In some exciting news, the New York Times reports that work done by Microsoft Researcher Eric Horvitz may be working its way into Outlook, among other products. The work would use machine learning algorithms to examine a user’s behavior and the emails they receive “to suggest whether a user wants to read each message that comes in.” This could be a fascinating improvement to the most popular desktop client in the world.
Horvitz has produced an amazing body of work since starting at Microsoft in the 90s, and a good deal of it has dealt directly with information and email overload. For example, Iqbal and Horvitz’s 2007 article “Disruption and recovery of computing tasks: field study, analysis, and directions” is a great example of determining real-world costs to task switching.
Adding intelligence to email clients, especially in the form of identifying user behavior and trying to determine message value, would be a huge step forward. Also important is that this would be implemented in the client that many middle- and upper-managers are using—the people who experience the most overload.
Now of course we don’t have anything more specific yet, but this news is very exciting to those of us interested in email overload, and sounds very positive. Great to see Microsoft continuing to innovate in this sphere, and hopefully bringing some of the excellent ideas of the future into reality.