Beyond Public Announcement Logic: An Alternative Approach to Some AI Puzzles