Assumptions can be very helpful as long as they are explicit and periodically tested against reality. One commonly untested assumption is that a new system stands on its own. In most cases it is connected in complicated ways to lots of other things. Another problematic assumption is that a system is readily usable. Again, seldom true—especially when it will be used in different locations by different professionals. A third risky assumption is that data integration is a technical problem. It is, but it’s also an organizational, political, intellectual, and managerial one.