It seems sensible if you're doing something with successive streams of data where working with sub collections as copies would normally cons up extra cruft. There are some reference optimizations that can be taken when you know a collection can't be mutated and some concurrency guarantees can be made.
I can appreciate his balanced view. I think it's of the utmost importance to keep Lisp a multi paradigm tool for the user. It's a testament to the language design that you can have your cake and eat it too. I'm all for language bending in Lisp as long as the culture stays one of open-mindedness.
The whole point of Lisp is that you get to be the language programmer. People should be able to dabble freely and go down any avenue that seems particularly interesting. If you try to shoe horn a Lisp into being single paradigm you throw away a long tradition of letting the programmer know best. So, I suppose I can say I'm just happy to see this kind of experimentation.
-Ryan