<< Back to the Blog
Better technical systems by observing social systems

Automation is starting to replace humans across all aspects of work. To automate work we must observe the social parts of the system as these are the part that learn and adopt to the environment. In order to make automation successful within organisation, we must embed the ability to automate within systems themselves. People have great knowledge of how to work within larger systems and how to adopt to change. These patterns of adaptation are a great challenge for automation initiatives. But there is a way to make automation successful.

Automation of human endeavour is accelerating. Everything from manual labor, professional work even creative work is being automated. I am one of those people who believe that automation will set us free and create abundance for all. Of course there are going to be difficult times ahead, as is true with any major transition in our society. But this post is not about that. What I want to discuss is how to implement automation in our organisations. The approach I want to explore is how we can create better technical systems by observing and learning from social systems. This is indeed the entire point of automation. We do not automate anything that does not add value to us. And since the dawn of our civilisation it is human work that has added value. So in order to automate well or create better technical systems we need to observe social (human) systems. This article is not about repetitive mechanical tasks automation, but knowledge work automation.

Let us explore this concept through an example. Consider the system illustrated below. The systems applies to almost any organisation. We have been asked to automate the “Processing and Analysis” capability.

Decision Making System - Data to Information to Knowledge to Wisdom

Before we can begin thinking about automating the Processing and Analysis capability let us consider the system as a whole and several patterns it implies . * Processing & Analysis must not only access data, but the type of data it accesses will change because work changes based on knowledge generated by decision making * Processing & Analysis is also guided by wisdom developed by decision making, so a way data is processed and analysed also changes.

So we can deduct from analysing this system that Processing & Analysis does the following: * Accesses data * Generates Information * Adopts to new data * Changes how it does processing and analysis based on new wisdom

Great, so above illustrates what we need to automate and probably in what order and complexity. The picture below illustrates how Processing & Analysis sub-system works.

Decision Making System - Processing & Analysis

In order to understand this fully, we need to consider few other variables: * How much and what variety of data exists? * What information and how often does it need to be generated? * How often does work change and therefore data it captures and stores? * How does information generated by Processing & Analysis relate to information generated by work directly?

So how would be automate this sub-systems? A typical approach is to establish an “Automation Project”, gather requirements and then implement the technical system that delivers those requirements. There are several problems with this approach. * The most valuable parts of the sub-systems are “Adopt to New Data” and “Use Wisdom”. The only way to automate these is to use artificial intelligence, which we don’t have. * Automating the other parts is definitely do-able, but the pace of change may make automation difficult. * It is safe to assume that most people work in the “Access Data” to “Generate Information” flow. It is also safe to assume that this is where most knowledge about the sub-system is.

So, finally we get to observe social to develop technical pattern this post is about. Instead of starting the project, we make the “Processing & Analysis” sub-system capable of automating itself. The approach we could take is: 1. Ensure easy collaboration of everyone involved in the system, by either co-locating them or giving them great collaboration tools. 2. Establish a new part of the sub-system that monitors work of “Access Data” to “Generate Information” flow and identifies patterns that are then automated. 3. People displaced by automation, move to do work on adopting new data and using wisdom to change how data is analysed and information generated. Their knowledge will be very useful in doing this work. 4. As “Access Data” to “Generate Information” is automated, the focus of people in the sub-system becomes improving the automation of the flow, not working directly on the flow.

There are few fundamental benefits to this approach: 1. As more automation is introduced into the system less people are required to do low value work and more are required to extract even more value out of the process. 2. New technology can be incorporated fast as the sub-system is itself capable of responding to new possibilities. 3. The sub-system can respond faster to change as more effort is put into adopting to new data or wisdom. 4. The new work monitoring part of the “sub-system” can over time shift focus to automating high value / complexity tasks such as using wisdom to change processing and analysis work.

Here are a few final takeaways: * Automating systems requires intimate knowledge of systems that drive work that is being automated (outside-in perspective) * People doing the work currently are where the understanding of how to automate will come from. * People’s work will change from doing the work to working on the automated elements of the sub-system. * Automation never stops, because the nature of work changes. People collaborate and learn and therefore collaboration and learning are embedded into every system. Machines currently don’t, this is where displaced people should be put to work.
* Learning for the machines and then changing them. * External automation by a project is extremely inefficient, simply because people with working knowledge of what is being automated are working in the system not on it. * Automation can’t be fully achieved by “freezing the system” which is what requirements will do. Some change capability can be automated, but how to incorporate wisdom can’t be. * Watch the people and let them automate the system with support by introducing new part of the sub-system.

Just out of interest, have a look at an interesting video about automation and work.

Share This Article References