************************************************

Excerpts from the reviews recommending rejection

************************************************

"The paper is very well written, thought-provoking, and

addresses a fundamental issue in AI: how to flexibly represent and

reason with knowledge at different temporal grain sizes. I greatly

enjoyed reading this paper, and would very much like to see it

published."

"The topic addressed in the paper is extremely important and of interest to

both decision-theoretic planning and reinforcement learning communities.

I think the paper could potentially serve as a basic reference

to this topic in future."

"Hierarchical options:  Why have the authors avoided this?"

"Options" must be clearly, succinctly, and unambiguously defined."

"...the section that defines Q-functions does not even mention Watkins

...and doesn't even call them Q-functions."