>>14417453first of all, when predicting shit we're talking probability, not statistics. statistics are posterior (aka, analysing data after it happened). probability is the good stuff you're after. that's what maths illiterate people often confuse (not really talking about you, OP, but for example media). We tend to use statistical models to predict the "future" events to a degree of certainty. But that can mean jack shit. that's where random events (variables) come into play. We can watch the past as long as we can, the "future doesn't give a shit". for our limited scope, it's just random. and once you consider that factor, you'd be arrogant to claim you can predict non-trivial (ex. next morning the sun will rise) future events even slightly.
sorry for the long "intro", now back to the good stuff - it's not though. ""statistically"" we're most likely to be halfway. funny thing is. whenever we look we're most likely to be halfway.
but, it only works if you have limited a priori knowledge (otherwise we'd have to apply the bayes/laplace rule).
if we pretend we only know that humans lived for 250.000 years, the most likely outcome would be that they'll live another 250.000 years. of course, in real life we can make some "educated" guesses and could/would have to apply Bayes' rule. I don't even dare to think of such a model
a fun reading about that topic is chapter 6 : Bayes' rule from the book "Algorithms to live by" by Griffiths and Christian.
this is just from a "scientific" way of thinking
