It is possible that the futures tree of all possible futures can be predicted by a sufficently sophisticated processor. Events can be accurately guessed, given enough data, the allowance for free will and the fact that people are almost as predictable as any other animal.
There is however, only one true future, on which I aim to find in my predictions. I can usually find the branch on which it lies on the tree. Each event is like a leaf that unfolds on the branch.
What is needed is a better analogy, maybe even a bad one?
Perhaps I should state this in the way I dreamt of it:
The universe is but one in a multiverse of parallel dimensions, we exist in a small pool of reality defined by the time that has past us by and the time yet to arrive in the future.
The main point is, we can only be uncertain of which particular pool the future will arrive from, to become our present depends on the decisions made by the inhabitants of the current reality. However, it is possible to put in a model the weights in the neural networks surrounding each decision according to what is probable.
In other words, futures can be weighed, assessed and prepared for if there is enough processing power and memory available. There is also no practical reason that I can see why people can't do this in their head.
Perhaps this could explain why some people can dream about the future, only to have it happen later and feel a sense of deja-vu.
Credit: I have developed some math for this, but it does require somewhere around 1.6 to 3.2 GHz of processing power with a 1 GB workspace. Works best on multiprocessors.