In my earlier days, when I was advocating the use of roles, or their dynamic equivalents, business rules (in combination with business attributes), I once ran into an enlightening discussion.
In an experts panels on an Identity conference I heard one of the experts state: "What sense does role modelling make, if, whenever we complete an organisations role model and are about to implement it, the organisation has changes considerably meanwhile."
Well "that's real life" we were tempted to say. But wait a minute! Can this really be true? I had to remember a second quotation from the same event "Roles are the DNA of an organisation”.
O.k. if you consider any commercial endeavour as a mild form of barely ordered chaos, if you run a company merely ad hoc, and role modelling is considered as imposing an unnecessarily rigid structure on top of a highly volatile ecosystem, that reinvents itself at every very moment - ok, in this case forget about any modelling up front.
- Introducing roles for process design or access control requires a certain process maturity – or it will fail.
The model driven corporationBut what, when it comes to automation? In order to automate an organisation you need to express it in a formal way, so that it can be processed by machines or - more often - in a blended fashion by machines and humans. You need to formally document your policies & rules, processes and roles in a complete, consistent and transparently executable way. Only then the processes can be repeatedly run by those above mentioned processors.
Considering this scenario, how can the organisation have changed while the role model is still controversially discussed? No way! The roles are the organisation, not just its mere documentation, or at least an important part of it.
This fundamental shift in perception of the importance of modelling becomes a vital precondition for any sufficiently complex organisation undergoing a digital transformation.
By gradually replacing human processors by automated actors the entire corporation turns into a hyper-system of interacting subsystems. The actors might perform simple, straightforward and deterministic activities or make use of heuristics, like deep learning based systems are doing.
The Behaviour of such ecosystems is hard to predict. They need to be tested carefully after changes having been applied and before releasing them into the wild. Otherwise unwanted effects might mount and result in cascading damages.
- Better don’t trust a model – unless it is sufficiently verified.
Simulation of complex systemsHowever test cases to be fed into the system consist of input data and expected results. Of these the latter may be hard to determine in such complex systems. So testing will rather morph into simulation.
Simulation hereby is understood as „Process of designing a model of a real system and conducting experiments with this model for the purpose either of understanding the behaviour of the system and its underlying causes or of evaluating various designs of an artificial system or strategies for the operation of the system“. Well a bit lengthy, seems to be correct however.
You feed in typical scenarios from everyday life or from some anticipated or even exceptional situations. For example if you run a seasonal business you may simulate low and high season, including bottle necks, cash flow minima and many more. For risk assessments you may generate a sufficient high numbers of random events to let some rare risks materialise. Robustness checks may be performed by some kind of perturbation calculation including a realistic number of cataclysmic events like strikes, political unrest, climate change effects or even regional wars. Adequate staffing, effects of fluctuations, diseases, might they be seasonal or epidemic, holiday seasons, labour market elasticity are just a few examples. The model will give reasonable answers to these vital questions.
- The digitally transformed corporation will be a model driven corporation.
Optimisation, the next logical stepMeanwhile the need for simulation has become widely recognized. There is however more to benefit from simulation than just avoiding traps and adding robustness. When running several simulations while varying the input parameters the desired outcome can be optimized. Variation of input parameters have to follow a strategy of course, in order to minimise the simulation steps.
There are several optimisation strategies in use for different purposes following different paradigms. The most robust ones follow evolutionary strategies.
Of course all enthusiasm for these powerful tools should not make us forget the potential pitfalls involved. Insights like these: "A model is a model is a model is a model" or "Essentially, all models are wrong, but some are useful" may remind us, that the model represents a stripped down version of reality only. We however regard it as the higher risk however to bypass the use of modelling, simulation and optimisation before transforming a whole business by applying leading edge digital technologies.
- But first you have to start with a model.
 As defined by R. E. Shannon in the seventies of the last century (Shannon, R. E.: Systems Simulation: The Art and Science, Prentice-Hall, Englewood Cliffs, NJ, 1975.
 Dierk Raabe referencing Rosenblueth, A.; Wiener, N.: The Role of Models in Science, in: Philosophy of Science, Vol. 12, No. 4 (October 1945), S. 316-321 (http:/www.jstor.org/stable/184253).
 Box, G. E. P.; Draper, N. R.: Empirical Model Building and Response Surfaces, John Wiley & Sons, New York, NY, 1987.