One quote used as a Platitude for the #NoEstimates discussion is...
This is from the Art of War. The problem of course is Lao Tzu had no awareness of probability and statistics, and the Box Jenkins Autoregressive Moving Average algorithm for forecasting future behaviour from past behavior.
Lao Tzu had no comprehension that all variances in any natural process - be it human warfare, flying to Chicago from Denver, spending money on the development of a software product - are random variables, drawn from an underlying statistical generating function for the irreducible uncertainties, or driven by a external probabilistic process for the event based uncertainties.
Lao Tzu was speaking about warfare processes against enemy's without real time Reconnaissance, Surveillance, and Target Acquisition (RSTA) capabilities. Without the benefit for the discovery of probability and statistics in the 16th century all the way to the Box-Jenkins algorithm in the 50's to our current non-stationary stochastic modeling processes used today in nearly every human endeavour.
This RSTA information, Bayesian models of maneuver warfare, were unimaginable to Tzu. In our modern warfare system, asymmetric warfare has caught the planners off guard, but Herodotus advised us on how to learn from the past in the current theaters we are failing in. Same for making forecasts of cost, schedule, and technical performance. Several #NoEstimates contributors are already doing this. They may not recognize it, but recording how many stories have been produced and making a histogram is the basis of forecasting the future, with statistical information from the past.
The Bottom Line is Simple
If you accept Tzu, then you're never learning from the past, assessing the engagement models, formulating possible responses, and testing those responses against your current knowledge and creating corrective actions - from the models - for future decision making.
A piece of advise from a senior leader of manned space flight systems may guide your application of estimating and forecasting
We don't make predictions because we want them to come true, but because we want them not to come true.
So if you are unable or even unwilling to develop an estimate of the total cost - total commitment for those owning the money - the total duration to deliver the needed capabilities to the customer paying for those deliverables, then there is a likelihood the project will fail.
The primary reason for this failure is...
You don't know what DONE looks like, other than we spent all the time and money.
No credible project management process, from eliciting the needed capabilities, to writing code for money that implements those capabilities, can be present in the absence of some form of estimating.
Take care following 500 BCE advice in the absence of modern scientific methods of decision making
If you do so, acknowledge your ignorance of those modern methods and your explicit refusal to consider them in your decision making processes.
To end with my own pithy quote...
Whereof what’s past is prologue, what to come In yours and my discharge. - William Shakespeare, The Tempest.
So if you don't seek to include the performance of the past, you'll never see the future coming. To use the performance from the past to forecast, estimate, make a prediction, a statistically sound prediction, you need to ignore Tzu's advice and get into the 21st century.
If you're looking for ways to do this. Like one #NE poster suggests, read here for actionable outcomes.
- Managing Projects in the Presence of Uncertainty
- Cost Estimating
- Credible Estimating Processes
- Probability versus Statistics
- Replacing Risk for Knowledge
- The Fallacy of Planning
