People are still talking about the recent news that the FBI's Virtual Case File system won't work after almost $170 million has been spent on it. A Wired News story has generalized the commentary to talk about government IT blunders. Your first thought might be that things aren't so rosy in the private sector either. There are three things, however, that make government IT projects even more susceptible to failure that the one in your company.
The first is scale. Even relatively small governments, like Utah where I was CIO for two years, have disproportionately large needs. For example, Utah processes over $1 billion dollars in welfare claims per year. The system that does that processing is currently being rebuilt. When it's done, the final price tag will probably be close to $100 million. I'm not a fan of big monolithic projects, but even split up into "manageable" pieces, the first phase of this project was $35 million.
The second issue is specialization. In the Wired article, Nancy Harvey says "Robert Mueller [can't] find an off-the-shelf product called 'Find Terrorist.'" While I sympathize with this, its not entirely hopeless. Most government projects can be built on top of commercial frameworks and use commercial tools. Too many government projects start lower on the food chain than they should.
The final issue is public exposure. Look at the beating that the FBI's gotten over the decision to cancel VCF. When I first became Utah's CIO, one of my staff counseled me that I could never cancel a project. The Legislature, the press, and the public would not understand. Even public companies can cancel all but the largest projects without anyone taking much notice. It takes guts to cancel a project anywhere, but in government its usually a career altering event. Consequently, projects get milked along until they just have to die and the result is more wasted money.