Some #Noestimates advocates are heard quoting the Standish Report, reporting the non-data date that 60% of IT projects fail and of course that NOT ESTIMATING would have somehow, by some means, through some magical force has saved these projects.
And like the return of the swamp monster, this long-debunked notion came back and needs to be put to bed again.
The numbers in the Standish report are still being used by "agile thought leaders" to establish the basis for their proffered solutions. Forgetting even more that "agile" is a bottom-up software development solution. And that most of the issues with large IT programs are leadership at the senior executive level, externalities from the market place at large, and general inability to connect the dots between business strategy and business execution.
And of late by a #NoEstimates thought leaders to further their unsubstantiated conjecture that decisions can be made while spending other people's money in the presence of uncertainty - violating the principles of Microeconomics of Software Development, Managerial Finances, and Probabilistic Decision making - all topics well established in the literature, higher education, and tools
So let's collect all the debunking materials to put an end to this nonsense
- Standish Report and Naive Statistics
- The Rise and Fall of the Standish Report Report Figures
- Finally, A Challange to the Standish Report
- Project Failure Rates
- The Non-Existent Software Crisis: Debunking the Chaos Report
- The Chaos Report Myth Busters
- Standish Numbers
- CHAOS debunked and Standish Discredited?
- How Large are Software Cost Overruns? Critical Comments on the Standish Group's CHAOS Reports
- Software Cost Overruns - How Large Are They and How Should They Be Measured? a Critique of the Standish Group
- Standish, the CHAOS Report, and Science
So next time you hear some IT project failure rate number, or how bad estimates are, you'll first ask where'd you get those numbers? No answer, run away, the speaker is just spouting anecdotes. And
Anecdotes are samples of One from an unknown population
And that's the same problem with the Standish Report, it's sampled from an unknown population and wouldn't pass the smell test of a college statistical sampling exercise.
