Reading the pdf makes me appreciate Haskell again. It seems really complicated whereas using STM in Haskell is simple. One paper by Simon Peyton-Jones explains STM much better than this large user guide.