Check out the episode:
You can find the shownotes through this link.
Are you interested in how to utilise digital twins in human-centred urban planning?
Our debate today works with the article titled Cities aren’t rocket engines: The maturity of digital twins in human-centred urban planning from 2025, by Lucas van der Meer, Lukas Esterle, Mario Cools, and Martin Loidl, published in the International Journal of Digital Earth.
This is a great preparation to our next panel conversation with Micah Gaudet, Ana Maria Bustamante Durante and Balamurugan Soundararaj in episode 450 talking about the challenges of data and digital twin use in urban governance.
Since we are investigating the future of cities, I thought it would be interesting to see how the urban mechanism of data and urban vitalism of human experiences can be connected. This article advocates for a human-centred approach where digital tools serve as a lens to explore possibilities rather than an objective truth to dictate outcomes.
[intro music]
Welcome to today’s What is The Future For Cities podcast and its Research episode; my name is Fanni, and today we will introduce a research by summarising it. The episode really is just a short summary of the original investigation, and, in case it is interesting enough, I would encourage everyone to check out the whole documentation. This conversation was produced and generated with Notebook LM as two hosts dissecting the whole research.
[music]
Speaker 1: Back in 1970, when an oxygen tank exploded on the Apollo 13 spacecraft, NASA didn’t just sit around and guess how to bring those astronauts home.
Speaker 2: They relied on a simulator back on Earth.
Speaker 1: Exactly. It was a physical replica of the command module that tracked the ship’s condition in near real time, and that simulator allowed engineers to test these extreme survival scenarios on the ground-
Speaker 2: Before executing them in the actual cold vacuum of space.
Speaker 1: Yeah. It was a miraculous feat of engineering, but it worked because it was a closed, highly predictable system.
Speaker 2: Absolutely. I mean, every valve, every wire, every variable could be perfectly quantified and managed by the engineers in Houston.
Speaker 1: Like a living, breathing city.
Speaker 2: That is a massive leap.
Speaker 1: It is. But today we are looking at the rapidly expanding field of urban digital twins, often called UDTs. These are highly realistic, dynamic computer models meant to simulate our cities. And we’re grounding our conversation in a really provocative Target article titled Cities Aren’t Rocket Engines: The Maturity of Digital Twins in Human-Centered Urban Planning.
Speaker 2: It’s a fascinating paper, and the core disagreement we are tackling today is whether a city can actually be mathematically modelled, predicted, and optimized like one of those engineered machines.
Speaker 1: Or conversely, if cities are inherently living complex entities driven by intangible human forces that fundamentally defy pure quantification. Just to lay our cards on the table, I’ll be representing the urban mechanism perspective, arguing that cities function through complex, knowable cause and effect processes that we absolutely can objectively model and optimize.
Speaker 2: And I am taking the urban vitalism perspective, arguing that the true essence of a city relies on unpredictable, intangible relational forces that cannot be entirely quantified or algorithmically managed.
Speaker 1: So before we get into the heavy philosophical disagreements, I need you to ground this for anyone listening. When we say urban digital twin, what are we actually talking about mechanically?
Speaker 2: It gets thrown around a lot in tech circles. It’s not just a static 3D map or like a video game interface.
Speaker 1: So it’s not just a giant game of SimCity.
Speaker 2: No, not quite. The concept originated around 2002 with Michael Grieves, who used digital models for product life cycle management, tracking a manufactured car from its creation to its disposal. But an urban digital twin is much more dynamic. It relies on a continuous feedback loop between the physical city and the virtual model.
Speaker 1: Through real time data ingestion, right?
Speaker 2: Exactly. Think about the Internet of Things. You have physical sensors on garbage trucks, smart meters on power grids, cameras at traffic lights, and air quality monitors on street lamps.
Speaker 1: It’s everywhere.
Speaker 2: Everywhere. All of those physical sensors are constantly beaming data into a centralized virtual model. The digital twin ingests that data, processes it, and then supposedly gives urban planners this real time four-dimensional view of how the city is operating in that exact second.
Speaker 1: See, that is exactly why I find the mechanistic perspective so compelling. Because of modern data accessibility and sheer machine speed, we can finally construct what biologists call dynamic mechanistic explanations.
Speaker 2: But we aren’t talking about biology, we’re talking about urban planning
Speaker 1: Sure, but the principle is the same. We aren’t guessing anymore. We can track traffic flow, energy consumption, human movement. We can mathematically reassemble all of those systems to capture the emergent properties of the city. We finally have the scientific rigor to predict outcomes and optimize efficiency.
Speaker 2: Hold on. You can track energy grids easily enough. A kilowatt is a kilowatt. But human movement? You can’t just plug human spontaneity into a mathematical formula and expect a perfect result.
Speaker 1: I wouldn’t say perfect, but we can find the patterns.
Speaker 2: But this is where the title of Vandermeer’s paper hits the nail on the head. Cities aren’t rocket engines. Engineered systems like Apollo 13 are closed. Cities are open social systems.
Speaker 1: Social systems still follow rules, though. They aren’t pure chaos.
Speaker 2: They aren’t chaos, but they also aren’t equations waiting to be solved. If you look back to the early 20th century, the sociologist Georg Simmel wrote extensively about the urban experience. He argued that the very essence of a city is its unpredictability.
Speaker 1: The chaos of the crowd.
Speaker 2: Exactly. It’s the excitement, the friction, the fleetingness of life on a crowded sidewalk. That is what makes us feel alive. When you frame the good life in a city as this mathematical optimization problem, you are fundamentally misunderstanding what a city is. You’re reducing human beings to mere variables in a digital machine.
Speaker 1: I see why you think that, but I genuinely believe that is a mischaracterization of what modern urban modelling is trying to do. Dismissing the machine metaphor like that feels like you’re looking backward at someone like Le Corbusier. Le
Speaker 2: Corbusier is highly relevant here, though.
Speaker 1: But his 1933 Athens Charter was a disaster. He viewed the city as a rigidly segregated, deterministic machine. He wanted to zone everything perfectly. People live here, work here, play here.
Speaker 2: Reducing people to mere occupants of his grand design.
Speaker 1: Yes, and modern urban mechanism isn’t Le Corbusier’s crude clockwork. We’ve moved way past that.
Speaker 2: I’m sorry, but I just don’t buy that it’s all that different. The legacy of that rigid reductionism is absolutely still alive in how we develop these digital twins today. The paper points to recent maturity models published by researchers like Massoumi or Haradiri. Do you know how they define the highest, most mature level of a digital twin?
Speaker 1: They define it as achieving complete self-governance, don’t they?
Speaker 2: Yes, complete self-governance and self-management. The implicit belief embedded in the technology is that an algorithm can, and eventually should, autonomously operate a city.
Speaker 1: In certain operational aspects, sure.
Speaker 2: But that is Le Corbusier’s dream dressed up in modern Python code. It assumes that data is an objective source of absolute truth that can entirely supplant human wisdom and, frankly, political debate.
Speaker 1: That’s an interesting point, though I would frame it very differently. Modern mechanism doesn’t look at a city like a simple engine. It models complex ecosystems. Think about how modern biology models a rainforest.
Speaker 2: Okay, but a rainforest doesn’t have a zoning board
Speaker 1: Fair, but a biological model doesn’t assume the rainforest is a simple machine. It maps complex interacting networks. And yes, it is difficult to map a city, but to reject the attempt to quantify these systems entirely, I worry that is just a way of coddling our own ignorance.
Speaker 2: That’s a bit harsh, isn’t it?
Speaker 1: Well, the philosopher Henri Bergson famously said that the vital principle, this intangible vitalism you’re defending, is often just a label affixed to our ignorance. Mechanism invites us to conquer that ignorance. Aren’t we just using words like spontaneity and vitalism as an excuse to avoid doing the incredibly hard mathematical work of actually understanding how our cities function?
Speaker 2: Not at all. Acknowledging what cannot be quantified isn’t ignorance. It is intellectual honesty. When we pretend we can model everything, we fall right into the McNamara fallacy.
Speaker 1: Walk me through how you’re applying that here.
Speaker 2: So the McNamara fallacy is the danger of measuring only what is easy to measure and then completely forgetting that the intangibles even exist. Imagine a digital twin optimizing a neighbourhood purely for traffic flow and commute times.
Speaker 1: Because speed and volume are very easy to measure with cameras.
Speaker 2: Exactly. So the algorithm might recommend widening the roads and removing crosswalks. It mathematically optimizes the system, but in doing so, it completely destroys the walkability, the local commerce, and the vibrant community culture of that neighbourhood.
Speaker 1: Because culture couldn’t be quantified by a sensor on a street lamp?
Speaker 2: Right. The model succeeds mathematically but fails practically. It creates a feedback loop that merely reinforces the status quo. We risk committing what Alfred North Whitehead called the fallacy of misplaced concreteness.
Speaker 1: Mistaking the abstraction for reality.
Speaker 2: Yes. If we aren’t careful, we will start adapting the real world to fit the model rather than adapting the model to fit the real world.
Speaker 1: I fully acknowledge the danger of the McNamara fallacy. You cannot mistake the map for the territory. However, we cannot throw the baby out with the bathwater here. The mathematician Erika Thompson talks about the incredible utility of what she calls model land.
Speaker 2: But model land is purely theoretical.
Speaker 1: It is an abstraction, yes. Model-land is a virtual environment where all of your assumptions are taken to be true. But humans have never survived without abstraction. What if simulations in a digital twin extend our mental abilities the exact same way a crane or a forklift extends our physical abilities?
Speaker 2: But a forklift moves real objects. A model just moves assumptions.
Speaker 1: But those simulated movements save lives. They allow planners to test high-stakes interventions in a safe virtual environment before execution. If we wanna know how a new sea wall will redirect floodwaters during a hurricane, we need model-land.
Speaker 2: I don’t deny the utility of testing a sea wall.
Speaker 1: We need that highly structured, mechanistic virtual construct to process variables that are simply too vast for the human brain to calculate on a whiteboard.
Speaker 2: That is a compelling argument But have you considered what it takes to actually leave model land?
Thompson explicitly states that you cannot just pull data from model land and seamlessly apply it to the real world without what she calls a qualitative escape.
Speaker 1: Meaning a translation process.
Speaker 2: Meaning human judgment. You have to step outside the math. And that brings us directly to how Van der Merwe’s paper redefines maturity in these systems. True maturity in a digital twin is not algorithmic supremacy.
Speaker 1: It’s humility.
Speaker 2: Yes, humility. A mature digital twin must facilitate a dialectical process between the real and the virtual. It has to know its own limits.
Speaker 1: I am actually in total agreement that human translation is vital. We can’t just hook the UDT up to the city’s infrastructure and walk away. And the authors lay out seven characteristics of a truly mature digital twin. They say it must be adaptive, constructive, diverse, honest, humble, responsible, and transparent.
Speaker 2: Which are all deeply human traits, not mathematical ones.
Speaker 1: True, but they dictate how we use the math. We need the human in the loop. During information acquisition, citizens should actively contribute context rather than just being passively surveilled.
Speaker 2: Yes, shifting the goal from data-driven automation to data-informed planning.
Speaker 1: Exactly. But I must insist, to be effectively data-informed, you still need that highly structured, sophisticated digital twin operating at peak technological capacity underneath it all.
Speaker 2: We definitely find common ground there. The technology is phenomenal, provided it remains a tool and doesn’t become a master. But to get there, we absolutely must abandon the rhetoric of the digital twin being a perfect replica.
Speaker 1: Because the virtual city should be treated purely as a lens to explore complex issues, not an oracle.
Speaker 2: Precisely. If we want to use a UDT to adjust the timing of traffic lights when it suddenly starts raining, that’s perfectly fine. But for long-term human-centered urban planning, the intangible, vital aspects of human life must always override algorithmic optimization. The math has to bow to the human experience.
Speaker 1: And I will maintain that as our computing power grows, the gap between what we currently consider intangible and what we can accurately model will continue to shrink. But I think we are converging on a really crucial insight today.
Speaker 2: That models are lenses, not objective sources of absolute truth.
Speaker 1: Yes. We simply disagree on exactly how much of the human experience can ultimately be captured by those mathematical lenses.
Speaker 2: I think that’s a very fair assessment, and the authors actually leave us with a fantastic concept to guide us forward, a Hippocratic oath for modelers, adapted from Emanuel Derman and Paul Wilmot.
Speaker 1: First, do no harm with your algorithms.
Speaker 2: Exactly. The core vow is to never sacrifice reality for elegance. It forces the modeler to remember that digital twins are just abstractions. They are math, but the outputs of that math impact real human lives.
Speaker 1: It is a powerful reminder. The true test of our future cities won’t be how autonomously our algorithms can run them. The test will be whether our technological tools will support human agency or subtly replace it.
Speaker 2: Leaving the listener to decide for themselves where the limits of simulation truly lie.
Speaker 1: The next time you hear a tech CEO make a grand promise about building a flawless, perfect digital replica of the world, remember Apollo 13. A perfect mathematical simulation might save a spacecraft.
Speaker 2: But a living, breathing city is not a rocket engine.
[music]
What is the future for cities podcast?
Episode and transcript generated with Descript assistance (affiliate link).
What is The Future for Cities? Podcast 
Leave a comment