Planning for an uncertain future calls for a shift in information management — from single numbers to probability distributions — in order to correct the "flaw of averages."
This, in turn, gives rise to the prospect of a Chief Probability Officer to manage the distributions that underlie risk, real portfolios, real options and many other activities in the global economy.
- Sam Savage, Stefan Scholtes and Daniel Zweidler
There are some very serious misunderstandings going around about how management in the presence of uncertainty takes places in business. The basic conjecture is
Management Science's Quest: in Search of Predictability †
Let's start with a basic fact for all projects, all business processes - everything is a stochastic process. So searching for predictability is not a goal for any informed business or technical person or organization. If it is, then that defines the maturity of that person or organization. It happens, but it states up front little understanding of the underlying stochastic processes that create probabilistic outcomes of - Everything
Here's a quick review of both processes in play in all activties.
In the Decision Making Business there are four reasons why they are hard.
- Decisons are hard because of its complexity.
- Decisions are difficult because of the inherent uncertainty of the situation.
- A decision maker may be interested in working toward multiple objectives, but progress in one direction may impede progress in other directions.
- A problem may be more difficult if different perspectives lead to different conclusions.
So to start with the notion of predictability - it is simply not possible in any real project or business domain, to speak about predictability in the absence of the underlying statistical processes that create probabilistic outcomes.
Any credible business or technical manager knows this. If predictability is assumed or even desired, then the naivety of the manager is the only likely source, or maybe the intentional ignorance of the statistical and probabilistic nature of business and technical process. But predictability is not possible in the sense of absolutes, only probabilities.
So let's look at some less than informed concepts that are popular in some circles ...
- Predictability is a form of causality - predicting is separated from the source of predictions. And certainty the causality associated with prediction need not be there. Bayesian statistics and Monte Carlo Simulation, need not connect the predicted outcomes with the source of those outcomes - other than the source of the random variables from a generating function.
- Planning rests on the assumption we can predict - a Plan is a strategy for guiding our efforts to change something in the future or arrive at some place in the future. The Strategy is a Hypothesis and that hypothesis needs an experiment to test the current situation to determine if it will result in the desired outcomes in the future. This is core design of experiments that we all learned in our High School science class. Plans describe an emerging outcome.
- Goals change with the observation of reality - This dynamic adaptation process is what we, in the Agile community, call a feedback loop - this is true, but a target value is needed to compare that feedback information to generate an error signal. This is called Closed Loop Control and is the foundation of all control systems including Statistical Process Control system. And control systems that are adaptive in the presence of emerging dynamic systems. This is the basis of Learning Systems in stochastic adaptive control.
- Management techniques must not be based on the existence of a perfect, predictable future - this is a naive understanding of management. Perfect, predictable futures simplay do not exist anywhere for anything. All processes are random processes, many times not even stationary random process.
The suggestions above indicate the lack of understanding of fundamental knowledge of making decisions in the presence of uncertainty as described in the Making Hard Decisions book. The Journal Operations Research and Management Sciences, will put the science back in management science that those conjecturing the topics above seem to have missed.
In Journal papers and many books and related sources all the suggestions that we can't make decision in the presence of uncertainty, that simple minded conjectures like:
The basic problem with most perspectives on management today is that they are static analyses of a future environment. And all decisions are made because we believe we can predict the future.
Are simply not true, and better insight as to why they are not true can be had with straightforward reserch available by joining INFORMS or a variety of other professional societies.
- Academy of Management
- American Association for the Advancement of Science
- American Mathematical Society
- American Statistical Society
- Association for Information Systems
- Decision Sciences Institute
- Institute of Electrical and Electronic Engineers
- International Council of Systems Engineering
- Mathematical Association of America
- Mathematical Optimization Society
- Military Operations Research Society
- Production and Operations Management Society
- Society for Industrial and Applied Mathematics
- Society for Decision Professionals
So perhaps before making unsubstantiated claims about how modern statistical and probabilistic management processes are applied to business, some homework might be in order.
