On the other hand, if we think of all possible statements “for all X, Q(X)”, there are more was to be false than true. Infinity more ways, even,
If you consider all syntactically valid Q in some sort of math notation, no matter the length, there’s the fraction that is simply 0=0*( ….....… X somewhere here........) , or your favourite less trivial statement of choice. Ditto for Turing machine tapes et cetera. There’s certainly a nonzero probability of constructing Q(X) that holds for all X.
That is a more detailed model than the robot uses. What it means by the ways to be false or true is more like
“true for 1, true for 2, true for 3, false for 4, true for 5...” The robot can’t look inside the statements while it’s doing probabilistic logic, it can only look at truth values and relationships.
On the other hand, the power of doing that is certainly a good reason to upgrade the robot :)
If you consider all syntactically valid Q in some sort of math notation, no matter the length, there’s the fraction that is simply 0=0*( ….....… X somewhere here........) , or your favourite less trivial statement of choice. Ditto for Turing machine tapes et cetera. There’s certainly a nonzero probability of constructing Q(X) that holds for all X.
That is a more detailed model than the robot uses. What it means by the ways to be false or true is more like
“true for 1, true for 2, true for 3, false for 4, true for 5...” The robot can’t look inside the statements while it’s doing probabilistic logic, it can only look at truth values and relationships.
On the other hand, the power of doing that is certainly a good reason to upgrade the robot :)