US Government has Crap Systems too
Posted on August 21st, 2006 by Tim Almond
Source: FBI Upgrade that Wasn’t
Looks like the FBI has trashed a $170 million computer system. What I don’t understand from the article is that they’re saying that it’s going to have to be started again.
I’ve worked on some projects that have, during user testing, been found to have a serious design flaw, but none have then been completely trashed. The system has been reworked and made to fit the correct requirements, or the problem has been worked around/lived with.
Still, it’s good to know that we’re not the only people with crap, overpriced computer systems.
RSS
I’m a bit surprised at how pathetic the FBI systems are, but not very surprised at how the upgrade has gone. I’ve seen a lot of government IT decisions plagued by very low IT understanding among decisionmakers, especially when the project is very big. They often get small projects done well; when something is big and high-profile, however, I guess too many low-knowledge high-ego bigshots try to muscle in and be The Decisionmaker.
I think SAIC may have it backward when saying:
Executive Vice President Arnold Punaro… FBI officials, he said, took a "trial and error, ‘we will know it when we see it’ approach to development." Punaro said the company warned bureau officials that such a method would not work
I think that the "waterfall" design betrays the government again and again. That design is exceptionally bad when the people drawing up the initial specs have vague ideas about what to do and why, which is generally the case with government IT projects. A good iterative design process would have been a better bet, I think, but it requires thriving interaction between customer and contractor, and it sounds like SAIC didn’t try to make that happen.
It’s also funny how much the unfulfilled needs they describe appear to be met by any of a dozen open-source projects you or I could get installed in an afternoon. The article also describes the choice as having been between COTS (commercial off-the-shelf) and homemade SAIC code… tragically neglecting the option of leveraging mature open-source code, which is indeed a common government error.