One of the escape clauses of #Noestimates is to re-label Forecasting as NOT Estimating, It is forecasting, based on empirical data. Ignoring for the moment the empirical data discussion since ALL data is empirical, otherwise you wouldn't have the data.
Empirical is defined as based on, concerned with, or verifiable by observation or experience rather than theory or pure logic.
And just to beat this horse some more
The origin of the term Empirical starts in the 1560s, from L. empiricus, from Gk. empeirikos "experienced," from empeiria "experience," from empeiros "skilled," from en- "in" + peira "trial,experiment." Originally a school of ancient physicians who based their practice on experience rather than theory.
Empirical also means observed, factual, experimental experiential, pragmatic, speculative, provisional. This is in comparison for estimating purpose for a theoretical model that produces data, parametric from empirical or theoretical models that produce data. In all cases the data is used to estimate some outcome in the past, present, or future. When that estimate is about the future, it can also be referred to as a Forecast. Weather forecasts, sales forecasts, market forecasts, earnings forecasts. All are estimates about some outcome. The antonym of empirical is theoretical, un-observed (as would be the case for a model), hypothetical, conjectural.
So just the repeat - empirical data is observed and empirical data can be one of the sources of making estimates about the past, present, or future.
Since Scrum is an empirical process, it is also shares its attributes with empirical process control systems. Project management and the management of work is a process control systems. This work - writing software for money - is a process. And as a process it needs to be controlled. The control makes use of empirical data. Empirical Process control systems have three major attributes. [1]
- Visibility - the attributes of the process that affect the outcomes are visible and known to the processes or people involved in the control of the process. You can't have a process control system, if you can't see what is being controlled, what the result of the control inputs and outputs are.
- Inspection - the various aspects of the process are inspected - sampled - frequently enough so that the variances can be measured frequently enough to keep the desired process under control. Yes, Scrum is a control system for producing the desired software at the desired time.
- Adaption - makes use of the data received from the Inspection to adjust the process under control within the desired range of outcomes.
This type of control is found from your home thermostat to the dynamic flight controller used to autonomously rendezvous dock the a vehicle with the International Space Station, to provide corrective actions to keep a project on track toward a planned finish date. It's all the same principle.
So Back To Estimating and Forecasting
No matter how many times the #Noestimates advocates make unsubstantiated claims that Forecasting is not estimating, it's not true.
Estimating can be based on empirical data or theoretical models, or better yet from my accelerator days, theoretical models informed by empirical data - Forecasting are estimates of outcomes in the future.
But forecasts are estimates,period. Anyone claiming otherwise needs to come up with reference materials from control systems books, financial modeling books, statistics books, and something to shows Forecasts are not Estimates about future outcomes. Even one of the founders of eXtreme Programming makes that cocla mammy claim that Forecasts are not estimates. Time to send them back to the High School math class.
[1] Modern Control Systems, Richard Dorf. This is the 5th edition. Mine is the 1st Edition, 1970 from the control systems course needed to write FORTRAN 77 code to control the sampling processes on the particle accelerator for our experiment. Google will find this book and Modern Control Systems, Dorf, in PDF form if you actually want to explore further. In the latter book here's the architecture of a general control system, which can be used to manage software development while spending other people's money, fly a spacecraft to Mars or you Boeing 737 you're riding on to home. Jeff Sutherland knows this from his days of driving around in an Air Force F-4 about the same time I was driving around in an Army CH-47 and his further development of Scrum from John Boyd's Air Force work of the OODA Loop. All project work, is all process control all the time. No way out of it, unless your conjectured method is open loop - then you're not controlling anything, you're just watching it fly into the ditch or crash into the ground. Not usually what your customer or commanding officer woould find desirable behaviour.
So time again to call bunk on this notion