Randomness is not usefull if it's public?
Are those publicly avaiable sources cabaple of producing random bits faster than the Bitcoin network?
Well... once its fixed, its no longer random. It is a definite number, and the same every time. It is only "random" to someone who is trying to guess what it is. Future hashes in the block are random, once they are published in the chain, they are fixed.
So I could see gambling or gaming scenarios where some random element depended on the next or some future block hash (or even the transaction hashes concatenated). Though, you would be limited in how fast you can use it, as some like to point out, block chains can rarely temporarily fork, leading to a potential change a few down the line, a possibility which goes down for any given block, with every block added.
Essentially, this is no different than illegal lottery games. My grandmother was telling me that when she was MUCH younger, she was a bookie (and still wont name names, even though they are all long since dead) and that is basically the game worked by people placing bets with her, then, the numbers were derived from numbers published in the newspaper.... so everyone could check and see if they won easily.
Thinking about Hal's comment, I don't think that is a real problem. He chooses nonces to make hashes. So yes, he can throw out any hash that he finds which is even or odd, but, it only matters if they are under the target. So a miner who was trying to cheat his bet would have to find a good block, giving him its tx fees and income, and then throw it all away, and keep searching.
That is for an even/odd bet though. Sure maybe with the right betting structure like that you can afford to mine and throw out half of the possible blocks, but... what if you implemented keno? I think putting together a betting scheme that would not be worth cheating in these circumstances could be done easily.