think it’s a pretty important piece for thinking about how good should we expect our forecasts about the future of TAI to be. It honestly updated me on something that I was more skeptical of before.
But I also think there’s a set of futurists that was missed here: the founders of the United States. The founders were able to foresee pretty far-out, non-obvious things about what would happen to their government structures as increasing competitive/adversarial pressure was placed on them, e.g. large republics being more stable, parties overriding branch loyalty, and a propertyless majority within a century.
In some ways I think the US founders’ case is actually more similar to the situation we’re in with TAI futurism, in that there were more skilled people debating the relevant questions and the stakes were higher than a sci-fi writer’s offhand predictions (especially bc it’s unclear how much effort was being put in and we don’t know how hard they were trying to be right versus interesting).
I think this should also be taken as evidence in favor of being able to predict the future, in the same vein as this blogpost.
A few critiques (and rejoinders) come to mind, though:
Selection effects: lots of people made predictions about governments, both laypeople and people who eventually affected governments in large ways, and we’re picking the most successful government.
I don’t think this is that big of a deal. There weren’t many people who had the chance to affect large-scale governments the way the US founders did, so the pool to select from is fairly small. And a lot of the best predictions came from the anti-federalists, who lost, so they aren’t selected on success at all.
There were federalists and anti-federalists, and both made accurate and inaccurate predictions, so “someone in the debate was right” is cheap.
Fair, but the debate was over the right questions, and what actually happened was usually one of the positions on the table rather than something neither side considered. Some people critique AGI futurism by saying we can’t say much because we won’t even know the relevant considerations; that really doesn’t seem to be the case for the founders. My guess is a proper quantitative assessment would put them well above chance, though I haven’t actually tried to run the numbers.
There were a bunch of ways the founders were very wrong, some of which still shape how government works today (Madison thought the legislature would be the encroaching branch; the Electoral College was supposed to filter demagogues; amendments were supposed to be more frequent).
Definitely true, but I’m not claiming they got everything right or that they were thinking about the relevant considerations correctly across the board.
I just recently read this article, and I think it’s a pretty important piece for thinking about how good should we expect our forecasts about the future of TAI to be. It honestly updated me on something that I was more skeptical of before.
But I also think there’s a set of futurists that was missed here: the founders of the United States. The founders were able to foresee pretty far-out, non-obvious things about what would happen to their government structures as increasing competitive/adversarial pressure was placed on them, e.g. large republics being more stable, parties overriding branch loyalty, and a propertyless majority within a century.
In some ways I think the US founders’ case is actually more similar to the situation we’re in with TAI futurism, in that there were more skilled people debating the relevant questions and the stakes were higher than a sci-fi writer’s offhand predictions (especially bc it’s unclear how much effort was being put in and we don’t know how hard they were trying to be right versus interesting).
I think this should also be taken as evidence in favor of being able to predict the future, in the same vein as this blogpost.
A few critiques (and rejoinders) come to mind, though:
Selection effects: lots of people made predictions about governments, both laypeople and people who eventually affected governments in large ways, and we’re picking the most successful government.
I don’t think this is that big of a deal. There weren’t many people who had the chance to affect large-scale governments the way the US founders did, so the pool to select from is fairly small. And a lot of the best predictions came from the anti-federalists, who lost, so they aren’t selected on success at all.
There were federalists and anti-federalists, and both made accurate and inaccurate predictions, so “someone in the debate was right” is cheap.
Fair, but the debate was over the right questions, and what actually happened was usually one of the positions on the table rather than something neither side considered. Some people critique AGI futurism by saying we can’t say much because we won’t even know the relevant considerations; that really doesn’t seem to be the case for the founders. My guess is a proper quantitative assessment would put them well above chance, though I haven’t actually tried to run the numbers.
There were a bunch of ways the founders were very wrong, some of which still shape how government works today (Madison thought the legislature would be the encroaching branch; the Electoral College was supposed to filter demagogues; amendments were supposed to be more frequent).
Definitely true, but I’m not claiming they got everything right or that they were thinking about the relevant considerations correctly across the board.