>Is hashing a technique that tries to make an unique value out of a file?

Yes


>Can I calculate the average hex value of the bytes in a file and call it the files hash value?

No (or at least I really dont think so)


>Does Metadata screw up the hash?

Different Metadata will result in different hash values (but then they are different files)


>Can Evil Forces fool the hash system? Like, can I make a (copy of a) file that has the same name, filesize and hash as the original, in order to f*** with those who want to download the file?

You could lie and say that you have a file with hash value X but when I dl and calculate the value I can see that you lied. Of course I have to have the full file if the hash value is from a full file hash. If I got different parts of the file from different sources I couldnt tell which one sent me the bad data. This is solved with the Partial File Sharing Protocol (see my post in this forum)