Complex Systems — Part 2: Managing complexity with bottom-up solutions
In Part 1 we saw the shortcomings of hierarchical decision-making and control in complex systems. In this second part of the mini-series, we look at three solutions that have been proven to effectively manage such systems. They result from a very radical shift in thinking: Namely, from top-down control and decision-making to bottom-up co-ordination and co-operation, which ultimately need a very important component to be successful:
You.
Do you want to learn about these powers to change the world to the better? Then let’s start.
Solution 1: Collective Intelligence for decision making
The first solution caught the eye of the mathematician Galton in 1906 at a fair where participants had to guess the weight of an ox. He found that although no one had guessed the exact weight, the average of all the guesses gave almost exactly the weight of the ox. This phenomenon is repeatable and is called the wisdom of crowds (when there are no interactions) or collective intelligence (when there are interactions between actors).
It states that no individual within a group needs to know the truth, but that it can emerge through the collective interactions between those individuals. In other words, the knowledge of truth lies in the connections between the nodes of a complex network (see Part 1).
Collective intelligence can be used for more complex challenges then the identification of an ox weight.
“A group of 3,000 ordinary citizens, armed with nothing more than an Internet connection, is often making better forecasts of global events than CIA analysts.” — Alix Spiegel
For example, in the Good Judgement Project, ordinary citizens were asked about future world events, e.g. “When will North Korea launch the next missile”, “How will refugee movements in Europe develop”, etc. It turned out that these citizens often predict future world conditions better than CIA analysts who had access to classified information.
This result is significant.
Some argue that this is why democracies work: By considering large parts of their population in the decision making, they are better equipped to sense the future than dictatorial systems and thus are better positioned to ignite the necessary changes to adopt to a changing environment.
An interesting branch of research in this direction are prediction markets.
So how does collective intelligence work? It can actually emerge in a complex system if the following three steps are guaranteed:
- Free access to a transparent and unbiased information source is essential as a starting point.
- Then, independent opinion formation/ solution exploration is required In particular, innovation needs a cultural setting that allows to experiment and to make mistakes in a protected environment, traditionally/ often called “private sphere”.
- Free information exchange that feeds back into the information source.
Interestingly, it seems that democracies over the world emphasis different aspect of this collective intelligence. For example, the U.S. emphasizes the freedom of speech aspect, while the German-speaking countries favor privacy (which is also enshrined in their “constitution”) and the Scandinavian countries emphasize transparent access to information (e.g., to the point of disclosing everyone else’s tax payments (which can actually make Germans cringe :) )).
New benefits can result from combining these three elements.
Solution 2: Digital democracy for decision making
Digital democracy is currently discussed as an upgrade to democratic systems to better cope with the challenges arising from a complex world.
If implemented well, it uses society’s collective intelligence and combines it with digital tools such that the best ideas are distilled, combined and implemented. Introductions to the concept can be found here, here and here. Concepts utilized are for instance citizen councils (e.g. citizens’ assembly), consul platforms (e.g. vTaiwan or pol.is), and participatory budgeting.
In general, in order to make democracy work in a digital world, two further steps are introduced to the previous three of the collective intelligence:
- Collective Intelligence: People share information, ideas and solutions (output of the three collective intelligence steps)
- Deliberation: Various solutions are combined in an innovative way by the means of a deliberative process.
- Voting: If the the deliberation process does not result in the convergence of a great majority, the people affected vote to determine the best solution. Here, majority votes are often sub-optimal as it can result in the “tyranny of the majority” which can lead to unpredictable reactions from various citizen groups (e.g. as seen with Brexit). This is most likely due to the fact that for complex and high-stake questions, a majority vote often is not perceived as legitimate by the voters.
Hence alternative forms of voting are explored, such as quadratic voting or the modified Borda count.
In summary, collective intelligence utilized in a digital democracy can lead to the identification of appropriate ideas worthy of implementation, and in this way provide a solution to the normative challenge outlined in Part 1.
Nevertheless, the question remains how their successful implementation should be executed/ controlled for. In particular, we have seen that hierarchical control mechanisms often fail in a complex world, even if the desired state is clearly defined (e.g. CO2 reduction).
Solution 3: Self-organization for control
Complex systems in nature, such as a flock of birds or the human body, tend to self-organize, resulting in decentralized systems that are both efficient and resilient. A flock of birds, for example, efficiently supports both (i) protection from predators and (ii) localization of food resources, while at the same time being (iii) inherently voluntary (no bird is forced to join it) and also (iv) resilient: removing a single bird from the flock would have little negative impact, as no single bird has control over the entire flock.
The flock is rather controlled by local feedback mechanisms: If a bird is flying too close to another bird, then it flies away, if it is too far away, it flies back. By such simple mechanisms the flock moves through time and space and supports its members. Actually these simple mechanisms illustrate the core concept of self-organization: Feedback is required for a complex system to self-organize.
Self-organizing systems are everywhere, also our human body and its immune system are examples of such self-organizing systems.
The question arises: Can we take this success mechanism from nature and apply it to man-made infrastructure and society? The simple answer is “yes”, as has been shown, for example, in the context of traffic light control systems:
Traditionally, traffic light control systems in cities are controlled top-down: A central traffic authority plans how the traffic lights in a city are switched, e.g. by prioritizing different traffic flows depending on the time of day (morning/evening traffic) or special events (e.g. football matches). The result of this planning can be observed by people all over the world every day: Traffic jams. Although, of course, not all congestion can be attributed to centralized control, the following evidence suggests that there are at least some major challenges associated with this type of control:
Namely, In order to mitigate traffic jams, researchers developed a traffic light control system that does not rely on a centralized traffic control authority, but utilizes self-organization. In this simple setup which has been deployed amongst others in the city of Lucerne (Switzerland), each traffic light is allowed to optimize locally the traffic light, only considering information (feedback) from its neighboring traffic lights. In particular, no traffic light or operator had the full overview of the system state, resp. could interfere in a local action. The finding of this experiment was not only that the average waiting time decreased for all traffic participants, but that also specific stakeholders (e.g. public transport) could be prioritized in a flexible manner.
This example demonstrates that self-organization can be utilized effectively in our technology-driven society.
So how does self-organization work? Actually, it is pretty simple:
Actors have to be able to act (freely/ autonomously). These actions change the environment which provide a (real-time) feedback to the actors on which they can improve their action. Self-organization does not work if either actors are hindered to act or feedback is too much delayed.
Conclusion and outlook
We have seen that our world is a complex system and that in such a system established hierarchical control and decision-making mechanisms fail (Part 1). The three solutions presented in this part of the mini-series together form a path to manage the challenges of our complex world more effectively, and thus provide to us an encouraging glimmer of hope but also a serious obligation:
A glimmer of hope because we are all needed for the emergence of collective intelligence, digital democracy and self-organization: Random individuals, just like you and me, are the most important core component of these mechanisms, not despite but precisely because of our individual biases, abilities, and (unique) perspectives. In particular, we know from complex systems theory that every small action can have profound effects on the system as a whole — and thus each and every one of us has the ability to change the world for the better.
Precisely for this reason, however, the knowledge of the power of the individual also gives rise to a serious obligation for each and every one of us to become actively involved — or, in other words, it is the responsibility to bring Imannuel Kant’s insight into the 21st century, namely to free ourselves from our self-inflicted immaturity.
In this context, I think, web3/ blockchain gained such a popularity because it (at least implicitly) utilizes successfully these bottom-up solutions by empowering the individual and in this way unlocking their creative potential and motivation to act. E.g, decentralized autonomous organizations (DAO) are instantiations of digital democracies that utilize collective intelligence to identify desired system states (e.g. as performed successfully in the MakerDAO or BanklessDAO) and then use their community to control for its implementation, often via token incentives. In general, these blockchain-based cryptoeconomic incentives in the form of tokens can function as real-time feedback enabling self-organization, as for instance studied in finance 4.0 (e.g., it can enhance the calibration of a complex systems).
In further blog posts, I will explore these concepts of DAOs and token incentives in more detail and try to underpin them with the theoretical foundations outlaid in this mini-series on complex systems.
Stay tuned!
This blog post has been strongly inspired by the work performed at the Chair of Computational Social Science at ETH Zurich.
Subsequent blog posts that elaborate on the concepts of this article and/or illustrate the mechanisms in practice:
