No rationality, or no Bayesianism? Rationality is a general term for reasoning about reality. Bayesianism is the specific school of rationality advocated on LessWrong.
A “world in which there was no rationality” is not even meaningful, just like “world in which there was no physics” is meaningless. Even if energy and matter behaves in a way that’s completely alien to us, there are still laws that govern how it works and you can call these laws “physics”. Similarly, even if we’d live in some hypothetical world where the rules of reasoning are not derived from Bayes’ theorem, there are still rules that can be thought of as that reality’s rationalism.
A world without Bayesianism is easy to visualize, because we have all seen such worlds in fiction. Cartoons takes this to the extreme—Wile E. Coyote paints a tunnel and expects Road Runner to crash into it—but Road Runner manages to go through. Then he expects that if Road Runner could go through, he could go through as well—but he crashes into it when he tried.
Coyote’s problem is that his rationalism could have worked in our world—but he is not living in our world. He is living in a cartoon world with cartoon logic, and needs a different kind of rationalism.
Like… the one Bugs Bunny uses.
Bugs Bunny plugs Elmer Fudd’s rifle with his finger. In our world, this could not stop the bullet. But Bugs Bunny is not living in our world—he lives in cartoon world. He correctly predicts that the rifle will explode without harming him, and his belief in that prediction is strong enough to bet his life on it.
Now, one may claim that it is not rationality that gets messed up here—merely physics. But in the examples I picked it is not just that laws of nature that don’t work like real world dwellers would expect—it is consistency itself that fails. Let us compare with superhero comics, where the limitations of physics are but a suggestion but at least some effort is done to maintain consistency.
When mirror master jumps into a mirror, he uses his technology/powers to temporarily turn the mirror into a portal. If Flash is fast enough, he can jump into the mirror after him, before the mirror turns back to normal. The rules are simple—when the portal is open you can pass, when it’s closed you can’t. Even if it doesn’t make sense scientifically it makes sense logically. But there are no similar rules that can tell Coyote whether or not its safe to pass.
Superman can also plug his finger into criminals’ guns to stop them from shooting, just like Bugs Bunny. But Superman can stop the bullets with any part of his body, before or after they leave the barrel. So him successfully plugging the guns is consistent. Bugs Bunny, however, is not invulnerable to bullets. When Elmer Fudd chases after him, rifle blazing, Bugs Bunny runs for his life because he know the bullets will pierce him. They are stronger than his body can handle. Except… when he sticks his finger into the barrel. Not consistent.
Still—there are laws that govern cartoon reality. Like the law of funny. Bugs Bunny is aware of them—his actions may seem chaotic when judged by our world’s rationality, but they make perfect sense in cartoon world. Wile E. Coyote’s actions make perfect some sense in our world’s rationality, but are doomed to fail when executed under cartoon world logic.
Had I lived in cartoon world, I’d rather be like Bugs Bunny than like Wile E. Coyote. Not to insist on Bayesianism even though it wouldn’t work, but try to figure out how reasoning in that reality really works and rely on that.
Then again—wouldn’t Bayesianism itself deter me from relying on things that don’t work? Is Wile E. Coyote even Bayesian if he doesn’t update his believes every time his predictions fail?
I’m no longer sure I can imagine a world where there is no Bayesianism...
No rationality, or no Bayesianism? Rationality is a general term for reasoning about reality. Bayesianism is the specific school of rationality advocated on LessWrong.
A “world in which there was no rationality” is not even meaningful, just like “world in which there was no physics” is meaningless. Even if energy and matter behaves in a way that’s completely alien to us, there are still laws that govern how it works and you can call these laws “physics”. Similarly, even if we’d live in some hypothetical world where the rules of reasoning are not derived from Bayes’ theorem, there are still rules that can be thought of as that reality’s rationalism.
A world without Bayesianism is easy to visualize, because we have all seen such worlds in fiction. Cartoons takes this to the extreme—Wile E. Coyote paints a tunnel and expects Road Runner to crash into it—but Road Runner manages to go through. Then he expects that if Road Runner could go through, he could go through as well—but he crashes into it when he tried.
Coyote’s problem is that his rationalism could have worked in our world—but he is not living in our world. He is living in a cartoon world with cartoon logic, and needs a different kind of rationalism.
Like… the one Bugs Bunny uses.
Bugs Bunny plugs Elmer Fudd’s rifle with his finger. In our world, this could not stop the bullet. But Bugs Bunny is not living in our world—he lives in cartoon world. He correctly predicts that the rifle will explode without harming him, and his belief in that prediction is strong enough to bet his life on it.
Now, one may claim that it is not rationality that gets messed up here—merely physics. But in the examples I picked it is not just that laws of nature that don’t work like real world dwellers would expect—it is consistency itself that fails. Let us compare with superhero comics, where the limitations of physics are but a suggestion but at least some effort is done to maintain consistency.
When mirror master jumps into a mirror, he uses his technology/powers to temporarily turn the mirror into a portal. If Flash is fast enough, he can jump into the mirror after him, before the mirror turns back to normal. The rules are simple—when the portal is open you can pass, when it’s closed you can’t. Even if it doesn’t make sense scientifically it makes sense logically. But there are no similar rules that can tell Coyote whether or not its safe to pass.
Superman can also plug his finger into criminals’ guns to stop them from shooting, just like Bugs Bunny. But Superman can stop the bullets with any part of his body, before or after they leave the barrel. So him successfully plugging the guns is consistent. Bugs Bunny, however, is not invulnerable to bullets. When Elmer Fudd chases after him, rifle blazing, Bugs Bunny runs for his life because he know the bullets will pierce him. They are stronger than his body can handle. Except… when he sticks his finger into the barrel. Not consistent.
Still—there are laws that govern cartoon reality. Like the law of funny. Bugs Bunny is aware of them—his actions may seem chaotic when judged by our world’s rationality, but they make perfect sense in cartoon world. Wile E. Coyote’s actions make
perfectsome sense in our world’s rationality, but are doomed to fail when executed under cartoon world logic.Had I lived in cartoon world, I’d rather be like Bugs Bunny than like Wile E. Coyote. Not to insist on Bayesianism even though it wouldn’t work, but try to figure out how reasoning in that reality really works and rely on that.
Then again—wouldn’t Bayesianism itself deter me from relying on things that don’t work? Is Wile E. Coyote even Bayesian if he doesn’t update his believes every time his predictions fail?
I’m no longer sure I can imagine a world where there is no Bayesianism...