Brains working towards machine consciousness

OK, so this is very silly - I just threw together a little test program (to generate data which might fool a LSA based tool I am planning) which produces a stream of text based on the probabilities of n-tuples drawn from a corpus of multi-authored documents (in this case, various academic papers I have on my hard drive).  This example output is on the basis of 3-tuples (that is, given a 2-tuple, it works on the probabilities of which word follows those two words in the source documents), drawn from a random sample of 40 documents from a corpus of 8,900 (some of which are in German, but

I tend to believe nothing is impossible.  Well, I say believe but it is more 'hope', really.  Richard H., colleague, friend and general loon, pointed out to me the other day that:

If nothing is impossible then
there is a finite possibility that something has a probability of 0, i.e. that something is impossible. 

Bother.  The fact that something can be impossible means that it is not true that nothing is impossible.