Over the years I have read a large number of pieces advocating the use of sensible defaults. Stephen Anderson's book Seductive Interaction Design devotes two pages to the topic with proof about the why and how (one of many reasons to order a copy as soon as you can).

Today Jared Spool provides even more validation. By performing a simple experiment with Microsoft Word he found that less than 5% of users change any settings. In other words 95% of people touching your product will assume there is a reason that the default settings are configured the way they are. Rather than fix them they'll go with the status quo.

I highly recommend reading the piece when you get a minute. The questions it asks apply not just to those of us trying to design the experience but the people actually writing the code. You may be surprised at all the assumptions that sneak through into the final design.