The USS Quad Damage

Clearly, this is a rhetorical question. I am referring, of course, to myself. I have just successfully built an arithmetic coder. It's like zip, only it's slower and compresses worse. On top of that, it's a stream based coder, so a single bit error will screw up the entire rest of the stream. If that weren't enough, it also doesn't know the exact length of the file, so it'll end up with some cruft at the end of the regenerated file.

However, I built it, so it must be good. Actually, I'm surprised how well it works. With 16 bit contextual coding on text (not a good idea), it compresses almost as well as gzip (bzip2 blows it out of the water, but let's not worry about that). It doesn't do well on jpegs, but then, neither does gzip. I suspect that with good contextual coding, it will be quite useful to me in the future.