This popped up on the XP mailing list last week. I don't think I could agree more.

My experience is that the single greatest risk in a software development project is building the wrong software. The money saved by reducing costs is miniscule compared to how much you are risking by building the wrong system (or even building a few features nobody needs). No matter how many SMEs and proxy customers you involve, and no matter how many prototypes and models you build, there is no better way to determine if you are building the right software for your market than the earliest possible release of working software to real end users.

Steven Gordon

Too often is seems that software teams introduce tools and extra processes to try and make sure they're building the right software before they start. How long will it be until we learn that this strategy doesn't work cost effectively?