Recently the IT Project Failures blog reported on a billion dollar project failure at the US Census Bureau. The idea was that the 2010 census would be taken with handheld devices rather than paper forms, and Harris Corp was awarded a $600 million contract to design the devices. The project is now $2 billion over budget, and the entire thing may have to be scrapped because it wasn’t built to GAO standards.
Obviously on a failure of this size, there are any number of areas where one can find fault. To me, the fundamental problem was starting such a massive project in the first place. According to the FDCA Web site, the contract was awarded on March 31, 2006. The RFP was released in late 2005, and planning for the project started in 2004.
I have no idea how you plan six years in advance for an IT project. The landscape is certain to change so much over that time that whatever you put in the RFP is likely to be obsolete by the time the project goes live, and who wants to go live with an application in 2010 that feels like it was built in 2004? Agile methodologies were a backlash against exactly this sort of project. When you read about failures like the one at the US Census Bureau, it’s easy to see why.