You can’t. But for example if they say what they rolled and you assume there is a correlation what they actually get and what they would say then you have a chance to narrow it down. If you know it’s not corrrelated to anything (ie is pure dice) you know it can’ t be evidence.
You can guess. You can roll the die yourself (and guess that it came up the same way). You can also examine the die, and then guess.
If I throw an additional dice it doesn’t help determine what already thrown dice are. Your expectation doesn’t shift so no probability can shift.
Also, this contains some assumptions that aren’t always correct. I can throw a die a bunch of times, and notice that it comes up “6” or “1″ an awful lot an conclude it’s weighted. (A shift in expectation.)
A guess would be equally good without dice throwing. Indeed if you have access to the dice that generated the result we want to know about you can infer distribution information. But if you have a different die and determine that it’s weighted it doesn’t tell whether the orignal die is weighted. If you knew the dice came from the same factory you could infer something. But you manufacturing a fresh dice is justified to assume to not be distributionally connected. If you have information that you know to correlate your manunfacturing process to be similar then that contains your information and the actual rolling of the die doesn’t tell you anything.
You can’t. But for example if they say what they rolled and you assume there is a correlation what they actually get and what they would say then you have a chance to narrow it down. If you know it’s not corrrelated to anything (ie is pure dice) you know it can’ t be evidence.
You can guess. You can roll the die yourself (and guess that it came up the same way). You can also examine the die, and then guess.
Also, this contains some assumptions that aren’t always correct. I can throw a die a bunch of times, and notice that it comes up “6” or “1″ an awful lot an conclude it’s weighted. (A shift in expectation.)
A guess would be equally good without dice throwing. Indeed if you have access to the dice that generated the result we want to know about you can infer distribution information. But if you have a different die and determine that it’s weighted it doesn’t tell whether the orignal die is weighted. If you knew the dice came from the same factory you could infer something. But you manufacturing a fresh dice is justified to assume to not be distributionally connected. If you have information that you know to correlate your manunfacturing process to be similar then that contains your information and the actual rolling of the die doesn’t tell you anything.