An intro into the world of systems

Both the 20th and the current century have seen humanity make leaps and bounds in technological advancements. These advancements have come at a big cost to the ecological balance of this world. We as a society are currently facing another major crisis today in the form of a pandemic which has devastated the globalized world by destroying lives, livelihoods and has questioned the benefits of the so-called globalization phenomenon. When you ask a layman about solutions to these massive problems, many of them answer that it’s the system’s fault or the government’s fault. But what exactly do we mean when we talk about this so-called system? Does the system simply refer to the government or does it go beyond that? This is precisely what the book Thinking in Systems by Donella Meadows explores.

I have structured this article around 5 themes. I have tried to introduce systems theory for people with no background in systems. Then I present the two categories of feedback loops introduced by Meadows in her book followed by a discussion on the general characteristics of systems and why our models will always be incomplete. Then, I summarize some of the solutions she presents to avoid some traps faced by the society. I have tried to stick to the original message of the book. However, some of the discussions do tend to deviate from the original book’s ideas and those evident thoughts are my personal opinions.

As a practicing engineer, systems theory is not an unknown school of thought for me which made this book on first glance an uninteresting read. But, I am glad that I picked it up upon seeing this book recommendation also flash up on Patrick O’Shaughnessy’s suggestion list whose podcast Invest like the best which is one of my go to podcasts these days. This book goes much beyond crunching complex numbers but rather presents the ideas of systems theory such that they can be applied not only to engineering systems but provides a framework (not direct solutions of course!) to understand complex issues such as climate change or even to think about things such as birth rates or death rates evolving as a function of time in Europe or Asia and the list goes on. This is by no means a perfect book and has its limitations but is a good primer on systems.

To someone who is unaware of the systems lens, it can be explained using a simple example of any team sport. Imagine a team sport such as football (please don’t call it soccer) or basketball. The team is the system. Each player is a part of the system (call it the sub-system) and performs specific functions for the team. You can further classify sub-teams into sub-sub-teams and beyond. For instance, a football team can be divided according to their role (defenders, midfielders etc.) which can be subdivided further. This can be applied to our ecology, human body, governmental structures, basically every object and institution you see. The human body as a system is another simple example with your brain, liver, heart etc being the parts of that systems whereas the tissues and the cells in those sub-systems are the sub-sub-systems and it goes on. According to Meadows, the ‘parts’ of the systems are most visible and as expected they are important to describe your system’s behaviour. However, it is the interactions or the inter-connectedness between these parts known as flows which are also extremely relevant and influence the system’s behaviour more than the parts themselves (not always true though). It is this flow in the system that makes it not necessarily equal to the sum of its parts.

An appealing way to understand systems is to create ‘flow diagrams’ as illustrated via an example (diagram from the book) in the Figure below. A term introduced by Meadows to refer to the existing state of the system is stocks (Example-Number of trees in a forest at a particular instance is a stock if you are monitoring deforestation in that forest over a period of time). The flow diagram below monitors the stock of lumber to the stock of trees in a forest. It shows both the inflow and outflow of the stock of trees with the inflow being the growth of new trees and the outflow being tree deaths and logging.

Example on flow diagrams to visualize systems (Courtesy: Thinking in Systems by Donella Meadows)

A trivial yet an interesting lesson is that the stock of trees can be controlled by both inflows and outflows i.e. either growing trees or killing them (a fairly simple idea!). According to Meadows, the human mind focuses mainly on inflows but misses the outflows which is true in many instances since we are too focused on the catchy or more sensational bits of a story which again brings me to the more general problems faced by our society. A pathway to start tackling the climate crisis can be to think in the direction of economic ‘degrowth’, a word which scares many! But how else can you manage finite resources in the world if you continuously keep expanding and growing is a question which needs to be pondered upon and is an open question to which I have no straight answer (Circular economists might have something to say about this). Meadows also emphasizes that stocks in a system take time to change because of delays, buffers, politics, inherent system dynamics and people tend to underestimate this momentum which implies that shifting from an existing state to a new one takes time. Hence, shifting our economy from oil to more sustainable sources of energy will take time as we can already observe not only due to technical challenges, but politics and policy level issues as well. On the other hand, these time lags are not necessarily bad in all circumstances and can help us in some ways since they allow us to experiment and fiddle around.

When talking about system, one has to discuss feedback loops in the system with two of those categories being:

• Stabilizing loops: The simplest example to elaborate on stabilizing loops can be an air-conditioner or a heating system where these conditioning/heating systems act based on the feedback in your room. If the temperature in your room deviates from the levels you desire, the systems react to minimize the deviation and heat up or cool down your room to comfort you according to your requirements.
• Reinforcing loops: Reinforcing loops can be elaborated using the simple example of compounding. If you invest x dollars and you get y% interest, you see your money growing at an exponential rate over a period of time (too many assumptions!).

In principle, you can describe many systems you observe in reality as a combination of these two categories of feedback loops which means that you can explain complex phenomenon by splitting them into simpler and more understandable phenomenon. Of course, reality makes it a mission to complicate things (delays, buffers, black swan events etc). A very large system such as our society has many industries (for example) intertwined which leads to delays of several kinds. These delays lead to oscillatory behaviour in the system complicating things further. However, it also implies that understanding these delays and accordingly framing a plan can be a good policy lever to influence decision making. Additionally, using a systems approach also provides you with a tool at hand to look at a policy problem (for instance) from both a qualitative and quantitative perspective. Crudely speaking, you can evaluate the problems you face by ranking these feedback loops and accordingly make your decisions. The utility of the model you create logically depends upon whether the feedback loops you consider lead to realistic patterns of system behaviour which also implies that a more complicated problem is likely to lead more the number of loops competing with each other making this approach not the easiest one for all problems. The brings to one of the limitations of this approach. It is not a ubiquitous solution and a reductionist approach can be much faster for solving a certain set of problems. However, I do believe that there is no magic potion which can solve all problems we face and therefore the onus lies on us to choose an appropriate method for tackling a particular problem. The simplest rule of thumb could be to decide based on the time frame of interest. For short term decision making, a reductionist approach can serve well over a more complex systems approach whereas for bigger problems with medium and longer term horizons, a reductionist approach can lead to blunders and a more robust systems approach can serve us well provided that all the points of leverage are accounted for in the systems model. From my personal experience as a CAE engineer, your model does not have to be very complicated to give you some very precise estimations be it a reductionist or even a system approach. What you need in most practical circumstances is a result which predicts your desired variable or variables to the correct order of magnitude.

Meadows proposes three reasons as to why systems (say human bodies, economies etc) behave so well in a particular environment:

• Resilience: Resilience is an inherent feature of any system since the combination of feedback loops which make up the system ensure that the systems corrects if there is any deviation from expected outcomes (As indicated, many systems are a combination of at least one stabilizing loop and at least one reinforcing loop). A relatable example is eating too much food from a restaurant at once can be pleasing for your taste buds not necessarily your gut with a possibility of a stomach upset. However, our gut is extremely efficient and tries to digest the fast food we ate. However, exceeding the threshold will certainly lead to a stomach ache. From a practical perspective, it is important to understand a system’s resilience limits (it’s not infinite of course) since it is the breakdown of these resilience systems that lead to terrible consequences (ecological disasters, chronic diseases etc). Understanding these limits helps one preserve the system’s restorative capabilities and improve it if the resilience is low. Regular exercise is essential for every individual since it ensures that the resilience of our organs is maintained. Basically, exercise can be thought of to be like the regular maintenance and oiling of your organs.
• Self-Organization: Meadows defines self-organization as the capacity of systems to evolve structurally. All biological organisms have evolved over millions of years due to this ability of systems to evolve. Self organization produces unpredictable outcomes. Therefore, policy makers should devise policies which promote self-organization by not stifling innovation and creativity. This idea or rather an extension of it is also laid out in many other books one of which (Creation by Steven Grand) inspired Jeff Bezos to focus on developing ‘primitives’ which then evolved under the right market circumstances which led to a successful product portfolio commonly know as Amazon Web Services (AWS) which is one of the most profitable businesses for Amazon Inc.
  
  Hierarchy: A third aspect which complex systems often lead to is hierarchies. This is kind of contradictory to the previous argument which emphasize on independence. However, a complex system cannot evolve if it does not have stable intermediate forms which is why hierarchies are common in many systems. This hierarchical structure ensures efficient pruning of information by different subsystems rather than a single entity controlling everything which for very complex systems can be very damaging. When talking about hierarchies within systems, what is important is to focus on optimizing the system’s behavior rather than optimizing the sub-system’s behaviour. Meadows puts it very well by saying that efficient systems or functional organizations of all kinds should have sufficient central control for achieving appropriate levels of coordination needed to achieve system’s optimum goals. However, this should be delicately balanced with the autonomy of sub-systems such that the sub-systems can keep flourishing and growing. I feel that this idea is extremely powerful in thinking about so many societal issues which is why I personally discourage any form of extremist thinking when it comes to politics or policy making and hence, whenever you find me discussing some ideas about politics or policy making, you will notice that I never prefer extremes. The extreme philosophies stifle either social innovations or economic innovations. I am trying to get a more nuanced and unbiased perspective on the different political theories and I hope that I can write a article expressing my thoughts in detail on this issue in the near future.

The systems approach as many other approaches of looking at a problem are limited in their scope and as mentioned earlier, not ubiquitous. This is because like other approaches, systems theory is a mathematical model of the universe, society (Whatever your system is!) and models often fall short of reality. The next few arguments discuss why despite such sophistication, why the systems approach is limited.

• System based models are designed for a given set of operating conditions or circumstances and often circumstances change! For example, if you had planned to buy a car in 2020, I am certain that you did not account for the pandemic and probably have decided not to buy one! In addition to the inability of humans to predict future events or future circumstances, even if we have fairly robust systems, these systems are generally designed to operate well in a linear environment since our mind perceives linear changes fairly easily compared with non-linear changes (and we are good at solving and interpreting linear equations!). However, the world is full of non-linear events. Add to this, some discrete events (black swans-check out Nassem Taleb if you haven’t) which can have a profound influence on all kinds of systems. There was an instance in history some 20 odd years ago when Yahoo had the opportunity to buy Google. A different outcome of that business deal than the one we have would have made the world a very different place.
• Meadows emphasizes on the fact that the boundaries we create when defining systems are virtual and non-existent in reality and the fact is that strange things often happen at these boundaries. Many times, organizations conveniently ignore things outside these pre-defined boundaries and market themselves in a positive light when in reality, the performance inclusive of those boundary events or circumstances might tell a completely different story. For example, when we consider the carbon footprint of a particular product say for instant an electric vehicle, what should matter is the entire life-cycle of the particular product which includes the footprint of the source (electricity is generated from many sources of energy most of which are currently non-renewable although that is changing fast), footprint associated with manufacturing and assembling the vehicle and then the tail pipe emissions. When considering many of these factors, EVs edge out engine driven vehicles by a considerable margin. However, that does not mean that EVs are without their own challenges since battery technology needs to improve significantly if they are to become the primary mode of transportation. Metal ions used in batteries are a significant cause of concern from an environmental perspective, a problem which needs to be acknowledged

https://www.youtube.com/watch?v=1Xwxe0wU4b8&ab_channel=RealEngineering

That being said, current data shows that electric vehicles despite their challenges are a far better option than ICEs to reduce the emissions in the environment. However, as mentioned earlier, there are a lot of delays in systems and these are so clearly noticeable when we look at the current scenario in the transportation industry and changes in the transportation industry will be slower due to these buffers than projected.

A more human reason as to why systems surprise us is the ‘observability problem’ (people familiar with measurement systems will already have an idea about this!). Real life system are multi-input, multi-output systems with a complex non-linear effect of a particular input or inputs on a particular output or outputs and as humans we emphasize on specific inputs/trigger points based on our limited observation set. A very simple example is that your body won’t grow in a healthy manner if you provide it a lot of carbohydrates but fail to provide it with sufficient sources of protein. Therefore, any physical entity should understand these multiple inputs and their effects on the outcome. These are called Layers of limit by Meadows. She also emphasizes that even if an organization understands all the layers of limits, it does not necessarily create a pathway towards a recipe for perpetual growth as you cannot have that in a finite resource environment although the space ambitions of Jeff Bezos and Elon Musk might just expand our resource pool significantly but that is expected to pan out far ahead in the future if it does pan out.

The idea that I am about to discuss is one of my favourite parts of the book known as Bounded Rationality (a term coined by Nobel Prize winning economist Herbert Simon). The idea is based on the simple premise that people make decisions based on the information they have. However, people normally do not have perfect information leading to under or over-utilization of resources. In addition to this dearth of information, our efficiency in processing the available information is not perfect as well. This idea has far reaching consequences to understand some general fallacies or ideas that we have as ‘lay people’. Whenever people are fed up with some corrupt politicians in a given society, they usually think that the best solution is to replace those so-called corrupt politicians and if they do that, all their problems will be taken care of. This is true only to some extent since not all their problems would be over even if they replace the so-called corrupt individuals with kind hearted ones. This is because individual decision making is ‘narrow-minded’ since the information flow to which you are subjected as a politician is limited-Bounded Rationality. Although, as emphasized by Meadows, this is no reason for justifying shallow ideologies but it explains the occurrence of common human behaviours. Being aware of this concept of bounded rationality allows one to look at the information flows, incentives, disincentives etc from a wider perspective which can certainly improve decision making. Meadows brilliantly sums up this idea that bounded rationality of each actor is determined by his or her incentives, disincentives which may not be optimal for the overall system’s behaviour. Therefore, replacing existing actors (or politicians in the previously considered example) will not solve the problem. The solution to this problem is redesigning the system, its structures, incentives and disincentives. Only then will the change of actors make a big difference!

The next question which arises is to use this understanding of systems to create a framework or to have solutions to the problems we face in our society. I felt that this part of the book is a bit narrow since it proposes solutions to only a limited set of problems which is fair. However, it limits the readers ability to think about problems beyond the use cases it discusses. Additionally, the solutions themselves are very generic with less emphasis on complex and nuanced problems. Systems approach is more broader than the examples discussed in this part of the book. Although the examples themselves are not bad by any means but are just limited and narrows the scope of systems thinking which I guess was not her intention. For the sake of completeness, I will still highlight the solutions proposed by Meadows using systems to solve general problems before concluding this article:

• Trap : When a resource is shared by people, not only are its benefits shared but also the costs of its abuse. Solutions: Educate the abuser . A second solution is to privatize the resource which is reasonable since the abuser faces the consequences of his/her actions. In case, the resource cannot be privatized, regulate the same. All the three solutions have their advantages and disadvantages which are not elaborated here since the article has already become longer than intended!
• Trap : Allowing performance standards to be influenced by the past can set up a reinforcing loop if there is a negative bias in perceiving past information drifting performance towards the lower end. Solution: Keeping performance standards absolute and using the best performance as the benchmark instead of using the worst performance as a reference. This proposal is a fairly practical and easily implementable solution by all individuals let alone organizations.
• Trap: Escalation- If USA builds more arms, my country should do the same! I cannot give a simpler example for this trap. It creates an endless race, a reinforcing feedback loop. Solution: Whenever there is a reinforcing loop, the best way to stabilize it is to introduce a balancing feedback loop as discussed earlier. This can be practically implemented by avoiding getting into the race or if you do enter the arms race for example, using diplomatic channels to stop the escalation.
• Trap: When winners of a competition are systematically rewarded, we create monopolies by creating a reinforcing feedback loop. Solution: Antitrust laws. Policies that disincentivize the rich, incentivize new players and level the playing field.

A few more proposals are made in the book. However, it is evident that the suggestions are very generic and not novel solutions. Therefore, I did not discuss them in detail. Keeping performance standards absolute is something I have firmly started believing in. When it comes to escalation or creation of monopolies, the situation is so nuanced that the solutions proposed fail to consider the complexities that exist in reality. Despite the presence of the Sherman Act and the Clayton (Anti-trust laws in the US which were the 1st anti-trust laws post the industrial revolution), we have seen the rise of the BIG-5 tech companies (Amazon, Apple, Alphabet, Microsoft and Facebook) dominating so many market segments and having a considerable grip on the entire globe, far greater than individual governments in some cases which is scary. On the other hand, these organizations have made things so easy for consumers that imagining a life without them is not possible which puts us in a strange position (Breaking up these almost monopolies is something that happen sooner rather than later in my opinion!). You have similar examples in India (Reliance Industries), China (Tencent, Alibaba etc. However, the situation in China is very different since the Communist Party has a considerable grip on these companies as is evident from Ant Group’s IPO postponement). With regards to fair competition, the competition commission of the European Union seems to be reasonable more active and are more frequent in monitoring anti-competitive practices and have penalized the tech giants on multiple occasions although the fines paid are only pennies when compared with their revenues. However, as discussed, systems are resilient and self-organizing and I do expect that the markets will fight back some way or the other (assisted by regulations of course). This is an area which fascinates me and you can expect a detailed exploration at a later stage.

To conclude, I felt that this book was a nice primer on systems theory. The books presents some very powerful ideas and I appreciate the author for these efforts. Some of the concepts are presented very well with easily understandable examples. However, some systems are inherently very complex to model given the multitude of variables involved, the approach is not easy to implement. Additionally, there is no discussion on discrete events or black swans which can have a profound influence on a system. As is the case with all books and individuals, there are some positives and some negatives!

Rephrasing one of her statements from the book: “The world is complex, messy, dynamic and turbulent. It self-organizes and evolves. That’s what makes the world interesting, that’s what makes it beautiful and that’s what makes it work”. So, let’s celebrate this complexity, keep learning and enjoy the world we live in!