It makes it trivial to write DSLs, for example, since you all you need to do is model the syntax tree as a series of Lisp data structures, and then evaluate them directly.

…Since it was designed at a time when there was One Processor to Rule them All, it didn’t go to great lengths to consider how its design might effect the needs of parallelism.

…Even if the arguments could have been computed in parallel, there’s no way to know for sure that the evaluation of one argument doesn’t cause a side-effect which might interfere with another argument’s evaluation.

…If then I do something in my function which needs some of those values, Haskell can start computing the ones it needs in parallel, waiting on the completion of the whole set before returning the final result.