>>19
There is nothing wrong with the current definitions. If a system was developed which, say, had 16 bit bytes they would become useful. Just now they are essentially synonymous.
It isn't nessecarily wrong to say a file is x bytes either, as long as you know the number of bits in a byte on the architecture the file was created.
Consider our hypothetical architecture again. To get octets of a file created on our current architectures you would probably do:
char ch = getchar();
octet1 = (ch >> 0) & 0xFF
octet2 = (ch >> 8) & 0xFF
Yes, I know this system does not exist.
I've never heard kibi, mebi, or gibi as prefixes before. They are indeed retarded.