Philip E. Agre's Paper – Lessons Learned in Trying to Reform AI (pages.gseis.ucla.edu)