Beautiful timing - I was just reading Land of Lisp for fun, and I use emacs! Questions: Why do you need SLIME instead of just writing into a buffer and evaluating it? Is SLIME just for REPL-ing? How do you organize a "bigger" project?
Since the interactive development environment involves an SBCL process running the Swank TCP server which SLIME running in Emacs talks to, I believe SLIME (or something like it) is still required. For example, when we press C-x C-e, even if we are not using the SLIME REPL buffer, we still need SLIME to send the Lisp expression to Swank which would then evaluate it within the SBCL process and send the result back to SLIME, so that Emacs can show the result in its minibuffer.