Wow. The argument in the original post that is linked to about scalability and functional programming is almost exactly backwards. Resource usage here is a sill

Wow. The argument in the original post that is linked to about scalability and functional programming is almost exactly backwards. Resource usage here is a silly red herring. Keeping code free of side effects means you _can_ scale it. Try scaling out to multiple machines when you have a fundamental bottleneck like serialized state in your code. All data-parallel programming techniques, dating back from at least 1985 with the connection machine lisp, and whose modern friends include MapReduce, come from unshackling ourselves from too much state. These benefits come from making software more functional, not less functional.