The Butler concept has had a solid year or so of discussion, and has produced a relatively good high-level plan with some outstanding implementation questions that would impact the features planned for its "generation 2" or "generation 3" features. While hashing out those things is good, with the release of Drupal 7 it's worth circling back around and assessing what we'd like to target for reasonable inclusion in Drupal 7.
Based on the previous discussions, the following phases look promising for initial implementation in the D7/D8 transitional period, with an eye towards a D7 contrib release and a D8 core patch:
Phase 1: Context and Phase 2: Response controllers both look promising, though there are some implementation details to be nailed down around the responsibilities of different context management subsystems. Phase 3: Clean up hacky systems is really about specific portions of Drupal that could be ported to use the new system from Phases 1 and 2.
Phases 4 and 5, while useful from an overall architectural perspective, are also much more involved -- they entail gutting large parts of Drupal's theme system and so on. I'd like to suggest putting those off until the next release cycle -- Drupal 9, or Drupal 8 contrib. While that's a long ways away, I think focusing on a tighter goal will pay off in terms of core reception and fewer ulcers.
Thoughts? Anyone? Bueller?

Comments
Core Conversations
I was drafted to give a Core Conversation on this topic at DrupalCon Chicago: http://chicago2011.drupal.org/coreconv/butler-project-phase-1
Well, fantastic!
Sounds like I jumped the gun. ;-) I'll be eagerly anticipating it.
Drupal 8
Sounds great.
Since Drupal 8 development is going to start soon, think we need to be wake (this time ;) ) and promote idea to implement butler to core.