Just a thought in response to this otherwise good Lars Christensen post.
There was a time when "discretionary" was the word we used to describe counter-cyclical policy. But I think after John Taylor that old distinction became a lot harder to justify. Rules is rules. Then with Lars Svensson's work on forecast targeting people started to realize that even old-school, non-counter-cyclical rules started to look a lot like the stuff they used to call discretionary. Nowadays it seems like "discretionary" is also used to talk about any unconventional mechanism even if it's used for the purposes of a very conventional policy rule.
So I don't personally see much value in the rules/discretion distinction except as a nice way to distinguish yourself from you opponents.