Erlang is largely a different kind of niche than Haskell and lisp. Its more focused on distributed and reliable processing than more general purpose. In effect its the Actor model taken to the extreme. Where I'd categorize Haskell and lisp more general purpose but they take different approach. Lisp the more "keep it simple" approach and Haskell the high theory approach. I've yet to encounter someone that claims expertise in haskell that isn't obviously lying or delusional as it is probably the largest language feature wise. I've been learning it for years but still have yet to find the "target use case" that it is the best tool for the job. It certainly is worth learning about even if only for the new perspectives on problems it encourages.
At least in my opinion:
Lisp is amazing for its simplicity and homoiconicness and the great powers that come with those.
Erlang is amazing for its approach to distributed computation and reliability.
Haskell is amazing for at least its theory, and probably more I'm not yet aware of.
At least in my opinion: Lisp is amazing for its simplicity and homoiconicness and the great powers that come with those. Erlang is amazing for its approach to distributed computation and reliability. Haskell is amazing for at least its theory, and probably more I'm not yet aware of.