>But the saddest thing is that the failure rate of large (or huge) IT projects still hovers around the 75% mark. Vendors walk in and sell the powers-that-be a bill of goods and then the IT department is stuck with making it work and the end users suffer.

It's important to break those numbers down.

Yes, there's a high failure rate of many projects. But it's important to understand what constitutes failure. A system initially targeted for implementation on January 1 might not get into testing until 2-3 months later and might not go into production until late spring. By many calculations, that's failure. Maybe the initial goals were too optimistic - or maybe there was serious feature creep - or maybe the project management wasn't very good - or maybe the development team had a high bug rate - there are all sorts of reasons. (and yes, sometimes a third party sells in a tool that isn't properly vetted by the IT people, and they're stuck trying to make something work. I'm actually dealing with one of those right now)

I'd love to see the breakdown of the failure rate into how many projects are scrapped or truly die, versus how many are simply late, how many evolve into other projects, etc.

Quoting generalizations as accepted science leads to fallacies. It's important to start some decomposition of what the numbers mean.