This is a brief account of the discussion on the future of Haskell at the Haskell workshop, Oct 3, 2002, in Pittsburgh. After Simon Peyton Jones discussed the copy-right issue of publishing the report, we had a brief discussion about the future of Haskell. The first point raised was that the addition of unsafe extensions to Haskell, like unsafePerformIO and unsafeCoerce, goes against the original design requirements for Haskell. The presence of unsafe extensions makes quick but dirty solutions possible, for example for problems that need concurrency. This hampers the development of semantically clean theories, and makes building tools that reason about Haskell much harder. Reactions on this point differed: some people heartily agreed, some other people thought unsafePerformIO was often just a cosmetic thing, and that whenever a theory/abstraction/method for an application for which unsafe extensions are used now is in sight, people will try to develop it. Another suggestion was to develop methods for separation of threads so that you can reason locally about threads. -- Johan Jeuring