Having briefly worked on the Child Support Agency project that went horribly my impression is that public sector IT projects are entered into for all the wrong reasons and with none of the incentives the private sector has to make things work.
Private companies generally buy IT services either to get a competitive edge or to reduce their cost base. Projects do frequently go wrong, but because the targets are usually pretty well defined this can be caught early and (with some luck) fixed.
In the public sector a competitive edge isn't really the issue and IT projects appear to be designed explicitly to avoid making anyone's job redundant. Whenever a project looks like it may be in danger of doing this the civil service change the specs and the contractors, paid on a per diem basis, happily take the extra work.
When someone eventually discovers that a project is giantly over time and budget (which, frequently, noone does, because noone is relying on it to deliver products or services the way a business would) there seems to be an agreement between the big contractors and the government. If the contractor takes the blame for things going wrong and the bad press coverage without unveiling the real reasons the project went wrong (or the fact that most of the public sector don't want the projects to go right) then the government will continue to hire that contractor no matter how badly the public think they've screwed up.
EDS understand this game best, which is why they win so much government business. As for the consultants on the ground it made us even more cynical than everyone else about paying taxes to fund the kinds of projects we now know to be incubation for groups of people sworn to protect their jobs and their budgets against modern technology.