My approach to research emphasizes careful, hypothesis directed
evaluation and recognizing applications that are well suited to
solution by AI techniques, especially planning and search. It is
critical to not just show that an
algorithm can perform better than a predecessor, but to
understand why it does and so motivate the next
design. This approach has led me to explore applications as diverse
as scheduling contacts with satellites for the Air Force to most
recently designing a security agent that is tailored to the
psychology of the home computer user.
