(This article was linked by QuantumDelta on IRC)
http://www.physorg.com/news205133042.htmlAs far as astrophysicists can tell, the universe is expanding at an accelerating rate, and will likely continue to do so indefinitely. But now some physicists are saying that this theory, called eternal inflation, and its implication that time is endless pose a problem for scientists calculating the probability of any event occurring. In a recent paper, they calculate that time is likely to end within the next 5 billion years due to some type of catastrophe that no one alive at the time will witness.
Basically, they seem to be saying (and I admit I have never before heard of "eternal inflation's implications in
probability"), that because you can't compare probabilities using relative frequency across infinite time frames (you'd end up with infinity divided by infinity), time has to
end somehow, involving this thermalization barrier thingie which sounds kind of like reaching heat death asymptotically, at least from what the article said.
Um, I thought that the only way to measure probability- regardless of an infinite universe- was by relative probabilities during a finite time period.
They don't claim that their conclusion that time will end is correct, only that it follows logically from a set of assumptions. So perhaps one of the three assumptions underlying the conclusion is incorrect instead.
The first assumption is that the universe is eternally inflating, which is a consequence of general relativity and well supported by the experimental evidence so far observed. The second assumption is that the definition of probability is based on the relative frequency of an event, or what the scientists call the assumption of typicality. The third assumption is that, if spacetime is indeed infinite, then the only way to determine the probability of an event is to restrict one's attention to a finite subset of the infinite multiverse. Some other physicists have already looked into alternatives to this third assumption.
The first assumption is the leading theory about the universe IIRC, and the second seems pretty self-evident to anyone who knows what probability
is. But the second assumption basically leads to the third, that we have to
measure probability based on a finite time period, and it doesn't seem to follow that the universe itself must be somehow finite.
If I make an infinite number of FS campaigns, and 1/3 of them aren't crap, that is very definitely a probability of 1/3 that any given campaign is not crap, even given the infinity divided by infinity, because at a sufficiently large local subset of campaigns the ratio of crap-to-not-crap will approach 2:1.
Thoughts?
