I usually find Robert X. Cringely kind of ludicrous - although the most enjoyable technology columnists are all about the ludicrous - but he had a good recent dump on IBM's recent woes. The whole article is pretty good - but he nails it with the closing paragraph:
IBM project management is not based on business results. It is based on documented deniability. A successful IBM project is completing everything as originally documented. If it works or not, it doesn't matter. The folks in the GTC will be quick to tell you eESM is a great success and it exceeded all the requirements. Sadly, its users needed something better and a lot less expensive. In most businesses, this project would be considered a failure. In IBM, it isn't. And the organization keeps spiraling towards the ground.