May 9, 2020
Networks, Goldilocks and the Future of Cities in the 2020s
One good thing about crises is that they cause us to collectively stop and ponder how, where and with whom we want to spend our time. Most of us live in or very near large cities. Why? The point of a city is to effectively produce and distribute goods and services, to increase the well-being of people. When most goods were delivered by ship and delivered to the port it made sense to centrally around those waterways to shorten distribution. When entertainment was the local theater, it was crucial to live nearby. In the world of Amazon and Netflix, both goods and services are delivered to us. So, do we need cities? Yes, but three things are changing as a result of Covid:
· One, the seal has been broken on remote work. People have now signed up for Zoom, bought the new AirPods and are well up the learning curve. Many won’t want to go back.
· Two, the attraction base of many cities is going to be far lower. Many restaurants, bars and theaters unfortunately will not be able to open back up.
· Three, there is a more acute understanding of the health costs of congestion and density. Cities require density to work, and such things as a popular transit system which have historically been put in the “pro” column might be less popular today.
All three decrease the attraction of city living. There will be secondary effects (lower prices bringing different users, etc.) but for now we can assume that big cities will lose people. What comes next for cities?
Let’s think of cities in terms of networks. Network systems come in three forms: centralized, decentralized and distributed.
Centralized networks are simple to understand and execute, with a single central node and lots of client nodes. Centralized networks struggle to scale after a certain limit and bottlenecks appear when traffic spikes. They are hugely dependent upon connectivity and risk abrupt failure of the entire system. This is the way we have built cities over the last 50 years.
The second network type is a decentralized system, every node makes its own decision. Bitcoin is the most well-known current example of a decentralized system. Scaling is very possible by adding more nodes and there is a low risk of bottlenecks.
The final network type is a distributed system, where each node makes its own decisions. Google is a well-known distributed system.
It makes sense for early networks to be centralized, as central server was a scarce resource and capabilities were more limited. As technology has evolved, the attractiveness of both decentralized and distributed systems has grown.
Similarly, for cities. The clean delineation between urban and suburban is a modern invention, reflecting a desire for top-down execution and accommodation of the automobile culture over the past 50 years. There is nothing magical about it. Like with all centralized systems, big cities requiring millions of commuters break down, lack functional resiliency and suffer bottlenecks at small changes in capacity (try being in traffic in a major city when two lanes are shut down). The outcome of these systems are an economy which has seen a further concentration of wealth into a small handful of cities, each of which has gotten more expensive, more congested and more unequal. All of these factors point towards systemic fragility.
A fully distributed system of cities would reflect one in which everyone simply works from home. These workers would not require physical clustering and all office space would be obsolete. In this extreme there would be little congestion but also far less dynamism. Larger nodes create a critical mass of energy that are very valuable. A fully distributed system won’t happen but Covid has nudged society in this direction.
The final system is a decentralization of cities. A decentralized system is robust and antifragile, with many points of redundancy. In this world, there are not just four or five dominant cities, there are 20+ great cities, each with their own distinct strengths and weaknesses and poly-centric nodes of activity and density around them. These cities have lower levels of stress on the system than in the centralized model and the close proximity between nodes allows for more continuity and more diversity.
Impacts on real estate development
The previous waves of development in the U.S. have been “further out” (in the 1980s, 1990s and especially housing in the early 2000s) and higher up (taller urban skyscrapers, especially residential in the last decade in he 2000s). The next wave should be neither, instead it should be “tearing down”. There are millions of square feet of existing real estate that was on a slow march towards death. Covid just put most of it on a conveyor built. Much of this dead real estate is in places that might not work well in the centralized world but will be great places to live and work in the decentralized world in which we are heading.
To offer an example, a 30 second Google search leads me to the Hi-Lake Shopping Center in Minneapolis. Unfortunately, it looks like a pretty typical center in the current world: a Dollar Store and a Chinese buffet, and empty. With a walk score of 84, it offers lots of access to everyday needs, and is across the street from the train stop. Without knowing the details of this property, sites like these are the development sites of the future. Projects in areas like these — rather than immediate downtown city center or far out suburbs — strengthen the strength of the signal in the network.
There are thousands of these centers in cities across the U.S. and many of them are no longer economically viable. They should be turned into both housing and small office space, within communities with existing amenities and transit. No need to build new roads and substantial infrastructure, and an opportunity to help people and create more tax revenue. This is the opportunity of the 2020s for both cities and developers.
