It's the economy, stupid - James Carville
This phrase is the basis of our failure when trying to fix dysfunctional conditions in any software technical domain. It's the basis of the principle of finding the root cause and stop treating the symptoms. If any proposed solution does not identify the root cause, the condition, and the action that produces the undesired effect, that proposed solution is at best treating the symptom, which will re-occur and at worst be ineffective at fixing anything.
So when you hear things like estimates are the smell of dysfunction, estimating provides no value, the cone of uncertainty does not decrease as the project progresses. While there is a thread of truth in each of these, the Root Cause supporting the claims are not presented. In addition, there is a body of response showing the statistical assessment, managerial principles, and other foundational processes have been violated in making the claim.
My most recent encounter is with the Cone of Uncertainty and the claim that it does not reduce as the project progresses. That is the estimating quality does not get better. For the reported data from one of the authors, this may actually be the case.
But the answer to Why it does not get better is not provided
Why does the estimating accuracy decrease, increase, or remain the same as the project progresses? This question is unanswered in
Same for estimates are the smell of dysfunction or any other conjecture about estimates
The answer to Why is not provided, so no corrective actions can be taken to remove the dysfunction, if it were to actually occur
This is a common failure mode for conjectures without root causes. If you see some effect, ask first - why am I seeing this. To not ask and answer this means you're just recording observations of some symptom but have yet to find the cause. Without the cause, you will continue to be subjected to the undesirable outcome of the symptom.
Here's a framework for performing Root Cause Analysis in a way that actually produces the Root Cause - not just the Symptom - so the Corrective Actions can be defined and applied.
Some sources of response to conjectures about estimating accuracy and precision can be found in
- Cone of Uncertainty Controversy
- Coping with the Cone of Uncertainty: An Empirical Study of the SAIV Process Model
