predicting

to be able to predict on a longer time scale the corrugations of the shorter term, i.e. for eg. to predict what will happen in a particular week a year in advance, what will happen in a year a century in advance - effectively, to be able to mimic the uncertainties of a shorter time frame in a longer time frame;

what's the entire point if I feed in the uncertainties of the hour into the equation and it predicts an hour ahead, but can't the short term uncertainties be clubbed to give me a long term statistic, just that another dimension increases :-);

trend lines->inherent assumption that long term effect are an aggregation of short term effects, i.e. what remains after the positives cancel the negatives gives the trend, but in most cases isn't the long term cause totally different;

fractal, an interesting step ahead, but why the self similar replication, inherent in nature or just that when you self replicate to levels so many, the differences ought to come down, that's a trick rather than a solution;

what then is the solution...let's see...atleast something intellectually challenging to work on :-)