Patrik Schumacher Lecture @ KazGasa University in Almaty, Kazakhstan

Good to see you guys. Thank you, Almaty. I'm very happy to be here. Thanks for a wonderful hospitality. I feel very welcome here, and very happy to visit Kazakhstan, Almaty. I had a wonderful conference yesterday, and I hope I will come back to do some work here. So, I will talk about architecture for the 21st century. You said I'm influential; I'm not so sure. Maybe I'm notorious, but the influence is yet to materialize. It feels quite lonely sometimes in terms of the work we're doing. There's not enough emulation and dissemination, let's say.

In terms of the world of parametricism, it's recognizable and also internally very diverse in terms of the morphology, materials, patterning, and spatialities which we develop. It's rich and diverse, and I compare it to the endless forms of nature. So, I would say that within parametricism, the diversity and versatility of formal strategies you can deploy, and the way these forms are adaptive, is richer and more versatile than all the previous styles of architecture put together. Because sometimes I'm asked, 'Why do you want everybody to work within parametricism? Doesn't this become too much the same?' I said no. That's a real takeoff of further diversity and versatility, because the space of exploration for creative and unique solutions in parametricism is at least a thousand times larger than, for instance, modernism or any of the other retro styles we still work with.

So it's all about opening up, working together, and exploring together that new paradigm of architecture. I think then it will really take off, flourish, and show its versatility. At the moment, we're way too small a group, so we need everybody's help to develop this paradigm forward. These are some of the projects we've been putting out. I won't detail it, but I want to show you this: this is nearly 20 years ago. I came to Almaty, and we made some sketches for a major site right next to the Square of the Republic. So parametricism has been around; this was already over 10 years into the development of parametricism, and by now we are 30 years in.

This was too ambitious for the site. Then we had another go at this complex; it became a bit smaller, a mixed-use development. Anyway, it didn't happen, unfortunately. More or less a year later, at the same time, we made this proposal for the National Library of Kazakhstan in Astana. It encapsulates a beautiful garden and has a big atrium at the peak, as you can see here, and a wonderful set of spaces to communicate. This is our architecture where we love to layer spaces vertically and use full three-dimensionality to have intervisibility between all spaces to enhance communication.

Finally—maybe it's slightly more restrained—our first project in Kazakhstan is in Astana, also very close to the center: the National Bank of Kazakhstan. We're integrating that with the landscape. The facade is all limestone, but the main feature is this atrium which connects you to the exterior and allows all the departments and workers to intercommunicate visually to see what's going on at all times—panoramic elevators, various terraces, etc. It's under construction, and we're very proud to contribute to Kazakhstan with this first project.

Now, I also wanted to show—I'm not going to show the Baku project, I have it coming later in the lecture, which is close by in the region—this is a project we're doing in Tashkent, in fact New Tashkent in Uzbekistan. It's a cultural building and research conference center. The exterior expression is the structure, which is at the same time wind funnels for natural ventilation. So when the sun hits the surface, it warms up and pulls the air out and through. I think it's a wonderful place, also integrating with landscape design. This is the principle of the wind tower, and again, double- and triple-height spaces for visual communication inside and outside across the various levels. I think it's a wonderful project and, after Baku, maybe another major landmark for us in the region. I think it's quite inspiring.

We will talk a lot about tectonism, which is the latest stage of parametricism, but I thought I'd just introduce parametricism a little bit. This article was published in 2009, so it's been a while—16 years ago. The year before that, in 2008, I launched parametricism at the Venice Biennale with a conference and a programmatic article in the catalog. The year before, I started to talk about it in SmartGeometry conferences, etc. But at that time, the actual development of parametricism was already nearly 15 years in the works, but it didn't have a name. I thought after 15 years of coherent work, as it was spreading throughout researchers, universities, and lead schools around some key spaces, and as we'd been developing it very coherently at the AA in London and at the studio, this deserves a name, a program, a theoretical summary, and canonization—a bit like the International Style program and formalization in the early 1930s, after 10–12 years of modernism, formalized this style. So that's the beginning of giving it a label, a flag, a manifesto.

This idea of canonizing means to look back and see: what are the principles we've been adhering to, even subconsciously? What are we no longer doing in architecture? What are we avoiding, and what are we always promoting? This is what I call the negative and positive heuristics, or you can say the taboos and dogmas which canonize and define the style. At that time, I realized we are no longer doing rigid forms. We don't use cubes, cylinders, domes, pyramids, etc.—these simple compositional elements which Le Corbusier still thought were the main elements of architecture, where it was just about composing these. We realized we've never used those for the last 15 years, so something is happening.

Also, when you develop systems in modernism in particular, but even in classicism—if you look at the Gothic cathedral, each bay is repeated—repetition is a very profound and fundamental element of ordering architecture through Gothic, Renaissance, but also modernism. Simple repetition, which is extremely obvious in modernism, we stopped: never repeat an element. Also, after modernism, there was postmodernism, which was bringing disparate elements together in some kind of collage attitude. Charles Jencks developed the concept before he called it adhocism, this idea of collaging; deconstructivism as well, interlayering disparate elements that remain unrelated. There's this collage of unrelated elements happening mostly throughout the '80s and '90s. We also said this is taboo. If you bring two different elements together, we need to mediate. They need to resonate with each other; they need to blend from one into the other. We're never just composing elements. Once you bring an element to another element, something happens to both elements.

So, no collage of unrelated elements; that's taboo. What we realized is everything which has happened in architecture for thousands of years is taboo. We're doing things radically different, and we have to watch out that we don't fall back. So what are we doing now on the formal level? All forms we're using, instead of platonic primitives, are parametrically malleable. They have parameters, and they take any form depending on what they get close to or where they sit; they're differentiated primitives. When we do systems with multiple elements coming together, instead of repeating, there is gradual transformation from one element to another. They're differentiated systems.

What you find in nature offers lots of examples. A honeycomb, when you look closer, is not equal cells; it's densifying, radiating out. Or snake skin, whatever it is—it is a differentiated system. When we bring systems together, they start to be correlated, like elements together, correlated systems to correlated systems. These are the principles of parametricism. But this is just formal principles; that's the repertoire you work with. You realize that this needs to make sense with respect to the societal tasks and functionality demands on architecture.

There's a parallel: for instance, we don't do rigid forms, but also, in terms of programs, we don't think in terms of fixed stereotypes. A lot of people still do, but in a modern society where there's a lot of variation, diversity, and differentiation of social conditions, you don't repeat spaces. Every space is unique with respect to place and purpose. If you do several of the same, you want variety. Because there's no social homogenization, we don't have this kind of simple repetition, and we don't segregate zones in the city and the building. These are sociological and programmatic functional differences in contemporary society which match up here.

Positively, all functions are variable event scenarios. A conference is different from another; a classroom should be different from another because the class is different in terms of composition, agenda, everything; an apartment should be tailored and varied to the situation. All program domains are differentiated. If you have one category of program, it comes with a lot of variety. But most importantly, when you have different program types and systems internally differentiated, when they come together, they start to correlate and communicate with each other. All the social events communicate with each other and influence each other relative to where they are.

I come at the end to this: when we do a program, we should not just look at a schedule of accommodation, which is a series of items on a list you take for granted and translate into square meters. When we talk about program, we're actually simulating the events. We're bringing crowds and people into the spaces with their agendas and communicative relations, simulating them to see that the spaces adapt to their dynamic interactions. That's the contemporary world, and you can only do that if your architecture is parametric and the social scenarios are dynamic and evolving, as you'll see later how this starts to look.

These principles have been operating now—we've been working with these principles for more than 30 years—and parametricism has a series of subsidiary styles. Tectonism, which I will talk about, is the latest style. It's a bit like modernism started with white modernism, then became richer: it had brutalism, showing materiality and larger structures where the structure could not be hidden anymore; then it developed into metabolism, into high-tech. Modernism has a series of subsidiary styles, but the principles of modernism in a high-tech project are the same fundamental principles as the early modernist: separating functions, optimizing them with components, but then each zone has internal repetition. This separation and repetition is always happening in modernism through all of it, whether white modernism, brutalism, or high-tech.

Similarly, we have principles carrying through, but the development of parametricism became enriched with an increasing repertoire, increasing rationality, and increasing articulation. Starting with this: when we started parametricism, we looked at so-called folding in architecture. That was the first slogan, the first name of what I later called parametricism: folding. Why? The idea was that instead of dissecting space with walls or boxes within boxes, you can articulate zones by making pleats and folding a single surface, and this could be adaptive. There are many projects you see which are just about folding a surface—splicing, cutting, folding, pleating a surface, and generating architecture through this. No walls, no enclosures: an open folding surface.

A project like this, with a continuously looping, folding surface going around—that's the way you can create architecture. It's very open and fluid, using elements where the stairs become the lighting. The purpose of this is to have a very complex space, but not filled with a clutter of composed elements; it's just one flow of system so you can focus on the space and not be distracted with all sorts of elements and different colors irrelevant to the flow of the space. The space was more complex, but the articulation of architecture becomes much more simplified and fluid. With this continuous folding of surfaces, instead of adding a staircase and a handrail, the same surface does everything. Instead of adding ceiling lights, acoustic elements, or seats, it's just the rippling of a surface. That's foldism.

The next stage was: okay, we sometimes need to make enclosures. This is the idea of blobs, blobism. What's the advantage? Instead of having Room A next to Room B and each room is closed, or three rooms inside a larger room nesting—which was the only logic by which you could organize architecture until deconstructivism, where it started to change—what you can do here is create degrees of separating out within an overall space. You're in a larger space that has subsidiary parts with degrees of articulation: more separate, most separate, relatively quite separate, but still a continuous flow of space where they nearly fuse.

Suddenly you go from a set theory where you say separate, nestling, or maybe intersecting—which was only introduced in the 1980s through deconstructivism—and we're going beyond that. You can have fuzzy logic degrees of being in one space and another, and we can have architectural articulation of this. In blobism, you can see quite a few projects fuse, with an open flow of elements, negative and positive blobs. Blobism also applies to furniture; it applies to everything. The style of parametricism and its different subsidiary styles are universally applicable to the totality of the phenomenal world. You can do that in urbanism, architecture, interior design, product design, fashion design—everything. It's total.

As a discipline, I'm thinking very much like the Bauhaus: unifying across all design disciplines so that everything you ever touch, see, or interface with—including graphic design and interaction design, everything you encounter in your life—was designed by a colleague. As a design discipline, we have universal and exclusive competency with respect to everything you ever touch, including landscape, because we have landscape designers, and there will be parametricism across all of these domains. Since every designer is looking at the scene as totally connected world architecture and world design, when you make a strong statement and reference, it will change everything that happens in the world and recreate the totality of the phenomenal world like the Bauhaus. The Bauhaus in the early '20s made experiments on urbanism, architecture, interior design, product design, furniture design, and Gropius and Mies lived to see in the 1960s the whole world transformed in the image of their research and development, including the Soviet Union, Africa, India, everywhere on this whole planet. We have to do that again.

This is swarmism: we have surfaces, we have blobs, and now we have swarms of elements. It's very good for urbanism. A swarm is a formation, a whole that has many parts, but you don't see the individual parts; what you see is a global form which compresses, stretches, and transforms. These elements don't have to touch, but they are connected through computational logic. This is what is beautiful about this: you can take an existing urban fabric, connect it, continue it, and then dramatically change scale and openings while remaining coherent.

Currently, every development in the world over the last 40 years is totally incoherent. Have you ever seen a city that has any sense of identity, beauty, or coherency built in the last 40 years? No. I call this garbage-spill urbanization. That's what we say is taboo: the agglomeration of unrelated elements which came after modernism and is universal. The last beautiful urban spaces were done in modernism: White Tel Aviv, Brasília, Chandigarh, the Art Deco District in Miami. You can recognize these as beautiful. Since then, we only have visual chaos without exception on the whole planet, and that shouldn't continue. We know how to overcome this, but we need everybody to participate, just like everybody in the 1960s was participating in modernism. Everybody in the 2020s must participate in parametricism and tectonism, because everything else is retro. Most of our colleagues are designing today as they could have designed 80 years ago; there's no advance. Why do we have an academic discipline and universities if we are just doing apprenticeship and craft? That's my challenge.

This is parametric urbanism. We won major competitions—I showed yesterday the buildup after 20 years of this master plan. When you come to a client, you explain that instead of making 100 of the same blocks, you can make 100 different blocks, offering the market anything from small to medium to large, making a coherent connective whole where the translation can vary. We're winning all these competitions now, and this has been fully built out. It works very well. That's what our cities should look like: they have a new identity, but they're adaptive to locality.

We're using sophisticated tools to do this. Instead of having different zones separated, we want to bring them together and overlap them in intricate ways. In terms of functional separation like Le Corbusier's plan—the business district, residential district, logistics, factories all separate internally with repetition—that's not how society works. We now have live-work integration, continuous communication, and so on. You have a model like this, and then you can script tectonic translations onto this vector field. Each color can translate the vector field's density, directionality, etc. into different morphological parameters. These are the kinds of urban textures you want to generate: open-ended, but with coherency and legibility.

The history of this is quite interesting. This started in 1992–1993. Columbia University was the first in '94 with a digital studio; everybody joined in. Then came the AA, Harvard, Yale, SCI-Arc, the Bartlett, and so on. Every architect of this generation realized that's what we had to work on. They were doing different things, but in '92 and '93 everybody started working on this; it was a total takeover, and we accelerated. The problem was that in 2008, the financial crisis blew that up to some extent. All these wonderful works were canceled—projects in Europe, Dubai, and everywhere. Then we had the Arab Spring and other economic and political upheavals which stopped this. There was a change in the mind of the profession: the Occupy movement, social justice, anti-capitalism. Architectural schools, education, and biennales went somewhere else; they were highly politicized. For this group of architects, the dynamism slowed; we continued, but we didn't grow as a movement.

What I find interesting is that Zaha Hadid Architects and some other firms continued, doing more and more projects, as did some commercial firms, particularly in China: Shenzhen Airport, Mumbai Airport, many of our large projects like Beijing Airport, large city projects. In the overall discipline, it's still a drop in the ocean, but it evolved. We matured; we learned how to integrate structures and environmental engineering; we learned how to apply this to all scales and programs. We grew, but within a narrow band, and we weren't influential enough.

What I find fascinating—and that's why I show this—is that when AI came on recently, you'd go to the Dezeen awards and see no parametricism; it's all retro. But when AI came on, suddenly you realized, first of all, we could use it. There was so much of our stuff out there on the web that you could generate prompts saying 'Zaha urbanism' or 'Zaha museum' and generate all of this. It was wonderful. I generated 100 new projects in an afternoon by myself because DALL-E 2—which was a breakthrough—and the internet were full of our references. We're using it, and our most recent work in tectonism is much more sophisticated, integrating engineering logic. These models could get poise, balance, perspective, light, and shadow right, and understood parametricism. We could also use it for urbanism.

We now train our own neural network models to integrate, and we can generate these urbanism models rapidly in multiple variations, feeding it with our own references—the big Stable Diffusion model combined with our model. We can parametrically proliferate and mix degrees of models. Because we have done about 100 master plans, we can feed them into the AI and generate a thousand more. We can do wonderful work where you want to visualize the final architectural feel in a building like this. As I showed earlier, we can do that even by programming with color: this is water, this is green, this is morphology, and it picks that color coding up and runs with it architecturally.

What I find fascinating: if you Google 'AI architecture' without any prejudice, without saying 'parametricism'—just Google 'AI architecture'—what do you see? 75 to 80% is parametricism. Something strange is happening: the world and schools of architecture are ignoring us, but when young architects show their desire, what they want is that. That gave me a lot of hope. Everybody wants parametricism, but it's hard to do. They would love to do it. Maybe in school or the professional world they can't do it, but at home after hours they can now visualize it. If you look particularly at all these teaching programs that want to teach you how to prompt well in Midjourney, what do you see? That gave me hope that this will take over; it must.

I tried some things more with tectonism, but the problem is at this stage there's no engineering intelligence. These models can get perspective, light, and shadow right, and even tessellate a body—they have learned from 500 or more examples of parametricism—but the current stage of parametricism, what I call tectonism, they can't do yet. There are not enough references, and they haven't learned that, so there is a lot of nonsense and missing members because the coherency requirement of structural models is higher. That's only temporary, because there hasn't been that learning curve. That was a challenge for me, and I will show later how we are now training AI to do this as well. If you look closely at these, they are very incoherent, with missing members and elements, and these attempts don't really have structural logic; it's kind of pretend.

That leads me to this recent style. Over the last 10 to 15 years, we worked on tectonism. There are not many built examples yet, so AI cannot learn it. I will now go into this later stage of parametricism. You can see it looks even more versatile and rich than the earlier world of parametricism. Instead of just being NURBS surfaces, blobs, or subdivision surfaces—say, certain Maya geometries—they have anticlastic tensile minimal surfaces, active bending, inflatable surfaces, compression-only surfaces, conic developable surfaces. A whole technical range of geometries is much more rich and diverse, and we can have an eye for this. Then there are all these coherent reticulations and patternings, whereas before we only had grey or white.

This is the world of tectonism, which is much richer than the earlier stages of parametricism because there's full integration of structural engineering, robotic fabrication, and structural optimization like topology optimization built into it. This is a series of my writings and books, and the latest book describing tectonism came out last year. That is what AI can't do, and it's much harder to work on as a designer because you need to learn a lot of the engineering logics and Grasshopper plugins. But it's still accessible because all these plugins have tutorials: you have mesh relaxation, and you can form-find. You don't stand there and model; you run algorithms which find the right form, which is much more beautiful, coherent, faster, and has an inherent rationality.

Tectonism is the most advanced stage of parametricism and implies the expressive heightening of engineering- and fabrication-based form-finding and optimization processes. Everything is run through engineering logics, but you can compose freely. The big advance of this type of engineering is that all prior engineering was constricted to platonic forms, symmetrical compositions, and repetition of elements, which is highly irrational. Here, you can take very irregular plans and run a fully optimized equilibrium vault across a freeform plan. Or you take a structure with an irregular railway, where positioning points are irregular and a bridge hits it—a totally asymmetric condition—and run a topology optimization to resolve that beautifully in a differentiated structure.

It takes all of this into account, which is impossible to calculate in traditional engineering where you have to think about pure systems like trussing, Vierendeel, or post-and-beam systems. These new structures are a mixture where connections are moment connections, and pieces are partly in bending and partly in compression, with the objective to increase information richness and legibility in the built environment. For architects, engineering optimization is wonderful because your buildings become lighter and eventually cheaper. But what we are interested in is that they achieve a richer, unique identity and legibility. You can understand what's going on: you're in the building, and you see where the bridge hits or where forces flow in a vault, telling you where the big entrance opening is. This legibility, which you cannot get under modernism, is what I'm looking for.

This is the new world, and this is parametricism before the tectonism advance, which is a massive advance in rationality. When we did designs between these surfaces earlier, we had massive poche of bizarre structures just serving what we wanted for social and spatial purposes, but now we can integrate this. This approach is much closer to the endless forms of nature. Within parametricism, you can have massive advances. If you compare high-tech—something like the Hong Kong HSBC building or the Pompidou Centre—to the Bauhaus of the 1920s, there's a massive advance within one paradigm. Here, we're having a massive advance within a new paradigm.

Tectonism matured in the last 15 years. These things take time. Again, I came out with the concept of parametricism 15 years into its development; tectonism was formalized and canonized 15 years after its development began. So we have a totality of 30 years: the first 15 years led to parametricism, another 15 years to tectonism, and there will be more coming. High-tech is distinct from early modernism, and tectonism is also distinct from early parametricism. You can see it, and it's more diverse. It doesn't mean it's always super expensive; these are simple tensile structures.

I want to show you that we're talking about paradigms and methodologies, but also styles, which I think we should understand as a very profound concept. There's something very profound when you move from Gothic into the Renaissance, into the Baroque, into Neoclassicism, Historicism, and Modernism. In the 20th century, there were manifestos and books; we need to understand that because this sequence of styles tracks socioeconomic epochs of societal development. That's my argument: epochal styles, with transitional styles between epochs, and subsidiary styles within them—like white modernism, functionalism, brutalism, metabolism, and high-tech within modernism. In parametricism, the subsidiary styles include foldism, blobism, swarmism, and tectonism as a big step.

It's most important to understand that we have socioeconomic epochs. These great styles of architecture are only epochal styles, not subsidiary or transitional styles. The first architectural style was actually the Renaissance. Why do I say this? In Romanesque or medieval times, there's no architecture in our sense; it's self-built vernacular and craft. There's no architect, no author. Even with Gothic cathedrals, you don't know who built them; there were guilds and contractors who didn't design with a complete set of drawings, but started staking out and building arches. They were sophisticated, but it wasn't architecture in the modern sense. In the Renaissance, you have full-on architecture as we know it, recognizing Alberti, Palladio, and Serlio as our colleagues.

Why? Because they did complete sets of drawings, simulated the visual impression through perspective, light, and shadow, and wrote treatises—the theory of architecture. If you do something innovative, you have to explain why: if you do vernacular tradition, you keep doing what's always been done, with no questions asked or answers demanded. But if you dramatically innovate, you must convince people and show what you're going to do because it will look different. You have to make a full set of drawings, including perspectives, and write why this makes sense. Architecture consists of authors who innovate, take full responsibility for something original, present their vision in full drawing sets, and provide a theoretical underpinning. These ingredients of architecture didn't exist prior to Alberti, and we still do that today.

The Renaissance was socioeconomically tied to early capitalism in Northern Italian city-states—a particular socioeconomic formation. It moved into the Baroque when you had larger organized states, like France under Louis XIV, with an integrated nation-state and economy, resulting in massive buildings like the Versailles government complex. You could not detail or design this in the Renaissance style, because the Baroque can integrate asymmetric parts with curvature, deep relief, and colossal order for distance viewing, creating massive symmetrical complexes you can comprehend, whereas the Renaissance is very additive. The Baroque expressed the capital of a 17th-century nation-state like France, or the Vatican as a global institution ruling across Europe. Looking at St. Peter's, the Baroque style is the only style that could comprehend that scale. You then have city developments like Karlsruhe: mercantilism and absolutism generate the Baroque.

In the 19th century after the French Revolution, Baroque and Rococo disappeared, and you had the society of Neoclassicism and Historicism. For the first time, they held architectural competitions and built many more building typologies: museums, ministry buildings, universities—whereas before it was only palaces and churches. There was a different spirit, which gave rise to Historicism. Modernism reflects a hugely different society: a mechanical, mass-production industrial society shaped by socialist revolutions after the First World War in Russia, Austria, and Germany, combining social democracy with industrialization.

Styles make no sense in themselves. They make sense as matching up with major socioeconomic epochs. They're aligning the epochal styles of architecture, and parametricism is one of those. Why? It's the whole collapse of the socialist world, which was the most far-reaching and logical expression of the era of mass mechanical production, Fordism. England was a socialist economy more or less, and so on. And so you have the microelectronic revolution and telecommunication revolution through the '70s and '80s, which led onto Margaret Thatcher, led onto Ronald Reagan. It's neoliberalism; it's the transformation of China, Deng Xiaoping, a reemergence of capitalism, the breakdown of many big capitalist bureaucracies into a network of smaller firms, moving away from this kind of greenfield sleeping cities and factory and office complexes in the green, with everything concentrating back in the center of the city.

So there's a dramatic socioeconomic transformation, and it should have an expression in the built environment and in architecture—and that's parametricism. It has, of course, then a series of subsidiary styles. So whenever you move from these big revolutions, you can't get it right immediately. There's some kind of brainstorming, a revolutionary searching period. So you have, at the end of neohistoricism, eclecticism where you start collaging these things, or you have Art Nouveau and expressionism, and then modernism comes in. So you have a 10-year period, 15–20 year period of transitional styles before you have 60 years of modernism. And similarly, this transformation to parametricism wasn't overnight. There was postmodernism for about 10–15 years, and deconstructivism for about 10 years, and then we established this. The problem is that this hasn't yet taken over the world like that.

So, this is the world under modernism, mechanical mass production. They can only produce millions of the same thing, and the whole economy is maybe five things: a car, a prefabricated house, a washing machine, maybe a packaged holiday, industrialized food. I mean, I grew up like this in the '60s, so they were going to 5 to 10 things produced en masse, which means everybody will have the same standard of living. It's called the universal consumption standard. It means sociological leveling because the prosperity gains were massive. In Germany in the 1960s, everybody of 60 million had that car, that house, that flowing water, that television with one channel and then soon two channels, a packaged holiday, industrial food, and everybody the same. There's no point in having five cars. There's no point in eating 25 meals. So it means physically you have to have an egalitarian society.

But now we have a thousand TV channels, right? And it's a totally different thing. It's huge differentiation of products. So suddenly it makes sense to have a lot of sociological differentiation, a lot of income differentiation, a lot of dynamism. And where are these people now who used to be here? What are they doing? They're doing research, development, marketing, finance, reprogramming the robots or this technology. With this technology, you could not absorb many innovations. You have a new idea—no, we're going to do another five million of this before this plant is fixed. Here, every day you can have an innovation and reproduce it. Or even more dramatically, software as a service, a new app, a new update—every minute you can have a new product brought to billions of people.

So, it pays everybody working in the knowledge economy on innovation and so on. Where are they going to work? In the cities. How are they going to work? In teams connecting up. Because a new product means you have to look at marketing, new material research, new science, new financing mechanisms, new advertising, promoting, and market. So you need to connect across. Before, for 30 years you did the same thing; you don't have to communicate bang-bang-bang in these bureaucracies. Now it's network communication, collaboration, networks of companies. It's a totally different world, and the architecture looks different. And that's a city, not this kind of modernist Brasília or something. It's R&D, finance, marketing, etc., continuously new.

And if you're not in the center of things—going out, seeing what everybody else is doing in the company, other companies, going in the evening to this exhibition, to this launch, to this conference, and staying in the pub and debating—the next morning you didn't get all these new inputs. So what you're doing this week is potentially wasting your time because you're not updated. You need continuous updates—everybody connecting to everybody else to update. Because if you haven't noticed the latest release of OpenAI, the reasoning version, you're wasting your time. So instead, everybody's work is recalibrated continuously with respect to nearly everybody else. That's communication, communication, communication.

We'll come to that. But also, you know, we're looking different because now everybody is a different specialist. There are these scientists, the engineers, the artists—you need all of them—the finance, the bankers, the entrepreneurs, the startups, the IT. In the crowd, you need to be able to spot them, and you do. So I'm talking about the semiology of the fashion system, and you have to calibrate how you want to appear. You don't want to be mixed up with an investor or banker, nor would you want to mix up with an engineer, so you have to find your style, and you want to express something different anyway. So that's that, and you want to reinvent yourself.

And the office starts to look like this. In some offices—this is a Google office—they also are differently organized: the teams continuously transforming, shifting, revolving. They're not separate; you belong to several teams at the same time with a different degree. The fuzzy logic of this kind of blobism comes through. Instead of strictly nesting, you have intersecting and overlapping. You have these kinds of spaces with where you both overlap and degrees of belonging and fuzziness. So this is the kind of model, and it's dynamic. We're moving from here to here. This is a project we're doing, for instance, for the Sberbank Technology Center in Moscow. Banks are IT companies now as well to some extent, and that's the kind of world we need to develop. That's where the prosperity gains are; that's what you have to participate in. And these are the kind of spaces: inter-awareness, intervisibility. This is a project we've done in Beijing—our own office is in there. There's co-working spaces, there's also restaurants and bars and offices and retail, and you see what's going on.

Okay, this is my book. I don't want to go into detail of this; I've shown this, so a bit repeating this. Okay, so just an aside: as said, tectonism has been a while in the making, and within parametricism the first glimpses started already in 2000. So it's not all my invention, and I'm the one who's kind of a synthesizer and productizer and promoter. Frei Otto became interesting again. I was actually Frei Otto's student in the '80s in Stuttgart because this is form-finding, and he was also someone early who used algorithms. But this seemed to be something separate; architecture moved somewhere else—it moved to modernism, postmodernism, deconstructivism, and so on, and he was forgotten. But then we rediscovered him.

It's very fundamental to these form-finding processes for tectonism. You see these shells: they're very adaptive. We have four stations in Innsbruck—underground, close to a bridge, on the hill, on top. So it's one principle, but it's parametric variation. Usually, say the stations now: you do a train station, it's repetition. This isn't; it's a set of principles, parametric variation. And they're also shell structures, so they're optimized. And then also, instead of cluttering details—plugging on a drainage pipe or this and that—the drainage is just one of the seams opening up a bit so you don't clutter the thing. And you have here a kind of correlation between two systems: the concrete platform and undercroft and the top. They have some different logic, but they correlate; they relate to each other. They're not separate; they have to correlate. We started this early on, 15 years ago: carbon-fiber shells, tensile structures. So this is the beginnings of tectonism.

And again, we have somebody like Mark Burry, so it's not only us, and Lars Spuybroek. It's been going on for a long time. I mean, Mark Burry was the architect computationally completing the Sagrada Família of Gaudí. And even somewhere like Reiser+Umemoto's *Atlas of Novel Tectonics*. So it also shows digital tectonics, novel tectonics—these phrases of tectonics were swinging around in the last 15 years. So that's why I picked up the name tectonism: that it was the desire of parametricism relating to the tectonic. And these are studies we've done many years ago at Vienna University. So instead of having a shell like, for instance, Foster—Foster has parametric tools and he did the great roof over the British Museum, but when you look at it, he uses this to smooth out so everything is used as homogeneously as possible. Whereas we love that this is different: open instead of ribs, a closed web, gridded tensile shell. So under this, you can imagine many different subsidiary spaces, different zones, but not separated by walls. You can orient and use that ceiling and space to differentiate an event in different subsidiary spaces.

The concept of tectonics as some kind of heightened artistic—in inverted commas—expression of structural logics, fabrication logics, material logics, is a very old concept in the history of architecture. And I do like that; I want to build on this because we're doing tectonics in a new way: parametric tectonics, tectonism. And of course, I looked at these. They're wonderful. They're structurally optimized and super-high-performance structures, but they also at the same time become ornament. And that's what we're talking about. Ornament is actually semiological expression—this kind of styling and giving identity and character to the different spaces. That's a task for architecture too, not only technical.

Or you have these kind of bizarre hybrids of Viollet-le-Duc or somebody like Frank Lloyd Wright. So we have these principles or engineering parts. We're using structural form-finding optimization as one major part, but also environmental engineering and then fabrication constraints: how to make things, whether you're casting, or whether you are hot-wire cutting the mold for casting or milling the mold for casting. This all implies different geometries: whether you use tensile structures, etc., what kind of fabrication you do. And what is interesting to realize—I said it earlier—some of these walls are compression-only walls. Eventually, the horizontal forces have to be taken out by what do you call these frames on the outside.

They only could do that traditionally by making perfect symmetry, because this is an equilibrium composition. And you can actually learn this when you read Palladio: that symmetry in all buildings is a structural necessity at that time, not only in these walls. Because if you build a building and you have asymmetry—put a tower here—you have differential settling of the foundations, and the building will crack up because you can't calculate it. So the symmetry was necessary; an asymmetric building would just have cracks all over the place, break up, and fall apart. So some of these formalized as an aesthetic structural logic: symmetries, of course; that you have the height-to-length of a column; that you have spans, maximum spans, and depth of beam, etc. So all of the symmetry and proportions of classical architecture are aestheticized structural logics.

Now we can do this: you can do any free form because we need to compose the spaces in complex relations, and it would really be a hindrance for a versatile and adaptive design that we have to make everything symmetrical, because institutions are not symmetrical. A mirroring over of spaces makes no sense. But now we can do this. What is wonderful is it also calculates each stone optimizing, and you can see how slender these shells are. These marble shells have no bonding—just pure compression. They're pure equilibrium. If you go to this model and you kick out two or three stones, the whole thing will rain down. And there are also ones which have—which I don't have here—compression and tension; there's a loop. And if you go in and cut this, you can learn from the traditional, but we can do it much more variably.

Or if you look at waffle slabs or any concrete, even the laying of reinforcement: all traditional engineering is obsolete and irrational because it's inefficient. You have to follow the forces. If you put columns in the corner, why do you have a waffle grid? Why do you have reinforcement going in straight lines? You're wasting a lot of material. You also can thin out in the center and bend. So this is what we have to do. And isn't that way more beautiful? And doesn't it tell you where the column will be and where the opening will be? So it's more legible and orienting. And you can run this across a shell also. So we have this kind of optimization of stress, primary stress lines, and that's what we should use.

This is a kind of three hypars connected up based on a Candela model, but using this and then using the stress analysis to layer those areas which have more stress in two or three or a single layer. But isn't it more elegant and beautiful? Wouldn't you cringe if you had to impose a rectilinear grid on this? We have a sensibility for that. So that's tectonism. If you look at this chair, here we have fabrication: we're doing 3D printing. So we are not totally unconstrained. 3D printing, particularly concrete printing, is a whole science with an enormous amount of constraints, and I'll come to this, but it's fewer constraints than we usually would have. So we can super-optimize.

We take the functional form, and then we stress-relax and optimize the shell form. Then we can run structural optimization on that surface for the reticulation. And then there are different ways of translating this: for instance, thinning out versus thickening up, plus degrees of perforation. You could do one or the other. So there's an enormous amount of creativity in the translation. It's never automatic, but you can play with it. And then you get something like this, and you can print that in plastic or in aluminum, and it's the strongest and lightest chair you could ever have. Well, of course there are other versions you can further optimize.

You can do something similar to larger structures, but then the fabrication sort of printing is more rationalized: it's straight lines and developable surfaces, but they correlate and resonate. So similarly with the shell, the first two steps are the same: shell optimization, reticulation optimization, and then there's a fabrication translation into straight lines and developable surfaces for the holes. Anyway, for instance, towers: I guess you've all learned—it might still be on the curriculum—that up to a certain height you only need a framework; then you need to have a core and framework; then you need to start putting outriggers; and at a certain height you flip into an exoskeleton, meaning to use the outer surface like a tube.

Now I'm saying there is a fundamental flaw here, which is the thinking in pure systems. You can take this diagram upside down and it tells you already how absurd it is, because the forces accumulate exponentially—wind forces and weight forces. And that's one thing: of course you do that by making the columns here much thicker and thinner there, but also the system separation is wrong. I'm proposing where you start with the tube and move into the outrigger and then into framing, and you can also phase-change systems. So that's the kind of thing we've been doing, and you cannot flip that around. And these are still rudimentary; they're not fully optimized—there's still symmetry—but it's a step for these kinds of projects.

And this way you can also then differentiate the product: three flats, two flats, one flat per floor. You can open up the column and get much more filigree and light on top, and you get much more framing action at the bottom. The other thing is, when you do exoskeletons, you can release an enormous amount of internal visual connection and freedom. So the tower should actually be like a vertical street, instead of everything separated floor by floor and, even on the floor, through a central core separating the rooms. Everything is segregated: you're in a building with 10,000 people and you all disappear in a corridor and never see each other. When you put the exoskeleton and run panoramic elevators, you can see what's going on. You're flying over the restaurant, the bar, the cafe, the conferencing, the lobby—you keep seeing things. You sort of look back into the urban spaces, etc. This is the kind of thing we're going to do.

Anyway, so many examples of this. You're doing this kind of shell optimization, articulation, and this is where you can start having an artistic—in inverted commas—repertoire. You can translate this into extruded ribs, more or less pleats, and you can go into openings as well. And you can have a whole variety of characterful differentiation, and you can therefore give each space a different character degree. Structure becomes ornament, like in all traditional architecture, and ornament is characterization, semiology, information, communication. This is where we combine tensile structures and shell structures—something Frei Otto also never did. So we started to say, instead of having the tensile structure run off pylons, you have the tensile structure running off walls. This was our winning entry to the Tokyo Olympics—couldn't make that—or we have done something like this in large structures like the airport in Beijing, and more recent projects, grid shells. And again, you don't have to make them even grids.

Frei Otto is so interesting. He had this kind of path network optimization, but with circulatory paths you can make the analogy to load paths. So these become—you materially compute irregular load path networks onto irregular support points, for instance. I did a project when I was teaching at Harvard and did this kind of project for the Google campus. One of the themes was using this and all the variation you can generate, translating into different materials: timber versions in the smaller ones, steel versions in the larger ones, all sorts of shells in this way, double-height, triple-height, etc. Another team was doing grid shells, active bending. Another team was doing catenaries, catenaries on catenaries, inverted and then differentiating out. So that's tectonism differentiating out.

My conclusion at the end was that each of these is intricate, beautiful, and differentiated. But for a huge complex like the Google campus, where you have 25,000–50,000 people and hundreds of various types of spaces and entrances, this is still too monotonous. But with tectonism, you can have all three and more systems coming in, differentiating out and layering. So I'm not happy with these up to a certain point; bringing many systems and variety together, that would be the ideal thing. And this Google campus, as you can see, was interesting. We pitched heavily for Google and Larry Page and so on, and they just didn't understand what they're doing because they said, "What we want is somebody who's delivering net zero." Okay, waste your whole company: 50,000 people waste their time because they're sitting in a net-zero bunker. You really don't understand what you're doing. But then in the end, in the second round, they actually did something similar to this: they got Heatherwick and Bjarke Ingels doing the big shell in open space, but they still made the same shell three times, so they still didn't learn.

This was an interesting tectonism project: a whole district talking about unity across difference. This is not garbage school, but it's very varied and intricate. That's a cultural district developed at the school in Vienna with the students for a new cultural district in Turkey, and different teams using very different principles of tectonics. You can see they have tensile structures, grid structural framing networks, shell structures, conic developable surfaces, etc. And it's quite intricate. If you go into detail, they're all quite different, yet they harmonize. They operated in parallel, observed each other, and followed a single paradigm. That's my paradigm of city development.

Okay, this is another one: if you do topology optimization, you get an enormous amount of variety. There are different tools: one which erodes the solid, which maybe you can translate into concrete; those which start with linear elements, and you can end up translating this into timber or steel. Even in this kind of optimization process, there are different packages and different sets of parameters, so it's an enormous variety. If you compare this to traditional structural thinking where you have three solutions, here you have 300 solutions which are distinct, but they're all optimization. Because what's the point of getting it down to the last 5%, right? Because you have lots of other things to think about: expression, the kinds of spaces, the kind of materiality.

So these are kinds of projects we're doing with students: the whole of New London, using this variety, that all of them have exoskeletons, atria, a lot of bridges. So this is something you can do if you come to work with me as a student: a whole city detailed to this level in one year. It's, of course, very long, because in the office we have to do this in two months. But what's very important for us as a firm: we have done shells and tensile structures, parametric urbanism, topology optimization, tectonism—various ways of advancing, always using the university, using this kind of post-professional degree. I've been teaching for 30 years at the AA, developing these themes. So before we did our first exoskeleton, we had 10 years of research at the school. Before we did our first shell and tensile, we had 10 years of research in Vienna, etc. The same happens now with the metaverse, happens with everything.

Okay, so environmental logic we earlier started. When you do—and I love the aspects of sustainability which are about passive systems, for instance, shading which is expressed on the outside. This starts with massing and volume configuration, self-shading, etc., you exploit. But then it's also sun exposure, and you can run a sun exposure map and script a geometry translation directly onto this. So you can translate degree of sun exposure into density of louvering. And of course, there could be 100 different systems and you can run them, and then you put another building in self-shading, these things thin out. And when you look at the building, you know, "Ah, there must be something here." And on an overcast day, you can see where southwest, east, north are, so there are a lot of things.

This is a project we're doing: the Central Bank of Iraq on the Tigris. And here we try to combine the double constraint: structure and shading brise-soleil in these vertical elements. There's a big atrium in front, so you move up, you can see the city, and you interconnect with levels. But what is beautiful is how it ties up: there's a solid wall at the bottom, and then you have the offices, and then there is a podium where the light and path connects up with the atrium. So there's also this kind of articulation of a pathway. You can see how this looks very different from different sides, so it becomes like a sundial. And it's also in the Middle East, very hot, so you can cut out the heat.

So there is no justification to have a tower symmetrical all around. It's irrational, but it's a wrong default we have had so long. It's taboo. We still do it sometimes, I'm sorry to say it, but it shouldn't be; it should be taboo. Some people decide to put louvers on their facades. How are they going to do it? They wrap it on all sides. It's insane. Instead, it needs to be optimized and differentiated. So this is another one of our flagship environmental engineering projects, and what I like about it is it gives character and identity to the place, to the project. You sense it's in that climate, in that region. This is a research center in Riyadh, KAPSARC. It relies heavily on small windows, deep-cut facades, but it still wants to catch a lot of light from where the sun is not hitting into it, and it wants to catch wind.

So we did the volumes, and the shading canopies have a related geometry. It's all based on hexagons, but not repeated hexagons—hexagons which squeeze and pull. And the hexagon has six sides to connect instead of four, so it's more communicative and connective. And you can see here, this is where you don't get so much direct sunlight, but you get the wind coming in. And what is beautiful about this: you have the solid part and the tensile part; they follow the same environmental parameters. They orient their oculi in the same way, so you get system resonances and correlation. And that's the beauty of it. And then you have outdoor courtyards, you have indoor atriums, and you have these very different situations. They're different and similar. Why are they similar? Because they're driven by the same parameter logics.

And if you now make many more different buildings around and they all heavily articulate these environmental logics, that's a unifying force which we don't actually do in reality because the discipline remains extremely unsophisticated. It's as if you wanted to write on a typewriter with your fists. Your tools are crude. Our tools, our repertoire, are so crude that we can only do very primitive things which you wouldn't want—it's like the surgeon coming in to you and cutting you up with a ruler and circle, you know.

I think there's something which becomes intuitively appealing to people. It's very loved and used for public events. It's 21st century, unmistakably. It's original, it's new, it's fresh, and it's still Arabic. It's still embedded in the tradition: certain typologies like courtyard typologies, and the environmental affiliation, etc. So I'm just racing through this, and the modernists got this right. This was before air conditioning; you had much more expressive architecture. After air conditioning, what do you get? Mirrored glass boxes spreading around the world.

Look, we've done that, and this is how we look at this and how we do our translations of this. This is Brasília. We now do this, of course, in a way which is very—in the south, you have emphasizing the horizontal louvers; they can thin out to the north, and then they become extruded; in the west, they become vertical because the sun hits very flat. So you get this kind of transformation of the different sides, and you sense it on the inside; you know where you are. This is a building just finished in Prague, Masaryčka. And so learning from these, these are our projects where we're learning to work heavily with deep relief, with various types of brise-soleil. There's multiple levels: there's the main ribs, there's the secondary, and then there is a grille which is also pleated directional. So in some spaces to the north you can look out and see the sky, but there's never direct light coming in in the daylight gallery.

This is a proposal we did for a Budapest blob with this kind of shading, intense densification, and degrees of extrusion. But then also formal affiliations to that—you're framing a surface with a backdrop so you can make it formally similar, because you have many ways of doing louvers and many ways of doing lights. So they could also share and participate in generating a space. So it's not pure engineering. It's engineering as a constraint driving and then composing a space. There's a big difference. Engineering is our input; we are orchestrating engineering options into a composition. The more exposed facade becomes to the exterior, and then it shallows out to the interior. But this also tells you public/private, so it has information. You can use these details to characterize spaces.

This is a simple idea of system differentiation: never repeat, but self-similar and parametric variability within a parametric element, parametric components. This is a default: if you do a system, which parameters are you giving it? Which variability do you give it? What is it driving? What is it relating to? And so the default is never repetition; it's always variation. It's very rare that repetition is the optimal solution. Particularly if you have a curved building, for sun exposure it gradually changes as the degree of undercut is different from exposure, etc. These are very old projects we've built, and you can express larger open spaces to the exterior, so there's a legibility to the exterior as well, integrating podium and tower, potentially tensile elements. There are many variations, studies, and options of all of this.

Shell structures: super-high-performance concrete mushroom shells and clear foil as a facade—you open it with a zipper. It's a house I was designing for myself in China, but it never happened. Timber roof overhangs as well, climatic for tropical climates. You can also affiliate to traditional tectonics of the region. So some of our Middle Eastern projects look Middle Eastern—climatic, but also typology, morphology, materials. Some of our Chinese projects gain a 21st-century China character. I was asked yesterday, "How about this?" There's something you can learn from those regions and features.

Not only that, you have relief and depth. This is just light and shadow studies: how a building, the character of a facade, could be very different at different times of the day. It's the same building. Or this, which I'm talking about: observer-dependent parameters based on observer positions. So you can also do this when from different perspectives it appears differently. Look at what happens: only the camera is changing. It shifts from a grid to circle packing. It's the same architecture depending where you look at it. You can affiliate it, you can make something homogeneous or differentiated, affiliate it to something else on the other side, etc. It looks like either a wave, circle packing, or a grid depending on the point of view. This is the opposite: it seems in the top view as circles, but then the circles disappear into a grid, into a wave. So that's another topic—I'm not going too deep into this—phenomenology: investigating the morphology with respect to perspectival positions.

It's also some kind of effects like this: the entrances here where we accumulate depth in the facade. This is student work. Again, this is a team in one year: they can generate, use balconies as shading and louvers, etc., including the self-shading integrated. And you can see the shift also from horizontal to vertical like we discussed. Or another team: shallow to deep depending on site sun exposure translated into geometry. So you get these kinds of visions for a dense urban cluster.

Fabrication: we've done a lot recently. I mean, I'm not sure how you do it here, but in the last 15 years there's also been a lot of investment in studios, student studios, master students, particularly at the Bartlett, for instance, and also at DRL, about experimenting with fabrication, pushing the industry, pushing the engineers as well, with the toolmakers as well, in building extruders, hot-wire cutting, all sorts of robotic fabrication technologies. And in Stuttgart they're doing it a lot: carbon-fiber weaving and threading, and you get these wonderful experimental structures. You use inflation and then robotic carbon-fiber-laying cells—wonderful stuff. This is not only us; this is connected to ex-students, etc. Aggregation of elements, minimal surfaces, but fabricating with aggregating small, super-thin elements.

This is our own material: a super-thin shell robotically folded. With this curved folding, the robot creases and then folds these elements, which then aggregate into a super-thin millimeter aluminum shell we showed at the Biennale. And then everything becomes part of an ornament as well, so we don't mind the bolting on, and they enhance the richness of the surface. 3D printing: we have an exhibition at our gallery where we're also hosting lots of young talent and experimental work in tensile structures, hot-wire cutting. If you do these molds with milling the polystyrene versus hot-wire cutting, it's about 30 times faster with hot-wire cutting. But then you have to live with these geometries, but they're actually quite lovely—they're all ruled surfaces.

And I mean there are others studying this: digital tectonics, 3D-printed tectonics, etc. So the enhanced tectonism for me is that it's not only technical. I'm emphasizing the communication aspect. You can see and learn what is happening—the natural differentiation of the structure. You look out that it's conspicuous enough; you look out for legibility; you look out for orchestrating the different systems to have both similarity and contrast where you want contrast. So that's the thing: you look at all these technologies as a palette, a painter's palette, but a painter with a purpose. The purpose is to paint the social picture so everybody can understand where they are, to make transparent the complex social institutions. And that's the idea.

Some of it comes naturally. If you run a script and translate tension into density of points, run a networking script, you can see where the forces are accumulating. If you're in the elevator and remember which floor you are on, you can sense, "Ah, now I must be somewhere up high," or "I'm quite low," because there's always something telling you. Whereas modernist repetition is incredibly disorienting. In row houses, coming home in the evening, you have to concentrate: where is my home? You can't find even your own home every single evening because they're all the same. In Brasília, which building is it where I'm working? Which floor am I on? There are two things which are incredibly disorienting: repetition, particularly isotropic—if the grid is equal in all directions, which direction am I going? Symmetry breaking is good: aha, at least I know I'm on the north-south. In Manhattan, I still don't know whether it's north or south, but I know I'm not east-west, so it's better to also differentiate this. Symmetry breaking equals information; that's information theory.

The other problem is, of course, visual chaos—the garbage spill. Repetition and garbage spill are equally lacking in information. So it's menacing in a world where you have to continuously move from one place to another and meet new people in different places, and then you have to look at the GPS blue dot, but then you don't see what's going on; you miss out on everything you wanted to see. So these kinds of logics we can run through the script: at a certain degree of density, I can flip into making these openings. Then I run a core through this. And when you run the core differentially far from the surface, if you then run ribs from the core to the surface and let them taper, they're differently thick when they taper back. So when you have a deep beam at the core, you know in this direction the space is deeper; if it comes back and starts from zero, you know it's shallow. So basically the structure will tell you in which direction is the largest spatial expanse. Everything in this correlated world becomes a piece of information for other things. I call it inference potential. It's very important.

that comes naturally with parametric modeling. So you can see that here with these ribs very far deeper, very close, very shallower. But these also have an inherent elegance and beauty if you optimize like this. Then you put an external surface on it. And this kind of orchestration of the elements I think is very satisfying and nearly comes natural. We still should then track and make conspicuous and orchestrate the visual correlations, etc. And what is interesting about these experiments, there is a kind of network of similarities you can orchestrate saying, "Okay, I have different timber structures, so they're different and yet similar in this way. But they're also similar here because this is from the geometry, but the material is different." Similar geometry, different geometry, similar materials—you can make a network of what belongs to what in a complex network. That's the way we can use this sociologically.

And when we come to the kind of ordering of society, this was the first kind of ordering: just make a loop, enclose something against the rest, and then have a series of nested sub-enclosures. So it's enclosures within enclosures and a subdivision logic. That was, for most of history, the only ordering. Overlap, as I said, is very important, came in very late in the '70s, or theoretically in the early '60s, and I showed this earlier. So if you look at the history of settlements, there's always this nesting thing. It's very important because without settlement structures, you could not have a social organism. You couldn't move from the animal kingdom into the human world because the human world is about structured larger groups. Hundreds and thousands, you need to recognize them through where they are. They need to be placed and relate, and all the activities you have together in various subgroups need to be laid out so you can find them, and the memory structure accumulates and grows with the society.

But also very important is this, and that's the importance of semiology. There's no human society which doesn't have settlement structures which structure the social organism, and where they have graphic transformation to distinguish different characters, hierarchy, social group affiliation, male/female, married/unmarried. Without this, you couldn't have an ordered social life. And as I said earlier, it's very important: the fashion system. When you go into a room, you know who is the boss, where are the engineers, where are the investors. And so you have this laying out the kind of cooperative organism, but you also have the graphic overlay because at this point maybe the physical options of making structures are too similar to represent the social differences. So you need to start putting the graphics on yourselves and onto the buildings. There is no society which doesn't have this ornamentation, either on themselves or on their environments.

And it matters because these tribes are bone-breakingly poor. They're on the edge of survival, but they're investing in all this ornamentation. And to this day, if you reflect how much you pay in your wardrobe of your annual income, this is vital. No society could survive; those tribes which didn't do this had no chance of survival because they had kind of social ordering dysfunctionality. It's universal. But also to ordering the space. If you don't do this, you will have physical hierarchical battles continuously or social ordering problems continuously, but also friend and foe distinction, hierarchy, professional occupation. It used to be extremely rigid. Now we can self-fit, but this was policed, of course. You're going to have choice in what you're wearing and how you ornament your house, but it's absolutely universal. It only exists in institutions which are extremely important, extremely tense. Everybody in their place—social locational semiology. And they still do that in England; even the barristers wear robes and so on.

That's why we're also working on this. I'm working on tectonism: 21st-century dinner jacket, super light, stretchy, I can run with it, it's breathing. You have all sorts of technical issues: you have cooling fibers, you have movement management—where is it stiff, where is it flexible, where is it breathing, where is it absorbing moisture. And you could differentiate the thing. We did a whole collection for sportswear, etc., and how do you transform different pieces and recombine, and so on and so forth. And so we did a lot of that because it's important, and the semiology goes through fashion, interiors, furniture, architecture, urbanism, everything. So it's really always been going on that you had the piano nobile, the kitchens and stables, the rustication. This is semiology. This is ornament. You want to express something: the fine polished, the private quarters, etc. And different orders, like I said earlier, the Doric, Ionic, Corinthian for different types of institutions: theaters Corinthian, the courthouses Doric. So all this became used; they use these tectonic systems to articulate, and it's a universal end.

I want to show you this because to distinguish our task as architects, the way we use functional subsystems and engineering and the way we orchestrate to articulate the logic of an institution enhances orientation and wayfinding. So this scheme is in the suburb of Strasbourg. It's two car parks, a tram station, and a bus station. So imagine in this kind of world—car parks, bus, and tram station—it would totally disappear in this kind of no-man's-nowhere texture. But we created something which becomes conspicuous by doing another thing. So these are the two car parks, but they belong together. So we tied it together across the street, and this is the roof, which is the tram station cutting in and the bus coming around, and it's tying it together. And we're using curvature to tie together. So when the cars are sitting there and occupying, they're also showing the way.

So this is a kind of strong articular piece. And again, you make the cutout to show the tram, the bus. It's a park-and-ride scheme, something which would absolutely disappear in this kind of no-man's land. Usually everything is technically perfect, but we're doing more. You can have a tarmac or concrete surface, and we use that strategically. So this is the kind of lines. And when we choose lighting, we wanted to use the streetlights to work with and enhance the swarm of parking lines to not clutter up with all sorts of random stuff. They align and run. They're growing in height the closer you come in to the station. And then something reappears: the same formal affiliation. Then these things become similar to the columns, and the lines become similar to lights in the roof and cuts in the roof. So we're orchestrating; nothing has a problem with functionality. We offer so many choices. So in the end, what we choose, we orchestrate for formal compositional reasons.

But formal compositional reasons are not just, "I'm an artist and I like this." This is formal compositional for legibility, for articulation, for communication, which is a fundamental core competency of an architect—with no offense for the engineers. But those spaces where architects haven't done that—basements, backyards—where the spaces disintegrate into some kind of visual chaos because they're not public space, it's okay, and the few people who live there, work there, it doesn't matter. But that's the difference between architecture and engineering. Pure engineering would make technical choices to the last degree of cost and bolt things together which are visually uncoordinated. That's the world of backyards and factory spaces and basement spaces. Now we need to realize that we can be tough technically and cost-wise within the 5% margin of that engineer, but we're generating—that's the difference between architecture and engineering, but it builds on engineering.

So this is a ceiling of this shed, and this is the floor, and the similarity comes out of they work together orienting. You can say it's an absurd formalism. No, there's a formalism in the sake of social functionality, and at night this kind of swarm pulling you through. Okay, so that's this idea of semiology. And you can go further if you actually look at graphic systems in the street. This kind of idea of semiology of encoding a visual language—we have it in traffic, but traffic is relatively simple because it's safety and you might crash and hurt yourself, but traffic is a very simplistic, primitive task of ordering social life. The world of our actual activities in the city and institutions—at conference, at universities—needs that kind of semiology much more. But I say it's a core competency of architecture; that's what we should be doing. Like this park-and-ride, it's a very simple functionality as well, but we can do so much more.

In this traffic sign system, there's a grammar. Each space should be a complex sign out of multiple affiliations, like a sentence with many words making many connections to other sentences, etc. So here you're actually combining signs into a complex sign; it's not each is unique. Red is prohibitions, yellow is warnings, blue is must-do, this is not allowed to do. And then there's the icons; they can iterate across these. Then there's the outer shape. So there's a number of components which come together into a visual language, and we can do something similar in the world of design and architecture.

This is university design. You can color code initially on a diagram, but why not color code the real space as well? I'm coding it a different way: how they're subdivided and connected and overlapping, etc. So we're saying we're doing something with shells. Shells give a nice unity where there's a center of space rather than a flat... Shells give you unity, but also they still could be open and yet give you unity. They could be different: you could make a creased, a faceted, or smooth shell. They more or less have similar shell action, so that becomes a choice where you can encode the school of science, the school of art, the school of humanities, if you like, and then have them across many different scales and variations. Then you have different things like lecture hall, library, workshop, seminar; they reoccur in all these different schools. So you can then run the different shells across these. You have four forms and three shells gives you already 3 * 4 = 12 at least combinations. So you get a combinatoric logic.

And then you can make combinations: you can combine by melding, you can connect through an interim, direct, one is dominant, the other is dominant. There's an algebra or grammar of connectivity of different types. You can also combine different shells, same shells, etc. You can nest them, overlap them, and articulate the functional diagram. Let's say if you think through the organization, you do a bubble diagram sometimes, right? And when you do them, you use curves. Why? Because you see the things properly. If you intersect two curves, it's easier to see the overlap than if you do two rectangles. There's something about convexity. When you then translate it into the final design, it's all rectilinear, it all disappears. The legibility of the diagram gets degenerated when you also have to tight pack. This doesn't fit into the site. So what if you would not degenerate the diagram, which for graphic reasons you make as articular as possible so you can actually orient and see the ordering in the diagram? Why doesn't the building become the hyper-diagram—doesn't degenerate, but it keeps enhancing? And that's the idea here.

So then you can have different tectonic articulations. There are different options and you can reaffiliate again along different dimensions, so you have distinct outer shell but similar reticulation, etc. You can also transmute and morph from one end to the other. If you have a gradual transition, instead of invented patterns you can have topology optimization and network optimization anyway. And so you can generate something which generates variety, complexity, layeredness. And you can generate something which now looks complex—we don't understand that, but when in a space, you know where you are: in the overlap between these two faculties, in the lobby between these three, between the workshop and the lecture hall, and this is the direction of the lecture hall, etc. And you follow the trajectories. This is the kind of project I'm talking about: using tectonism for semiological articulation.

This is the same with diagramming. For instance, something like convex spaces are meeting zones of various scales; concave spaces are working zones. But these two are distinguished from closed versus open networks of open social and circulation spaces, but each is parametrically variable, so it could be many shapes. You can always track it's a working space if it's concave; they are always convex. You can then think about also morphing from one to the other—degrees of pure work, pure meeting. So they have a parametric grammar, and you can then also see how they nest: meeting nest into work, work nest into work, overlap. Two works can generate a new work, two works can generate a meeting, a meeting and a work can generate a meeting or work, etc. So you can have a lot of options of articulating.

And then we're going into 3D. We can also make it kinetic and you can still legible at each state. And you can then have an overall color gradient, for instance, or an overall gradient from more closed zones to more open and fluid where working spaces in the lobby kind of quick work on part meeting quick. So you can have them like this semi-articulate, and then you can three-dimensionalize this. The same logic can be used for the furniture. So the concave form is a work desk, the convex form is the meeting desk. You can make them embedded in the open landscape and so on and so forth. And then you can animate it by bringing the crowds in.

So this is the idea: that we can generate something where everything has an information. You can intuit the language like any child. You can all learn a visual language. If you use it, you learn it. And therefore, everything now to some extent, of course, the environment gives you lots of clues. You intuitively know where you are and what's happening. But it's very stereotypical; it's not very nuanced. There's few categories. You know whether you're in an office building or residential, and sometimes even that gets obfuscated. There's five types of spaces: you know when you're in a corridor, but I want these kind of bleeding and in-betweening. For instance, if you go to Google campus New York, the corridor is not just a corridor; there's lots of things going on. It's widened into a space. So it's lots of in-betweening, lots of subcategories: ceilings high, ceilings low, carpets thrown in, quieter. We get some of these clues, but this is systematizing such a language to give you a lot of ordering and allow everybody to quickly, intuitively find each other, notice what's going on, what is to be expected if you're lingering out here, who might come around, what might be happening, what is the social protocol inscribed here. We need this to be much more richer and nuanced. And I'm saying it's like a blind person can read the whole text by driving their finger along these dot makers; we can go through a space and take in all of that and intuitively empowered.

Q&A, you said is open-ended. So what we're doing here is therefore different places, in this case, similarly encoded, trigger different behaviors. You slow down if in the carpeted area; in the public space you're more quick. In the public space, somebody is wanting to proxy, you talk to what is you want; in a private space you have to engage hello and this. So each type of space, the ambiance puts you in a mood receptive to what is allowed, what is not allowed, what is expected. So this is what we're trying to do. And also when it reconfigures, there's new things possible. So what we're doing here, we're saying the meaning, the social meaning of the spaces is actually the behavior of those people. What should they do? How do they interact? Are they sitting down? Are they staying long? Are they moving through? This behavior is the meaning of this semiology. So we can bring the meaning and purpose, which is the same thing, into the design, and we should do that.

So my challenge is this: if you draw a building or a piece of furniture, animate the purpose. What is this for? What could it be used for? Then you're an architect or designer. If you just draw the piece, then you're not a designer because you just draw something which can be measured, can be materialized, can be technically translated, but you don't test and investigate and imagine the purposes. Of course, if you only reproduce stereotypes—this is an interesting form, it changes and it's slightly amorphous. What is it? Well, the interesting thing is it could be many things and could produce a form and could be an in-between situation. If you only draw stereotypes, of course, then you know what it is for; you only have five things: this is a dining table, this is a working desk, this is a sofa, and that's it. But even in the '70s, when you have social revolution, you have sit-ins and all sorts of situations, you break out of these typologies. Then you have to investigate what they mean, what they're for.

And we're also doing empirical work, and then we are simulating, testing our environments: different kinds of groups, different teams, visitors. What might they do? What would they do? And you can encode their behavior. They're attracted by certain things. Different groups are attracted to different things. They have different protocols here: when there's somebody sitting down, you sit either close to it because it's a private space; here, you sit further away from it in this space. So it's encoded. And we can run this and simulate and optimize for social interaction. That's the idea. So we are saying every model which is an architectural model should have always the crowds running. And if you shift the sofa, new things happen. And if you re-encode the ceiling and say this belongs now to the private, different things happen.

And if you want to coordinate thousands of people within hundreds of meeting rooms and zones and 50 entrances, and you want to test differentiated types of agents, we can now run thousands of agents. Then you can say how many people will encounter each other? How many people will convert encounters into communications? Will it be long communications, small communications, one-to-ones, larger groups, smaller groups? You can anticipate this. And you can also design strange things. If you do this, you don't know what you're doing; you need to run and imagine and play it out. And so that's what we've been doing: bringing the meaning and purpose into the geometry together. And also kinetic transformation: a big thing that instead of having 100 spaces, you can transform one space into several things. That's very safe and it's not so distanced.

And you can help cajole and enhance and signal a small group is forming so more people can see and join or separate out, and walls opening up. So as a kind of dynamism, robotic intelligent architectural agents facilitate formation of social interactions or break it up if it's too long. And you can also have these zones overlap, integration furniture coming in, and dynamize the situation. So in the bureau, in the corporate environments where it matters that you quickly move in from one team to another, from one situation to another—small groups, large groups, big gatherings, one-to-ones, concentrated works—and we even do physical models of this. But then we're also running testing it with agents, and that is really the meaning: that the social ordering effect and the kind of situations you generate is what it's all about. You don't create architecture for other architects and say, "Oh, nice ceiling." That's really not what we're about. Is it a system of ceilings where this tells me and invites for a certain situation versus another? What is it similar to? What can I remember and anticipate?

The furniture tells me a lot of what can I do, who to expect, how to behave, can I join, can I not join? Body postures and filling out helps as well. So that's what it's all about. We're kind of for the lovely social life and productive social life, because you go to an event and you don't get it, you miss half of the event. You had a hundred hellos, but nobody to meet longer, or you got stuck next to somebody and you only had for two hours listen to this one person at that dinner table. I mean, you could have a horrible time in the wrong building with the wrong orchestration. So you can waste your time continuously, missing everything, get locked into meaningless conversation, never get beyond a thousand meaningless hellos, and the conference was a waste of time. Or every day in work is a waste of time because we don't have the right intelligent environments: open yet intimate, dynamic, transformative, inviting, deep vistas where you understand and see many things and you can guess what's going on, where to go. And this is all student work where we—and we also have professional versions of this—where you get these kind of deep spaces and they articulate and transformative, etc.

This should be your kind of world where you don't waste time, where you miss nothing, participate in the things you should participate. And when you sit down on a certain table, you invite others to come; on another table, to say, "Leave me alone." These gestures could be embedded in the space, in the architecture, in the transformation. And it's very important, otherwise you get disturbed every five minutes, or you sit there and nobody's joining you, misunderstanding the situation. We also do material experiments on how you transform spaces. This is with the same amount of particles, how you can do various types of partitioning. This is important as a kind of physical substrate which reconfigures, and that can imagine across the whole field. There could also be events cascading through, radiating through, etc. But then bringing in, obviously, the end users to understand the situations.

It's interesting to work through. You have many of those, but it also operates not only on the furniture scale; it operates on the zonal and architectural scale, also in the urbanism scale. These are different projects or different teams; we locate the projects next to each other, their shared zones. They're different yet similar. So there are many of these kind of studies. And so in the end, the principle is architectural models always have the meaning and purposes modeled at the same time, not model only that geometry. That only makes sense if you can rely on stereotypes. But it also encompasses exterior spaces, public spaces.

Now, this is research we're doing at the Design Research Lab in the AA. We then developed some of this modeling into PhDs. I have four PhDs on this developing real thousands of agents properly articulated, then have an office internal research team develop it further until we then apply that as well. This is, by the way, Sberbank Technology Center, this kind of huge office landscape. There's 20,000 square meters on a plate, thousands of people. The passages we actually elevate so you overlook and see people moving, multiple atria in this building. And I'll show you later that we then run these simulations and test different configurations, not on the building side, but on the furnishing and layout scale and corridor transformation scale.

We then develop these kind of agents, different kind of characters. They have internal states, they have desires, they have a schedule of so much concentrated work, they should meet at least this much, who they should meet—maybe meet more people from a different discipline across department, within department, across hierarchies—and then run these agents, thousands of agents across different configurations, and then measure encounters and measure how encounters get converted into conversations which might need a prop: you meet somebody in a corner in a quiet place, or you can have a stand-up table, or if there's a sit-down you'll generate longer conversations, but less frequent. So you can then distribute this, and then you can calibrate a mix of a range of situations of types of communication. And you can quite specific set targets together with organizational, managerial demands and requests and desires for optimizing communication. So we can run this and you can also visualize it and run individual agents, and you can run hundreds and thousands of agents simultaneously. And they particularly belong to a particular department. We can input who they've been communicating with, who they know, who they don't know, etc.

So that's what we're doing. And just finally, we started to look at AI and the incompetence of AI with respect to tectonism. We're now kind of teaching tectonism to AI so to have that palette to organize the semiological projects and run the simulations in a more enhanced and nuanced, articulated thing. So this is just where we started to teach and train these neural networks to run structures onto an envelope. We can have particular tectonic articulations saying fabrication logics: these are developable surfaces, so sheet material, you can translate with sheet material and you can embed that into the AI with this degree of tectonic detail, shell structures, and then bringing in more detail. So these are AI-generated based on what we trained up. So we're bringing in our repertoire, but we're also generating new synthetic training material.

So this is, for instance, the tensile, the different tectonic materiality tags we learned. And when you then prompt back, you get this applied to the geometry. 3D printing is a very intricate tectonism. And then this idea of the tower I said earlier, translating between systems: you can do that also with tectonic systems, but also material: concrete at the bottom into steel and then timber. And this is the public spaces, the office space, and the residential spaces as well. And that is something we developed—this is not AI yet, this is something we then input. This is kind of a tectonism again. And then we've developed training up the AI of various topology-optimized structures where we actually generated the training data through topology optimization. This was a one-week workshop we did at Tongji with students. We gave a massing model, and then we generated these topology optimizations as training data, and the AI is generating them.

Similarly, with different types of fabrication details we inputted, the AI is learning those for different components. And environmental models: we generate a lot of differentiated brise-soleil systems and then train the AI, and the AI is outputting that as well. So then you have to combine these prompts and generate—this is the structures, and you generate these kind of systems. Look, they're very crude at the beginning, but at least there is some kind of beginning. And this is all the result of a five-week workshop at Tongji where we're teaching tectonism to AI on that basis, with synthetic data generated by the students, training up by the students under our guidance, and outputting the results and putting the results into some kind of massing model on site.

And final chapter very, very briefly: we are simulating on an urban scale with detail and interiors, where we put many of our buildings together. That's about 30 of our buildings: towers, museums, public spaces, blocks, cafe types. I think my computer is at the end of its tether. Okay, it's at the end of its... But this is kind of where we are with the technology: Unreal Engine. It's a live model. You can jump in and roam around and meet with friends and show clients around and move into the various spaces and get a photorealistic rendering real-time for this kind of scenario.

And in this case, not all our repertoire is yet tectonism, by the way. It's tough. Our output is at least, we say, only 20% tectonism still. It's a learning curve; it's the opportunity. It's only segments and parts, but we're getting there. And I want to finally end on this: we're also doing the metaverse and this idea of this intensive knowledge economy communication. In this case, it's a school of architecture hovering above London. Virtual spaces have a similar kind of agenda, and we're also thinking about integrating virtual and physical spaces. But I'll just show you this: this is a virtual Bauhaus, like an extended virtual extension of the AA hovering above the AA in London.

There's another one. So I have these kind of three metaverse schools, which is a new topic, another lecture—you'll have to invite me back for the whole world of the metaverse, which I'm fascinated with. But similar principles: inter-visibility, inter-awareness, semiological encoding, many dynamic interactions, crowds in the virtual space as well. So this is a design school. Okay, so while this runs, this was the lecture. So if there's time, we can have a few Q&A.