systems complexity: what we learn in school

More years ago than I care to remember I taught, briefly, at a parent controlled school – along the lines of Michael Gove’s free schools, but in those days parents had to stump up the cash themselves. One of my first tasks was to draft a curriculum. The experience stood me in good stead when I found myself educating both my children at home.

What had concerned me most about my children’s education at school was not so much what they knew or didn’t know but what they understood about the world they live in. As my eldest put it; “We were taught about the Egyptians, the Greeks and the Romans, but I never understood why, or what they had to do with each other.”

After some trial-and-error (the standard school timetable was a non-starter) we adopted the history of the universe as a narrative spine for our learning. We started with the Big Bang and proceeded from there. We made a timeline of the universe that stretched the length of the house. The periodic table filled one wall of our dining room and the rest of our home was festooned with posters from the excellent Edugraphics. We found out what life must have been like for the young Mendeleev and for the inhabitants of Darmstadt during WWII. We studied evolution and creation stories, unearthed skulls with Leakey and watched our distant ancestors farm and develop city-states. My youngest returned to school just after the fall of the Roman Empire. I must remember to let him know what happened next.

Few teachers would think of introducing an eight year-old with special educational needs to sub-atomic theory, but for my son, that knowledge made sense of everything. Once you have a basic deep structure understanding of the connection between energy and matter, how elements interact, what DNA does, how brains process information and how people tend to behave, you have a broad framework into which all new surface features of knowledge fit. So new knowledge, whatever it is, makes sense.

But the school curriculum (in the UK at least) tends not to start from first principles. It usually begins – understandably and justifiably – with building on young children’s existing knowledge (My Family, Our Town, sand and water play). It’s later dominated by the requirements of academia. What undergraduates are required to know largely determines the content of A level courses, which in turn determines what is learned at GCSE level and so on. Add to the mix what politicians or other interested parties believe children should learn and you have a curriculum that is derived neither from the deep structure of knowledge nor from how children learn.

Using deep structure as a starting point has a number of advantages. It enables you to understand:

-how everything is related to everything else (however distantly)
-how skills and knowledge are related
-the importance and relevance of different skills and different types of knowledge

Schools have always had a problem with non-academic skills like plumbing or painting and decorating, partly because they are non-academic skills but also because of their social status. Because fewer people have the skills needed to become lawyers or doctors, these professions command high salaries and high status. Schools tend to measure their success by the number of their graduates who go into high status professions. Not on how happy those graduates are with their work or how useful they are to their communities.

We are frequently reminded that our knowledge about the world is growing at an exponential rate and that specialists can’t hope to keep on top of their own field, never mind others. This has led to increasing specialization and as a consequence there is pressure on the school curriculum to become fragmented and unconnected. Increased specalisation might be inevitable but it doesn’t follow that economists don’t need to understand human behaviour, or that doctors don’t need to grasp the principles of nutrition or that journalists don’t need to know how the brain works. Nor that it’s OK for politicians to understand only politics and not the principles that link everything together.

Dr Beeching, I presume?

When it was nationalized in 1948, the UK rail system was already inefficient. It had evolved piecemeal over the previous century into an overextended sprawling network. Passenger numbers had been hit by the increased use of motor vehicles and continued to decline in the post-war period. A commitment to full employment by the Labour government and increasing union power due to the post-war economic boom set the scene for a two-week rail strike in 1955, forcing rail freight users onto the roads, where many of them stayed. A White Paper in 1960 recommended splitting the integrated national transport system. Rail – then making a significant annual loss – would be run by a new British Railways Board, and a programme of complete modernization was proposed.

Richard Beeching was a research physicist who had risen through the ranks of ICI to become technical director when in 1961 he was appointed chairman of British Railways. His task was to make the nationalised rail network profitable. Quite why someone with no experience of the rail industry was appointed to this post remains a mystery. Maybe it was thought a physicist would understand the technicalities of rail. Perhaps it was felt that someone with a rail background wouldn’t be sufficiently ruthless. Or maybe Beeching was considered thick-skinned enough to take the blame for savage cuts. I won’t speculate on the motivation of Ernest Marples, then Minister of Transport.

According to Robin Jones’ fascinating account Beeching: 50 Years of the Axeman, one of Beeching’s criteria as to whether or not a service should be spared his now legendary axe, was direct profitability. On the face of it this seems perfectly reasonable. It certainly made sense to replace a branch line carrying a dozen passengers a week, with a bus service. Unfortunately for Beeching, many unprofitable branch lines contributed much of the traffic that made mainlines profitable. And according to Robin Jones, Beeching assumed that long-distance passengers whose branch line had closed would drive would drive to their nearest mainline station and complete their journey by rail. Instead, partly because of the new motorways, car owners found it more convenient to keep driving.

The way Beeching wielded his axe is an example of a classic systems-change error, known as sub-system optimization at the expense of system optimization. In an interconnected system, changing one component of the system will affect other connected components. The tighter the connection, the greater the risk of unintended or unwanted outcomes. Since rail branch lines are tightly coupled to mainlines, the effect of closing branch lines was considerable.

In addition, Beeching committed a second common systems-change error; making unfounded assumptions about another interacting system – in this case human behaviour.

The problem that we humans have with complex systems is that they are complex. With incomplete knowledge and a working memory that can hold seven-plus-or-minus-two bits of information, it’s very difficult for us to look at systems as a whole. That creates a lot of problems. We tend to optimize our own immediate situation regardless of the impact that has on other people or on our own long-term outcomes. Governments tweak sub-systems oblivious of the impact on whole systems and then have to tweak other sub-systems to compensate.

Beeching’s systems errors and Marples’ policies had a lasting impact on the transport infrastructure of the UK, with incalculable cost implications for the economy as a whole. It’s only been since privatization in the 1990s that rail passenger numbers have recovered to levels comparable to those prior to the Beeching cuts. (Whether or not the increase in passenger numbers is due to privatization or due to traffic congestion, petrol prices and difficulty parking at stations is a moot point.)

It’s interesting to speculate on how a biologist might have approached the task of making the railways efficient, since the systems that biologists are familiar with are significantly more complex than those that engage the attention of physicists.

Next, I plan to look at levels of complexity in systems.


Jones, R. (2011). Beeching: 50 years of the axeman. Mortons Media Group.

Miller, G. (1956). The Magical Number Seven, Plus or Minus Two: Some Limits on Our Capacity for Processing Information, Psychological Review, 63, 81-97.

form and function

In the next few posts, I plan to look at some examples of how systems issues affect organizations, especially public sector ones – education, health and social care. I then want to explore how people’s theories about the content of education, health and social care interact with the form of public sector systems. My particular areas of interest are in education, the neurobiology of developmental disorders and conceptual modeling in scientific research, so later posts are likely to focus on those topics. In the meantime, I want to pay tribute to the pioneering work of an organizational researcher little known outside her field.


In the 1950s Joan Woodward (later a professor at Imperial College) carried out a study of the structure of manufacturing firms. Manufacturing was of vital economic importance in the aftermath of WWII, and Woodward’s team was trying to find out what made efficient manufacturers efficient.

What they found didn’t make sense until they took into account the type of technology firms were using. Woodward identified three main types of production system; unit or small batch, mass/large batch, and continuous process. In essence, what she discovered was that whether a manufacturing process was efficient or not depended on the technology used, and the technology depended on the nature of inputs and outputs. It would be inefficient to manufacture packs of granulated sugar from sugar beet one at a time – continuous processing would be much more effective. Similarly, customized wedding dresses couldn’t be produced efficiently using an assembly line – a unit process would be more appropriate.

What Woodward also noted was that once you knew what technology is being used, you could accurately predict what the organization’s structure will look like. Unit/small batch organizations had ‘organic’ structures, with most workers reporting directly to CEOs, whereas mass production organizations tended to be run as bureaucratic hierarchies with many managers.

Woodward’s contribution to organizational and management theory was an important one, but she had stumbled on a principle already familiar to biologists; that organizations are as much subject to physical constraints and affordances as living organisms, and that both inhabit their own ecosystems.

Gibson, J.J (1979), The Ecological Approach to Visual Perception, Lawrence Erlbaum.
On my reading list.
Pugh, D.S. (Ed.) (1997), Organizational Theory, Penguin.
A good introduction to the history and development of organizational theory.

logical incrementalism

When I was thinking about a name for this blog, logical incrementalism sprang to mind and wouldn’t go away, despite being a challenge to say and spell. The term comes from the decision-making literature. In the 1950s Charles Lindblom suggested that minor decisions in organisations tend to be disjointed, not related to each other or to an overall strategy – decision-makers tend to ‘muddle through’. JB Quinn, 20 years later, thought that incremental decisions could be logical and integrated within a strategy. Whether Lindblom or Quinn or both are right depends on what you think is logical.

Incrementalism isn’t confined to decision-making. It applies to all systems. An extra proton creates a new element, a tiny change in genetic material gives an organism a better chance of survival, a small price increase moves a business from the red to the black. Each incremental change makes some future changes more likely or inevitable, and makes others less likely or impossible, so all incremental change has an inbuilt logic to it, even if it isn’t obvious. And all incremental change has logical outcomes, even if these are sometimes unpredictable because the system is complex.

I’m interested in systems – biological, social, economic… When people think about a system, they often focus on its content (what does what) rather than its form (why things happen how they do). What’s interesting about the form of systems is that regardless of their content, systems have many features in common. Commercial companies behave in similar ways to organisms competing in an ecosystem; hooks catch hold of things, whether the hook is on a seed casing or an industrial fastener. In this blog, I plan to explore the form of systems and how features of systems impact on everyday life.
 Thank you for reading.


Lindblom, CE (1959). The Science of ‘Muddling Through’. Public Administration Review, 19, 79-88.

Quinn, JB (1978). Strategic Change: ‘Logical Incrementalism’. Sloan Management Review, 20, 7-19.