In Johanna Rothman's post it was suggested that predictions about cost and schedule are almost guaranteed to be wrong. Here's some guidance on why you should ignore that advice if there is not a statement about "doing the wrong thing will not usually produce the right results." In other words "bad project management is not the excuse for failing to understand it is bad project management."
One school of thought says we cannot manage what we cannot measure. Of course this is not true, since we manage things all the time that can't be measured - product innovation, manuscript development are managed without direct metrics. At best they could be managed with inadequate metrics. In the intellectual and creative domains, metrics do not provide insights needed to forecast the outcome.
What's missing from this approach to measurement is the answers to questions like:
- How good are your processes?
- What quality can we expect from our efforts?
- When will we be finished with this project?
- How much will it cost when it is finished?
What happen is project managers measure intellectual processes - like software development - with qualitative metrics rather than quantitative metrics.
Monitoring activities using metrics to evaluate broad issues like progress to plan are usually counterproductive. Measuring specific progress against identified quantitative constraints to determine the likelihood of meeting that constraint is a realistic process. Measures like defect rates, sufficiency of assigned resources, whether there is enough time and money to complete the project - the Earned Value Measurements - for the basis of the Program Controls function for any non-trivial project.
Software Sizing, estimation, and Risk Management: When Performance is Measured Performance Improves, Daniel D. Galorath and Michael W. Evans, Auerbach Publications.
Some well know thinkers have something to say about measurement and especially performance measurement:
- Apparently - all things being equal - it is better to measure than not to measure - Winston Churchill
- Quantities are measurement of qualities - Paul Kirchner
- Now, things do not, in general, run around with their measures stamped on them like the capacity of a freight-car: it requires a certain amount of investigation to discover with their measures are - Norbert Wiener 1920
So in the end it is very simple
No Point Estimate can be correct in the absence of statistical variances.
Suggesting that predictions about cost and schedule are almost guaranteed to be wrong without stating that those estimates are missing the underlying statistical model is both right and wrong at the same time.
Never provide point estimates without the statistical variances. To do so puts your in the same position as Johanna - your right, but wrong.
