I can't understand why you're being so bloody minded about this. You say, "The original information that we want to store, exchange, and process with computers doesn't come in sizes of integral numbers of bits."
No kidding! You're not talking to a novice. I've been involved with computers (very intimately) for just over 45 years, as a programmer, systems analyst, systems programmer, systems engineer, product marketing manager, director of systems and programming, consultant, etc., etc., etc. I'm also a Fellow of the British Computer Society (since 1969). I was writing code in the late 1950s that packed and unpacked data that didn't fit neatly into the computer's own arbitrary structure (10-decimal-digit words on the IBM 650 on which I originally worked). So I do know these things, you know. "Most of the data on my 120 GB (120 * 10^9 byte) hard drive is compressed." Of course the data is compressed. However, the device in (or on) which that compressed data is stored is structured as 8-bit bytes. Like anything else in a computer, a compressed file is a string of bits. The computer stores that string of bits, 8 at a time (I'm ignoring parity bits and error correction codes for the sake of simplicity), in the bytes into which its memory is structured. The fact that your 120 GB drive holds what was originally much more is, again, irrelevant. (Actually, overhead considerations such as pointers, incompletely filled blocks at the ends of files, file fragmentation, etc. can reduce that considerably.) Yet again, I was talking of a byte of storage, not a byte of original data. However, compression simply means that a given number of bytes of uncompressed data will, after compression, occupy considerably fewer bytes of physical memory. The fact that the compressed version straddles physical bytes is irrelevant to the discussion we were having about physical memory size and binary prefixes. Let me repeat three of those words: PHYSICAL MEMORY SIZE -- the original topic in this discussion. "Don't feel so bad, we all make mistakes. It's no big deal." Let's just say we have a semantic disagreement, John -- and please don't be so damn patronizing. Bill Potts, FBCS, CMS Roseville, CA http://metric1.org [SI Navigator]
