On Progress

German Precision, Japanese Work Culture, American Patriotism, Middle Eastern Family Values, or Scandinavian Liberalism are stereotypes associated with a few of the so called first world countries. From such stereotypes one can derive societal properties and assign to them positive values due to the relative status that first world countries enjoy. But reflecting on a consistent set of “good“ societal properties in an absolute sense, a question arises about what actually the basic units of society are.

It is possible to imagine a hierarchy that structures societal properties into related classes at multiple levels. It is relatively easy to think of properties and their relations on the level of individuals, family and friends, institutions and communities, and (national) policies. It seems similarly easy to subjectively assign positive or negative values to the different properties. Obviously, a society of truthful people that respect their elderly while arriving at work on time would generally lead to a better overall society than one consisting of criminal insurgents, right? One could imagine an exercise in which members of first world countries – possibly experts on the matter – come up with such a hierarchy of properties and assign values for everyone to adopt. After weighting all their inputs, you would get guidelines for measures to be taken, to be enforced. So, why are there still people who are struggling in this world?

Three possibly related problems come to mind. The first is a pragmatic one: people in power might not want to adopt a better models of society to protect their power. The second one is fundamental: as society is a complex system and there is an inherent danger in assuming simplicity when mapping out complex systems. Often, the result is a combinatorial explosion because of feedback loops and external influences. Consequently, the resulting model would be a mere subset of the system under study. It is arrogant to assume that such an arbitrary model can be pushed onto a system without resistance, thus resulting in unforeseen side effects. Likewise, it seems presumptuous to define and enforce a model of society made up from a snapshot of subjective properties. After all, a society of criminal insurgents would have been the best option in the third reich. The issue is the lack of an absolute authority for validating different models of society, which would also have to consider the current and goal state of society. The third problem is that society is often confronted with novel problems for which finding a solution based on historical data might not be a valid approach.

Nevertheless, there seems to be a contemporary notion of societal progress that – for all intents and purposes – is considered desirable. Regarding this notion, it is interesting to consider that throughout human history progress was not always conceived as something that was possible as societies plateaued. In ancient times, progress did not exists, as the timeframe in which things changed was usually greater than a human lifespan. An individual could not imagine a future different from the past. Heraclitus famously said something along the lines of: it is not possible to step twice into the same river. Thereby, he acknowledges the constant change of things but also remarks that things stay the same by changing. For most of us today, it is the case that we watch as bystanders while change on a societal level happens. Some of us are more actively engaged in that change than others. However, there is a lack of guidance that could be used to accelerate or steer societal change in the “right“ direction. How can we avoid progress that leads to a dystopian society? We look at models that are sold to us as being worth striving for. Thereby, one problem is that some potentially good models do not seem trustworthy, as people applying a hive mindset are more likely to continue utilizing an apparently unworkable paradigm until a better paradigm is commonly accepted. Another similar problem is that we are blind in our intuition when evaluating models and we perceive progress in the frame of the existing social environment.

Throughout time, a set of societies could rely on the struggle between them to reflect on success and failure. As we approach a global society and differences fade, the danger is that our societal model reaches a stable (locally optimal) plateau while better alternatives are not pursued. Even if a hypothetically improved societal model could be conceived theoretically, there is no simple way of telling how it would behave in reality. Consequently, conservation wins: never change a running system, as the alternative is to follow a blind leader who sometimes is lucky in his reasoning and sometimes is not. This is a never before seen challenge in the age of a globalized society.

However, the system of society is subject to external influences and internal permutations. Corresponding progress happen as a systemic feature. Systemic problems can emerge from this approach. There is a real possibility that an uncontrolled society might destroy itself like on planet Krypton. To avoid pitfalls in societal development, let’s consider the following hypothesis: Societal progress can be purposely designed in an efficient and secure way. But as previously noted, a top-down approach where an elite class of leaders enforces their simplified models of visionary societies very rarely succeeds. Combining this with a bottom-up approach where leaders are democratically elected by the people helps, but is not a solution either. It is simply too hard to model the future behavior of society, yet alone to optimize it. The typical logical and emotional reasoning mechanisms based on these models fail to accurately predict the actual outcome when a model is enforced in reality. However, some lingering societal models still maintain their destructive potentials. Without a controller offering guidance, society changes on its own in a systemic way and might follow a possibly chaotic path that can lead to its own destruction.

The need for a controller acting on society as part of society is inevitable. However, to avoid the failure of model-based reasoning that is subject to the combinatorial explosion, the proposal here is an alternative mode of reasoning that utilizes the engine of complexity: evolution. Instead of building a controller for society that arbitrarily compares snapshots of various societal elements taken at a given point in time (which are a simplification of reality that are not inherently suitable to enable future predictions), the idea is to build a controller that fundamentally utilizes the same basic instinctual goals of all living things: reproduction and survival.

The goal of this controller is to make progress towards a future grand design which might become necessary due to external influences from the systems surrounding society, e.g. an extinction level event. But instead of working only with a model of society, the controller also interacts with the real thing. This is the idea of generative design. In analogy to evolution, generative design can be understood in two phases: exploration and exploitation. In the exploration phase, a first model of society is built. This can be a model as previously mentioned with the addition of different levels of goal abstraction. The model is a first prototype relating abstract goals for society to societal properties at different levels and their values. The exploration phase then applies random mutations to the model and stores the modified models. The result is a set of models for which no-one can really predict their behavior in reality. In the next phase, the exploitation phase, simulation of the models is performed. The simulation results are then compared against previously defined goal parameters, e.g., suitable goals to prevent an extinction level event. Thereby, the nature of simulation is crucial. Some randomly mutated models will be total garbage and not even pass formal verification. Others can be sorted out by model based reasoning. But for models where model based reasoning reaches its limits, a proof of concept simulation in the real world can be performed. Illustrating the last point, it would be possible to set aside a representing percentage of the population somewhere for testing a societal theory. This utilization on the proof of concept idea is not new in today’s world. For example, India and Finnland defined dedicated areas to test different realizations of basic income. 

The benefit is that generative design is a progress rather than a model definition. That progress is also able to adapt to new requirements in an agile manner. The strength of the approach lies in a combination of systemic factors and randomness through the two phases of generative design, which utilizes the conflict between thesis and antithesis in constructing societal progress. The advantage over a purely model base approach is that a large amount of small but scaleable changes that are are quickly evaluated in a grounded manner and in parallel. If set up correctly, this is a possible approach to deal with the combinatorial explosion while making progress towards a global optimum. 

To set up a system for generative design, its is crucial to reuse the knowledge we have. The initial model defines a structure that allows for focussed exploration. A solid core of this model can be defined to start exploration from. Depending on the scale of random changes to the model, societal progress can be engineered at different speeds. However, such a solid core also brings the danger of not changing elements of society due to our intuition which might be based on wrong beliefs. It is subject to the paradox of efficiency in which stepwise refinements and improvements offer only diminishing returns. Rather, the alternative to optimized inefficiency is a comprehensive redesign, but this would lead to great disruptions. Over time, the system can optimize itself as the mechanism of generative design can be a part of the model that is refined through generative design – like in nature where living things evolved to mutate faster or slower.

The bottom line is, society should be open for these proof of concept simulation runs. This would imply a higher diversification and the possibility that experiments fail. However, the price we will pay without actively testing new ideas for society will potentially be much higher. There is the danger of course that random mutation bring destructive features that, when tested locally, could push a global society as a whole to its limits. Testing them would be a bad idea. Not doing anything would be worse. The solution is to combine generative design as tight as possible with human knowledge and expertise, at least in the beginning. The current state of society could be a solid core and only small changes could be allowed, thus slowly converging to a local optimum in a semi-supervises manner. The evolutionary proof of concept results will be the examples that people & leaders understand and willingly adopt, which is the mechanism to drive a future global society society out of local plateaus. 

Previously such experiments would have required institutions like governments and separated geographic areas. Those of us who have a family with migration background might have heard similar stories, Chinatown is everywhere. But in todays digitalized world this should no longer be necessary. What separates a group of people from others might be virtual. As a consequence, the cost of experimentation goes down. We should be much more open than ever to experiment. A lot of the values stopping us from doing so are based in a long gone reality. But imagining a world that is able to escape the local optimum is a really tough thing to do.