Re:Asteroid ZTm0038 with a >3% impact probability
by bulatb on Wednesday August 16, 2023 @03:57PM
(#37152823)
Someone tells you, "Hey, there's a car." And walks away. Did they mean, like, there exists a car? Presumably they meant there's a car on the road, but who knows. On this road? Was it moving? Yes, No? Maybe?
There are parts of the road you can't see. If you were forced to estimate the chance you'd have to move for this particular purported car with just this information, you might throw up your hands and say P(they're telling the truth) * 1/4 until you can see it.
If a bunch of people like to lie, and you have nothing but one person's word, your move/no-move prediction for this car is likely to be wrong despite your model of the road-car-person system being perfectly right.
As more people say they saw the car, the estimate from these reports approaches 1/4, dominated by your understanding of the system. But you might have fewer early estimates that say you need to move per actual car and actual move, and many estimates for cars that don't exist.
I hope we're in a world where that makes sense and I'm actually right about this. But who knows!
There are parts of the road you can't see. If you were forced to estimate the chance you'd have to move for this particular purported car with just this information, you might throw up your hands and say P(they're telling the truth) * 1/4 until you can see it.
If a bunch of people like to lie, and you have nothing but one person's word, your move/no-move prediction for this car is likely to be wrong despite your model of the road-car-person system being perfectly right.
As more people say they saw the car, the estimate from these reports approaches 1/4, dominated by your understanding of the system. But you might have fewer early estimates that say you need to move per actual car and actual move, and many estimates for cars that don't exist.
I hope we're in a world where that makes sense and I'm actually right about this. But who knows!