Unpredictable Patterns #91: The city as software and problem solving organisation
Organisation of groups and problem types, the age of city-sized AI:s and why we might want to bet more on cities in the future, even though they have been hyped for so very long.
Dear reader,
Moving the London has been eye-opening in many ways. One of the things that I have really enjoyed is living in a very large city. London is an organic, living thing, always changing and exploring the possible futures. It has made it clear to me how central cities are to our lives, and made me wonder if we should not expect cities to become even more important in the age of AI.
Organising ourselves
Different problems require different group organisations - the size, structure and interactions of a group can be matched to a problem in different ways. A very simple example of this is a soccer team: they are organised, sized and their interactions designed to solve the problem of winning a soccer game. Another example is an orchestra, often matched to the piece they need to perform - with varying skills (instruments) and size (from chamber music to symphonies) depending on the nature of the problem they are solving.
The same principle applies for solving more multi-faceted problems, and even companies and other organisations should be designed to match the problem they are trying to solve, but here we run into problems: many organisations are set up to not just solve an existing problem, but to also be able to take on new problems over time - and growth has become a key component to such organisations.
The modern company is measured, in many ways, on its ability to chart a path from the simple problems it solves when it is small to how well it can solve ever larger challenges in different ways. This is not a necessary state of affairs - it is easy to imagine a world in which a company sets out to solve a particular problem and never grows beyond the size needed to solve that problem, but that would seem to limit its economic growth over time, and so we resist the idea.
Does this mean that we should expect larger organisations to seek ever larger problems that match their size? Not necessarily: what seems to happen quite often is that an organisation will take on many more smaller challenges, under the assumption that there is something large that connects these smaller challenges into a coherent whole. Companies thus expand horizontally or vertically in markets depending on the logic they apply top their search through the space of possible problems.
The challenge then becomes to ensure that the subsets of the large organisation really are efficient (this is what has led to rules like the two pizza team rule1) and that the sum total of problems make sense in some way and reinforce the whole. Ideally the smaller problems could be fractal in relationship to the larger one, so that the organisations are solving variants of the same problem in different contexts and across different domains and learns more and more about the underlying structure of the fractal problem space.
This principle also applies to solving larger and more complex problems, like how we adapt to evolutionary pressures.
When biologist E.O. Wilson argued that we are an eusocial species he noted that we care about other people’s children (collaborative brood care) and divide labour much as other such eusocial species do.2 But what we find in other eusocial species is also that they organise in specific ways. The ants organise in the anthill, the bees in their hive and so on. It is not farfetched, then to ask if cities are our natural form of organisation.
There are many different theories about why we organise into cities. Some argue that it had to do with taking up agriculture and so benefiting from staying in one place, others argue that it had to do with the defensive advantages that a city could bring and yet a third explanation could be largely ritual - that we can believe more efficiently in a city than we can if we are a scattered tribe, and as our need for religion grows stronger, our tendency to seek to organise ourselves for it will grow as well.
In all cases, the city is a solution to a problem of some kind, and seeing that we can also play with the idea that the city is software implemented in buildings and roads, in shops and connected with other cities through trade. In this metaphor, the city is not the buildings any more than your word processor is your computer, but a complex set of algorithms executed on a clunky computation device put together with steel, concrete and humans.
This view of cities implies something else that is quite interesting to explore as well — and that is that maybe we should think about our civilisation less as a series of nation states and international organisations, and more as a complex network of cities. Visualisations of human interaction on the Internet will mirror the underlying structure of cities, and our technology and infrastructure also builds on the city as the key component.
This is not a new or revolutionary insight - in fact, it has been hyped for more than 30 years now. There are even political initiatives around cities, like the C40 - the 40 largest cities in collaboration, that have been founded on the idea that cities together can help solve global challenges. The C40 is currently focused on climate change challenges, realising that climate change also is a problem that can be understood as strongly connected with the city network that humanity has built.
If there is a software stack here, it is the city, and then individual human beings running their own lives on the city software, making decisions and exploring the world - with companies as special problem solving applications.
This leads to an intriguing possibility, built on a very simple question: how large is an artificial intelligence? Often, in fiction, the answer is that it is a robot (and tries to kill us) or it is a universal global system (and tries to kill us). This view - that an AI is either very big or very small - seems unimaginative. What if the optimal size of cognitive infrastructure also mirrors the city network that we are a part of? What if we should build and prepare for the age of city-sized artificial intelligence?
Admittedly, this is a fanciful idea, but it is worth thinking through, at least if we believe that there will, indeed, exist cities in our long future and that they still will be central to our human experience and how we organise ourselves.
Now, we do not need to think that future artificial intelligence will only be city-sized, that seems overly simplistic. But we can imagine a world in which individuals have AI-tools of different kinds that all run on and interface with a city-sized AI. The reasons for such an infrastructure are many:
The city sized AI will allow for some control over bandwidth, redundancy, security - securing a global AI would be tricky at best, impossible at worst.
A large part of our cognition, economic activity, public services and our tasks are related in some way to the city we live in - for one part of the future AI-layer to relate to that fact seems not entirely crazy.
A city-sized AI might find ways to optimise complex networks with long enough data series, and for it co-evolve with a data archipelago that is focused on the city would make sense. It would be a bit like search and search space growing together and co-adapting in a red queen problem.
A city sized AI could build a deliberative supportive layer that would help get the politics of the city right, and perhaps even help birth a re-commitment to our common polity even as nation states slowly fade into gridlock and populism. A city may be an evolved answer to the question of what size is best to solve the problems of the polity.
And so on. I am sure that you can come up with more reasons for why ensuring that our future cognitive infrastructure is modelled on cities, rather than on global systems. This world would in many ways be a much more natural state of affairs for us - for the longest time humanity was principally organised in cities - and so would not be an anomaly, as much as a closing of the short and troubled period during which we organised our selves in nation states.
And at the heart of the question is which organisational collaborative group can be predicted to best solve a basket of diverse problems - this is a question we should ask more often.
City skepticism
If cities are so great, why do we still have nation states? That question, or versions of it, has emerged as a very reasonable counterpoint to the city-utopianism that sometimes bubbles up in this discussion. There is, these critics continue, need for global coordination and national coordination for a large number of problems - including national defence and climate change.
The city advocate need not disagree with this at all, but can note that cities can coordinate quite well - and even better now that the coordination costs can be reduced with new coordination technologies. The challenge is to ensure that collaborative deliberation technologies are evolved that can work across the city network that we have built over time.
But there is something to the city-skepticism. Nation states have been declared dead many, many times - but seem to be soldiering on quite nicely. Why is that? If cities really are our natural form of collaboration and organisation - why don’t we see more change driven by cities?
Well, we could ask if we see more change driven by nation states. It is true that nation states meet in the United Nations, and that they seek large treaties and international agreements - but how effective have these mechanisms been for international political action? When we tout the effectiveness of nation states, we need to be clear about what it is that we are measuring.
There may be other explanations hiding here. One is that the nature of war could have shifted our organisation into nation states. The kinds of weapons we have are weapons of a certain resolution and granularity. Modern weapons are expensive difficult to imagine in some kind of city army. The idea that one city would threaten another with nuclear weapons is silly, especially if the cities are close - since that kind of weapon would be mutually destructive (not necessarily an argument against organising in cities!). The nation state may simply be a better equilibrium for the organisation of military force, than a city would be. The logic of military force is a logic of empires - where the size of the army and the projected force of its weapons wins over any smaller organisational model.
Rome may have been organised around cities, but its army was the army of an empire.
If this is true - that the drift away from city networks largely had to do with the organisation of violence - then maybe there is another insight in this thought: maybe a more peaceful world is possible only if we re-organise in smaller polities?
Surely this is naive, the critics say. What you are imagining is some kind of communitarian ideal in which small communities connect and trade - a return to tribalism. Why would that not lead to many more wars? The amassing of force and military might is actually the driver of the relative reduction in wars that we have seen over the last hundred years. Maybe - but at the same time the destructiveness of those wars seem to have increased, and the emergence of nuclear weapons that can destroy the entire planet seems to be a pretty big negative externality from allowing ourselves to organise in ever larger political spheres.
A more reasonable counter-argument is that there is no path there from here - in a world that has evolved nation states we have no way back to a city -based network with a weak federal infrastructure. This may well be true, but that is why it is so intriguing to think about what it would mean if we built new technological infrastructure to mirror that city network and built our artificial intelligence systems in - at least partly - city sized chunks.
Cities and companies
Geoffrey West analyses the relationship between different kinds of human organisation in his excellent book Scale.3 He notes that companies tend to die young and that cities tend to (almost) live forever. West argues that while it is easy to kill a corporation, it is surprisingly hard to kill a city.
We can pause there and make an interesting point about the way we typically see the future. In science fiction it is very common to envision a future in which mega corporations take over and all politics is organised in different company structures. This is such a common theme that it has almost been accepted as a “future truth” - a given in any discussion about the future.
There is, at the same time, suspiciously few science fiction stories in which power and political organisation is completely done in cities. Why is that? What are we missing? The answer is perhaps that the power of the short lived company is much more focused and direct than the long term power of a city over the world, so we over-index on the company and forget the city.
The one organisation built around a city - the Vatican - is the shining exception. It would be surprising if we did not see more corporations think actively about how to build themselves into city infrastructure in the future (assuming that we can call the church a company or corporation - and I mean no offense here at all, of course).
West also suggests that we need a predictive model that is built on how cities act and plan, a way to look at the long future through a city lens. This is, again, a very interesting thought - and one that suggests that maybe cities can become more relevant predictive elements in future scenarios than we think - especially if we start to design technology at city-levels (as in ensuring that every city is collecting and curating data about itself in a highly structured manner — what we need at the outset is city-sized bureaus of statistics).
The difference between companies and cities is multi-faceted - but the key to understanding why they are problematic is that they are sub-linear. Stewart Brand writes, about West:4
Are corporations more like animals or more like cities? They want to be like cities, with ever increasing productivity as they grow and potentially unbounded lifespans. Unfortunately, West et al.'s research on 22,000 companies shows that as they increase in size from 100 to 1,000,000 employees, their net income and assets (and 23 other metrics) per person increase only at a 4/5 ratio. Like animals and cities they do grow more efficient with size, but unlike cities, their innovation cannot keep pace as their systems gradually decay, requiring ever more costly repair until a fluctuation sinks them. Like animals, companies are sublinear and doomed to die.
This insight, that companies scale badly and cities scale well is also interesting when we think about the age of city-sized AIs: the way a cognitive augmentation of a city could work super-linearly suggests that it would be an excellent investment!
But there are intriguing questions here too: why does most innovation happen in companies, if cities are so good? Well, the answer is probably that the city is a set of loose connections and a network designed to allow for other connections (such as those that help us form companies) and the company is tightly connected - but maybe the idea, then, should be to make companies align much more with projects? Today, companies are formed as projects, but they slowly become systems. This is the truth expressed in the fact that they scale sub-linearly (they become worse with size) - and maybe companies should only be allowed a certain life time?
Imagine a world in which corporate law allowed for limited liability only for a limited time. So you would end up with a company that ran for 40 years, but after that had to be dissolved. Would that increase efficiency in the economy or create huge waste? How would a company behave if it knew that it would dissolve in 40 years? Would innovation slow down or speed up?
The invention of the limited liability company has been heralded as perhaps one of the greatest innovations in recent history - and that is probably right - but that does not mean that we could not change it. Allowing for a maximum length might be a simple thing - and maybe no company should be allowed to live for longer than its founder?
It would put a sharper edge on the distinction between corporations and cities: the first can live forever, the second only for a few decades. And it would not be terribly cruel either: most companies, after all survive for fewer than 10 years.
So what?
Our thought experiment here has been built on the idea that cities are - on some way - a more natural and resilient, and perhaps even efficient, way to organise human civilisation. One possible conclusion from that was that maybe we should mirror the city structure and network in rolling out future innovations like artificial intelligence. If this thought experiment is right, there are a few predictions we should be comfortable making:
Political talent is likely to flow to cities rather than national parties and parliaments. This will be a problem because there is much still decided on the national stage, and this “level brain drain” may open opportunities for demagogues in national discussions. At the same time there is an opportunity to re-invent democracy city by city - through citizens.
City-sized innovation will become more popular: from urban transportation to gig work (often organised in cities) we will see more innovation on a city basis than nationally. Successful venture capitalism will start to look at how a business model scales on the city network, and city policy work will become more important.
City collaborations may become more and more common. Bilateral agreements and collaborative projects between cities may come to be popular ways to deepen the city economy.
Urbanisation will continue and not be broken by remote work - because cities are still the most functional unit of learning and problem solving for a large set of challenges that we face.
Even if the rhetoric around cities has been over-heated, it is interesting to think about what they will look like in the next decades, and especially with artificial intelligence applications in mind.
As always, thanks for reading,
Nicklas
This rule has been attributed to Jeff Bezos and other entrepreneurs and states that no team should be bigger than to be able to sustain itself on two pizzas. This, of course, depends on the size of the pizza.
See e.g. Wilson EO Kaspari DC. Genesis : The Deep Origin of Societies. London: Penguin Books; 2020.
See West GB. Scale : The Universal Laws of Growth Innovation Sustainability and the Pace of Life in Organisms Cities Economies and Companies. New York: Penguin Press; 2017.
From the Long Now presentation West made and Brand wrote up.