Arithmetic compression represents the current dominant form of statistical compressor for most common data types: JPG, MPEG, LZMA, BZIP, WEBM, WEBP etc. It's wide range of use is dominantly because of it's un-rivaled ability to encode data as close to entropy as possible, without bloat. But what's not understood by most developers, is _how_ Arithmetic Compression actually achieves this result. Most public descriptions are mired in technical details of the transform, but don't actually describe the statistical reasoning for the compression savings. Well, Fear not, because +Colt McAnlis walks through the core concept, and how all the crazy math produces one amazing compression algorithm in this episode of Compressor Head.