Problem of Induction - most briefly stated as "the past does not predict the future". For example, the proposition that "every time I log on to the web I get a fast connection", might be based on the fact that, up to now, the 5276 previous occasions on which you've logged on have all been good fast connections... Anyone still think that this guarantees that the next one will??!! (of course, in about 50 years hence - 2050 onwards - they'll look at that example and say 'yup - whats yer problem?' Well just remember, young whippersnappers, the first versions of this were written in the days when a good connection allowed us to upload or download stuff at perhaps 2k per second. Yes, thats right, 2 kilobytes of 8 bit data per second. And yes, you could go away and cook dinner, walk the dog and have a shower while a 10 Mb file was downloading!)

(of course, in about 500 years time, they'll read that bit about 'walking the dog' and think 'uhgh! primitive savage!') (and, in a couple of thousand, they'll absorb that bit about 'reading' and be amazed we managed to store as much data as we did!) and beyond that... ...who gives a shit?