The Standish and similar reports paint a bleak picture of project success. Ignoring for the moment the validity of these reports (see Finally a Challenge to the Standish Report, The Standish Report, and Standish Report and Naive Statistics).
Here's a sample of the problem classifications from an academic study of DoD Programs. http://pmchallenge.gsfc.nasa.gov/docs/2010/Presentations/Romeo.Mitchell.pdf
We can see from this study that poor estimating and scheduling is source of problems. Why this is so is another problem. But number 2 is inadequate staleholder management. Then the causes go down from there.
The discussion around what Done looks like is off track. There only definition of done is contained in the Statement of Work (SOW), the Statement of Objectives (SOO), the Concept of Operations (ConOps) and the Integrated Master Plan (IMP) and the connection of the Technical Performance Measures (TPMs) with these three documents.
Attempts to define a universal description independent of the description of "done," is pretty much a waste of time. There are no units of measure outside those defined in these documents, and the performance parameters applicable to the program management processes - cost and schedule variance parameters.
When Standish says "on budget" and "on schedule," they don't say what the acceptable variance is. This of course is meaningless. No variance meassures, means the is Failure for any variance other than Zero (0). Doesn't sound like a rational approach to successful project managment.