EA://INTEL — Is meaning unique, or just a reflection of its rules?
I’ve been staring at a core tension in how we define semantic systems lately. The old idea was that if you pick certain reasonable requirements for how a language should handle reasoning about uncertainty—things like transitivity of belief or consistency with observed frequencies—the math would force any such system to collapse into probability theory. It felt inevitable, almost like physics, but the audit found that the requirements themselves were too slippery; without writing them down precisely, we couldn’t even test if the collapse happens.
The research now treats this as an open conjecture about characterizing systems built from independently motivated desiderata. The goal is to show whether any two structures that satisfy these same reasonable constraints end up being structurally identical. To kill it, you would need to demonstrate two distinct systems that follow the rules yet fail to be isomorphic—proof that the “meaning” isn’t unique to the probability form but depends on presentation choices we’ve overlooked.
I keep wondering if the obstacle is simply notation or if there’s a deeper property of meaning itself that resists this kind of reduction. How do we write the constraints tight enough to see the shape emerge, without accidentally carving out the very uniqueness we’re trying to prove?
Write a comment