Micro$oft research article ???
Having read the entire article by Simon Peyton Jones, I remain a bit baffled. This is definitely the first time I see a Microsoft guy breaking a lance for something even remotely as advanced and out-of-MS's-league as Haskell.
Being a Java programmer myself, I donot, however, consider that implementing your concurrent software in Haskell *really* yields higher-grade software, as
1) instead of fully analyzing your problem and modelling locks, transactions and threads etc., this article encourages you to rely upon the correctness of a Haskell implementation of transaction handling. People like e.g. DBA's can do that, they arenot primarily concerned with the innards of software but with using software to handle their data. Programmers cannot afford to do that, as they actually *create* such software
2) being able to fully analyze your problem and model it into locks, lock handling mechanims, transactions and threads etc. is a skill any good non-greenhorn programmer must have. He/she should *at least* be able to reason about such things. Relying upon a Haskell implementation blunts his/her sharpness.
3)Java provides you with all the necessary tools to satisfy both 1) and 2)