Is it time for more Systems Thinking in IT - DevOps and IoT, in particular, seem to be all about feedback controlled systems

Written By:
Published:
Content Copyright © 2016 Bloor. All Rights Reserved.
Also posted on: The Norfolk Punt

As I think about where business automation is going, I begin to wonder whether we need radically new ways of thinking about automation and the way it interacts with human systems, so as to deliver “people centric computing”.

By “radically new”, I do mean (as is usual in IT) applying ideas developed in the middle of the last century by Ludwig von Bertalanffy and others. In fact, Systems Thinking formed the most interesting part of my graduate course in computer science, back in the late 1970s. So, perhaps not “new”, but “ahead of their time” – and that time has now come, at least in an IT context (again, as usual, other disciplines may well be in advance of IT).

I think we need to think of business automation in terms of open feedback-controlled business systems, exchanging information with their environment, and delivering business outcomes in a state of dynamic equilibrium. If we want to change the business outcome, we need to nudge the system into a different state of dynamic equilibrium. However, there is a danger of “emergent behaviour”, where the system state changes in unanticipated ways, as a result of complex feedback interactions, resulting from the interactions of so many things that the unlikely is almost guaranteed to occur relatively frequently.

How do we govern such systems? Well, we need to define the systems boundaries and we need a controller with more “variety” than the systems being controlled – more degrees of freedom than the systems enjoy. The “controller”, I think, will be a human being interacting with automated systems, making use of models, in order to understand the complexity of the systems being modelled. This is somewhat similar to “model-driven development”, as espoused by (for example) Mendix here, but extending well beyond the world of technology. Perhaps decision modelling (the Sapiens DECISION approach is interesting) will be the way such systems will be governed, using business rules logic, with simulation used to validate the behaviours resulting from changes in the feedback loops and the micro-behaviours of system components.

Is this relevant to the practical issues facing the CIO today? Yes, I think it is – the CIO has to be, in effect, the controller with the requisite variety needed to manage the automated business. Or, rather, the CIO is part of the “systems of systems” that provide good governance; he or she has to be the controller with sufficient requisite variety to manage the systems that manage the systems that manage the automated business…

In a more practical context, I think that systems thinking should be driving the way the CIO responds to the major trends he or she is facing today. For example:

  • IoT is the concept of intelligent things talking amongst themselves in a state of dynamic equilibrium and producing business (human) outcomes as a result of system state changes, influenced by a human controller and human-imposed rules and feedback loops;
  • DevOps is, in essence, treating business automation as a system controlled by feedback loops, culminating in a loop between the end user’s business experience and the requirements management process behind development or services orchestration. It results in a delivery pipeline, that delivers people-centric automation, starting from a process improvement vision.

I would be interested to learn whether this resonates with any of my readers.