#250: Thinking Systems

A toolbox for framing and solving problems.

Beyond Design Thinking: This post is based on a talk I’ve done a number of times. The idea started as a reaction to people constantly asking me to do a ‘design thinking’ workshop when what they really meant was ‘do something creative’, and create an interactive session on the subject (which was usually unspecified), preferably involving post-it notes. I realised that there was also a tendency to think of design thinking as a panacea to almost every problem - from strategy, to supply chains, or risk management. It’s a powerful tool, used in the right way but it’s obviously not the answer to every problem.

That led me to the idea of Thinking Systems - which is the idea of multiple modes of thinking including but not restricted to design thinking. Where each one is necessary but not sufficient. They are the tools in your thinking toolbox, and you need to find the right combination of tools for each problem you encounter. The more complex problems often require more than one tool.

(In Praise of) Design Thinking:

There are hundreds if not thousands of articles on design thinking. So I’ll keep this brief and share my perspectives in a few key points:

1. The popularity of design thinking in the digital age is due to the decades of badly designed software where the focus was on automating and standardising tasks rather than making things easier for end users. Software development, in a nutshell lacked empathy, and this culture extended to customer facing software as well.
2. In the digital age the explosion of choices meant that the competition suddenly got serious and the number of apps and web based tools for doing things increased dramatically. Suddenly, if you app didn’t work for the user, they were gone. Design was in the limelight.
3. Very quickly we understood that rather than beautifying screens once all the ‘thinking’ had been done, design was actually at the core of the development process helping to define what was going to be built and in which order.
4. Design thinking is really all about empathy. And being able to put yourself in the shoes of the user(s).
5. When done well design thinking often generates counter-intuitive answers. Years ago, when we we were designing an app for a major airport, the brief was to bring the brand experience of the airport front and centre in the experience. Our design led analysis showed that actually nobody wants to notice the airport. All travellers care for is the destination. And like a referee in a football game (or the government perhaps), your best work is done invisibly. You only get noticed when you’re messing up.
6. When we say users, we often focus on a narrow definition, but Google Glass showed us that the people in front of the spectacles were also stakeholders, almost as much as those wearing them. Not being empathetic to their concerns was arguably the key failure of Google Glass.
7. You don’t need to run workshops and projects to have empathy. Joseph Friedman noticed his daughter struggling to drink her milkshake in a San Francisco store sometime in the 30s, and went home to invent the bendy straw. A lot of design thinking today happens in corporate offices.
8. Probably my most contentious take - I’m not aware of any great product or service that was born in a ‘design thinking session’. Design thinking is great for validating, polishing, and improving products, or even invalidating assumptions that will help with the same outcome.
9. There are plenty of problems that can’t be solved via design thinking. For example, discover new drugs, build bridges and roads. Solve complex and interrelated problems such as easing traffic in urban centres. Design thinking can possibly contribute, but you’re going to need other weapons.

Engineering Thinking

Which brings us nicely to Engineering thinking. If design thinking gives you the insights into what people nearly need and want, you’re still going to need the mastery of tools to make it happen.

There’s a great story about Lego, who were a wooden toy manufacturer in the 1940s, when Ole Kirk Kristensen, the founder and owner had a vision for a plastic toy for kids. He wanted to create bricks that a 6 year old could snap together and take apart, but wouldn’t fall apart by itself. So far so good, and a great example of empathetic thinking. But that was just the start of the journey.

When Lego bought Denmark’s first injection moulding machine, it cost them half of previous years profits. And it still took them 10 years of effort, to get to that point described by Kristensen of how the bricks should behave. It was achieved by a specific design of the brick base and the top, so every piece could fit into every other piece.

(Fun fact about Lego, which I realised much later - despite it being plastic, it’s almost never wasted or discarded - every piece is useful and can just go into another set. Nobody throws away Lego pieces. It’s arguably the most environmentally neutral plastic in use anywhere in the world.)

But back to Engineering thinking - it’s the master of the tools and materials that helps you to design and build the things you envisage. Without it, everything is a pipe dream. To solve problems of product development - you’ll need the engineering to go with the design thinking. This is why Apple famously takes the metal shop as seriously as platform development.

If you look around you - your television, the door hinge, the boiler, your car and all its components, your phone, and all your electronics, the zips in your clothes, or the lock in your door. All these, and a thousand more things you probably encounter on a daily basis is the result of engineering.

From the MRNa Vaccine to fibre broadband, once a break through product has been created in a lab, it needs to be made at scale, and reliably, and at an affordable cost, and using environmentally friendly ingredients, and be easy to maintain, for mass production and consumption.

And yet, like Design thinking, Engineering thinking has it’s own overreach in terms of the Engineering Mindset. Engineering flourishes in environments where components behave in predictable ways to deliver preferred outcomes, and built to specifications. The engineering mindset to any problem therefore tends to be component focused. If you take a part a complex engine, identify the faulty piece, fix it, and put it back together, the engine should work. And in the industrial era which is dominated by engineering - this mindset has permeated in organisations and even governments.

This is a problem because the engineering approach tends to break down when the specifications aren’t clear, when the behaviour of the components aren’t predictable, when the interdependencies between the components are too complex to isolate, or when the components themselves evolve and change, or when the environment you’re designing for keeps varying beyond the limits you’ve set. Solutions such as replacing people, restructuring, or changing a part of the organisation often don't work because organisations aren’t like machines - they are far more organic. Which is also why organ replacements aren’t simple, or why a body might reject a perfectly healthy new organ.

When policymakers and governments adopt this mindset, they risk significant failures because you can’t componentize society, or the economy, without interdependencies. You can’t fix housing simply by building more houses without paying attention to the impact on traffic, on environments, on local schools and hospitals, or the knock on effects on roads.

A lot of organisations and managers resort to this kind of engineering thinking even if they aren’t engineers by education. Yet, as we stand at the end of the industrial era, we might be seeing the end of the primacy of engineering thinking. Engineering will continue to be critical to product development, and it’s the only way to translate insights obtained from empathy and design research into products that actually meet the uncovered needs. But like design thinking, it can’t be used indiscriminately on organisations, complex systems, or policy making.

Systems Thinking

Systems thinking is a way of understanding complex problems with inter-related components as a whole, by understanding not just the components but also the relationships between them, including the flows and dependencies. Because complex systems don’t behave simply like the sum of their parts. Economics students might remember the paradox of thrift, where everybody saving a bit more leads to a collective drop in savings in an economy.

One of my favourite stories is about the village in which there was a snake problem, so the local government issued a reward for villagers to kill snakes. To claim the reward, they had to bring the dead snakes in. Some enterprising villagers then decided to breed snakes, so they could occasionally kill one and bring it in for the reward. The local government realised what was going on and stopped the reward. So the snake breeders let the snakes go, as there was no money to be made. Consequently the snake population of the village went up. This is an apocryphal story and no animals were harmed in the telling of it), but it beautifully illustrates the challenges of complex systems. This is almost always true of economic policy decisions, where the net impacts, and potential harms are always hard to predict.

In the 1950s, Jay Forrester modelled systems behaviour at the MIT and showed how in supply chains, small imbalances in demand vs supply could be amplified through a system to result in significantly high levels of stockpiling, thanks to positive feedback loops Forrester’s field of work - System Dynamics - is critical to understanding of how systems behave - especially in terms of change, or reactions to an external input.

For example, a city might react to congestion on the roads by adding new roads but the availability of roads leads to faster transits, which in turn may lead to more cars being bought and driven, ultimately leading to similar levels of congestion on a wider network. So reinforcing as well as balancing loops can influence these interrelated parts of a system. System dynamics is a quantitative science, but even qualitative systems thinking can be applied to understand the interdependency between elements of system and to make better choices about the outcomes.

Another area worth exploring is Network Science. The mathematician Albert-Laszlo Barabasi has an excellent book on the subject. Networks tend to have common characteristics and behave in predictable ways if you know how to model these characteristics. For example, in starling murmurations, each bird is ‘connected to’ or ‘keeps track of’ 7 others at any point of time. And birds at the edge try to move into the centre, for safety. This is how the dynamic network of a murmuration is constructed.

A lot of biological and ecological models have been studied through history but it’s only in recent decades that formal systems thinking has been recognised. It’s a complex and perhaps less understood model, and shouldn’t be used for simple problems, but for large and complex problems it’s an invaluable part of the toolset.

Systems thinking is usually not actively used in organisations, and it should be formalised as an area of competence, especially for innovation and problem solving functions.

Data Thinking

The 21st century has definitely added a new layer to the thinking systems - AI and Data thinking.

Data has externalities. This is incredibly powerful and very undervalued. For example, tracking the environmental impact data of a business over time seems harmless, but could give you a good idea about its business performance and cycles. The map below is a map of all global shipping lines but it also creates a relief of the world map. This is data externality at work. Data collected for purpose A, may turn out to be really valuable for purpose B.

Every digital action creates data. So our ability to capture data from multiple sources and make sense of it can become a super power. There is of course a signal to noise challenge. The problem is also that every organisational system is designed to keep data in, so sharing data is not easy. And a significant percentage of all data collected is still discarded today.

There’s an interesting story about a chap called Henning who created a Fantasy Football team and he also created and tracked groups of people who played the game from each of the big football clubs. And he would publish his findings and plans online. Now, when a player gets injured, the first people to take him out of their fantasy teams are his own club members. And so just by tracking that Henning could predict who was injured and unlikely to be played. Which he published and managers found that their squad information was being “leaked” before games.

Data can also be misleading - most people know the story of Abraham Wald. In the second world war, the US airforce was evaluating fortifying planes but faced the problem that the weight of the metal meant that they had to choose which parts of the plane to fortify. Data from all the planes that had returned from aerial warfare suggested that the damage was on the wings, so that’s where the planes needed strengthening. That was until Wald showed up and pointed out that the data was biased because it only included the planes that came back. Once you considered the universe of planes that went into battle, it became apparent that the planes that perished were actually damaged in the fuselage. Which completely flipped the logic of where the metal needed to be added to the planes. This is an example of survivorship bias, which was also at play in the Challenger disaster at NASA.

Learning to exploit these properties of data is one additional thinking model for any innovator.

AI Thinking

We are now entering the era of AI, so it’s important that we reflect on what AI thinking is all about. I think of a few key areas, where AI has added something new to how we think about problems.

First, it’s the fact that the technology is both improving and is a learning tool. So it’s capabilities may naturally be better tomorrow, compared to today, at learns and evolves.

Second, AI systems can be connected to a common central “brain” with a vast range of artificial sensing collecting data from any number of sources. So the rate of learning may actually be many order of magnitude higher and may be driven by the number of nodes on the network.

Third, some domains are extremely complex in terms of input variables and data sets. Think healthcare, macro-economics, or managing environmental impacts of cities. Historically we gather data based on our pre-articulated hypotheses, so we can prove/ disprove our hypotheses. Now we might be better of throwing the data at the machine, and letting the AI identify correlations around which we can build hypotheses. This has sometimes been referred to the New Radical Empiricism.

And lastly, (for now), I’ve spent much of my career arguing that the business problem comes first, and then you find the right technology to fix it. But with the rate of change of AI, its becoming apparent that the boundaries of what is possible is constantly shifting, so we do have to go the other way as well. Sometimes it’s useful to ask what can we do with this tech that we couldn’t do before, and how could it improve our business (or our lives)?

Philosophical Thinking

It’s apparent that the current technological capabilities spearheaded by but not restricted to AI gives us near superhuman powers. Once you combine this with immersive environments, quantum computing, robotics, drones, and blockchain, for example, it gives us superhuman capabilities. And with this comes both the power and the responsibility. Technology can do great good or great harm (both intentionally and unintentionally as I’ve suggested above). So the key question of our times is not just can we, but should we? We have to think about the first and second order impact of our technology choices, and at least make an attempt to identify the unintended consequences to minimise risk.

There’s also the opportunity and need to look at the bigger picture of the planet, the environment, of the global health and education challenges, and the raging inequalities, and pose to ourselves the ethical questions around how we deploy tech, and what responsibilities we have to those displaced by the technology, or those bypassed by it. And what broader impact might it have on society and the environment. It’s critical that we approach problem solving also with an ethical framework.

As AI gets smarter and more capable, it will force us to shine the torch on what it means to be human. AI generated art may be beautiful but is it truly art without emotion, or experience? What is the value of human interaction? Or of being human? What will give us meaning if / when AI does a bulk of our work?

And finally, when AI generated content becomes very real, every video and image has the potential to be fake or AI Generated, what will we trust? Will need to revive empiricism or become cynics?

Your Tool Box

This is therefore a long and rambling post that started with my issue with design thinking as the answer to everything and led to this exploration of the different types of thinking models. Collectively I call it a Thinking System, and it’s like having a toolbox of these different modes of thinking, and being able to use the right tool. Thinking in terms of design, engineering, systems, data, AI, and philosophy are 6 very basic models. There are potentially dozens more and the possibilities of combinatorial models is also endless. But starting with 6 and being competent in using these in the right way would be an excellent start.