Maybe I misunderstood, but it seems to me that without set-car!, Scheme becomes more like Erlang in respect to "unchangeable variables" (don't remember the proper term right now).
I had recently evaluated Erlang but found some things very difficult to achieve. For a contrieved problem, how do you program an effective algorithm to calculate prime numbers in Erlang?
So I felt quite good about the destructive operators of Scheme. If they are now wanted anymore, I might reconsider using Erlang.
My understanding of the situation is that the PLT people had some optimizations they could do if lists were immutable, and they noticed that there were very few places in their code where they mutated lists. So they decided to just make lists immutable and see who complained. And it turns out few other people were actually mutating lists, so they kept the optimization.
PLT Scheme still has other mutable data structures that can be used instead of lists, though; I don't know if this is the case for Erlang.
