While this is very interesting, maybe the fact that very few software projects are failing means that we’re being unreasonably risk-averse in software project management. A good software project can produce a return of hundreds or thousands of times what was initially invested. So we should be willing, as a society, to invest in many software projects that are almost certain to fail, just to get the benefits of the few that don’t.
I know, that sounds a lot like startupland (hi other site!) but startupland is still a tiny minority of software development; bespoke in-house projects are still a huge chunk, although surely less than a decade or two ago. And there’s lots of software that produces huge value but doesn’t have a reasonable way to capture that value with any confidence.