We work with semantic attributions intuitively to avoid misunderstandings and clarify what things are for, but it's not an explicit project. We're not systematically working on semiology or anything like that. The research arenas are on the level of social functionality. We had to bring the engineers in, force them in, and we have to also do 3D printing, manufacturing, computational fabrication to start this. But now we should hand back to the engineers, fabricators, and contractors and not substitute ourselves for them. That was my message with Parametricism 2.0 about seven, eight, nine years ago. We have to focus back on our project.
We're talking about spatiology and I have people like Chris Alexander and Bill Hillier and space syntax working on this. Also people like Kevin Lynch, the legibility of the city, semiology — this was a big story in the late 60s and 70s. They brought in Eco and postmodernism had a lot to do with this, what the meaning of different iconographies and architecture must speak. I'm not saying we're starting from scratch but it was not very widespread. We need to have network analysis and all the kind of potential patterns, learning about space syntax — not only street networks but also connecting these are the true ordering logics of these plans. Even if they look similar, there's a hidden kind of syntax you have to bring out.
We can go beyond that and have intersection, overlap, boolean subtraction, the whole set theory and make that very beautiful and complex, and then go into fuzzy logic. Fuzzy logic is much more powerful and information-rich to solve problems. The same as in architecture — if you're not only in or out of a zone or in an overlap, you're to a certain extent with a certain degree in several of these with different degrees. There is much more sophistication in this and you can order and organize much more subtly and adequately. For instance, corporate campuses where you have domains of competencies intersect, where you're part of several teams at the same time. In the bureau, these kind of free-flowing furniture swarms — you can see where this goes. These are the most high-productivity places on the planet, Google campus, etc. And that's what we should focus on.
Now we have vibe coding and it takes you 20 minutes to generate a new kind of algorithm which is neither concentric nor a grid — some kind of multi-concentric street pattern that has a certain legibility as well. When you have this complexity organizationally, you still have to worry about the legibility, the perceptual tractability, because we're not cattle and rats shoved through channels. We need to navigate these places. We need to read a complex situation and there is a science of perceptual psychology of perception. We need to learn how scenes are decomposed which are so complex to have some kind of legibility, figure-ground conditions. For instance, depending on context, these are actually the same color or something — they are amorphous as a precondition to do the strongest of symmetries, or if you put another occlusion layer you actually start seeing the fields behind. There are some interesting effects we have to work towards and learn about to do our job, which is making an empowering environment where you come in and intuitively you see things.
Figure-ground traversal. Maybe you want to have figures which have a certain simplicity — the void is a figure or the mass is a figure and you can move from one to the other. These are things which are in the tradition of our discipline. It's not that everything is starting from scratch. Nothing here is physically moving — it's camera moves, observer triggering transformation, triggering animation of a scene. These are all aspects of what I call the phenomenological project, where you worry about what is perceived, because things are not objectively there. What you identify is what you can relate to, and if you don't understand what belongs together, then you don't know your way. It has been in the discipline these kind of conditions — how do you compose something which is ambiguous between what belongs together and there's overlap of function, overlap of articulating this. This has multiple readings and the environment is very powerful if you have multiple readings of the composition. The very same physical condition can have multiple decompositions and relate to different audiences which find their conditions.
We've done that — you could do a park and ride scheme. Imagine a kind of suburban condition with supermarkets and sheds and car parks, and you put in two more car parks, a tram station and a bus station. It would be invisible in that visual kind of no-man's-land. And we work with a strong formalism. Not technically everything works perfectly, but you have so many ways of technically translating this. You use formal means to communicate and integrate the composition across the street — two cars using curvature, using a field of lines of lamp posts aligning them and working together formally with the demarcation lines. And then they translate into fly-in to become columns and light slots. So you have actually the ceiling of the shed and the picture of the ground. They're formally similar because they're working together on what project? The phenomenological project. Because you strongly are pulled into this shed. The tram cuts in here. The bus cuts in here. All of that makes sense intuitively. That's the project — the architect, the engineer. You wouldn't find your way. You wouldn't know what's where. It would be that kind of garbage spill. That's the difference between engineering and architecture — the articulation of making transparent, because in that complex world, we have to not be distracted by all sorts of items which are not essential. They're just visual junk which we need to suppress and then we need to accentuate those which we rely on. That's the phenomenological project. That's the compositional project.
We all know that. We all do that. And you know where it breaks down — in the backyard and the basement. There's no audiences. You get absolute visual chaos when the engineer optimizes without considering that. You get totally impenetrable and illegible spaces. We're working with conditions where the viewpoint makes a difference, where different readings of an open composition — you can make the sense of a closed space but yet it's totally open and fluid. These are compositions which work in my view. We're also doing the next step — ease of orientation, recognition, ease of navigation — making that explicit using AI. There's a PhD project where we're moving from intuition to science, explicit principles. Yes, contiguity — we're seeing lines here and horizontal lines here and here. What do we see? The similarity is required to make this reading. Now we have vertical lines again, or do we have horizontal lines? There's an ambiguity because we can go by similarity or by continuity and proximity. Here we see the white and here we see the black. Why? Because we privilege symmetry. Symmetry is something we focus and give substance to. There is closure — this decomposes like this but it could also be that or it could be that. This is more likely the spontaneous decomposition. But you can also make different readings and you can make a little gap and suddenly this breaks from here to here. You can transform a space with minimal or nearly no physical transformation — you can make a totally new space.
We can also go in and use AI to simulate the perceptual tractability, the legibility. That's the kind of project where you model something and in your layer structure you know what is a soft, what is a ceiling, what is a balustrade, what is an escalator, and you have a truth and then you train a model with a lot of these examples and they do semantic segmentation. You can machine learn this and allow these items to be recognized, but you have to calibrate. This is a lot of testing — people going, giving them an app. There's an app where you can click on quickly find the escalators, click on them, and you can know how what the approximate ranges and how it correlates with complexity. What we found is that with the human user, the more complex, the more mistakes they make. Similarly then the AI — we selected one which has a similar graph but there's always certain while having a lot of variety which is good when you want to have many different kinds of spaces for functional reasons. That usually makes the space illegible, but there are ways of having a lot of variety and still grouping them and arranging them so they remain legible in the face of complexity. There's a kind of research project which goes through these different designs and then we can pick out these minimal variety and then pick up the most legible one and then we look at all these examples and see there's a certain grouping and alignment conditions which make this more legible.
We're trying to operationalize the phenomenological project and also with path systems. When you're confronted here, you have a destination — which of these 12 or 15 pathways would you select? There's a certain rule of thumb you might use and which environment would make it easier for you to find something which is approximately the optimal path. This is a condition where you have a certain rule and different empirical data collection — how do people select? Because you can't compute the optimal path. You have some kind of rule — well, I'm looking at the destination and then picking the ones which are in line and I don't care about the direction. There are certain — we've tried many different rules. They talk to people also which one, how you think you're selecting, comparing these rules and then choosing one which is very similar to an outcome with proximity conditions, presuming that this will be the rule, the way you collect, and then trying out in different environments and different designs to predict the navigability. There we also discovered again — grouping, alignment — despite complexity you have a much more navigable condition. So this is an attempt to operationalize this.
Finally, the semiological project. In a complex society, we're no longer 25 apes. We are building up this social order and there's no human society which doesn't first of all have a settlement structure — so you know where the activities are and you can structure access, relation, and hierarchies and social order. You don't have to fight out who is the strongest every year to have the simple dominance hierarchy. No, it could be much more subtle relations. But also then there is this kind of graphic overlay of self-transformation — to know tribal affiliation, occupational role, whether you're married or unmarried, or you kind of cross the threshold into adulthood. All of this has meanings and it's very important. It's very expensive to generate all this ornamentation and no tribe that wasn't absolutely essential would do it, and only those who did it survived and grew. It is a huge cost but the benefit outweighs the cost. So we need to understand the benefit of the fashion system, decoration, ornamentation, because when it comes to physical construction, often times the repertoire is more simplistic than the functional differentiation. So you bring in the graphical ornamentation to mark out and highlight. You use all sorts of repertoires of order, positioning, ornamentation, color, and it's non-arbitrary. Only this guy is allowed to wear this crown, and so on.
This goes through all of history and it still continues. It used to be much more strict. They were literally — when society opened up a bit more in the 18th century and people were kind of more impostering in urban environments where you will have more anonymity, larger city, you needed laws to prevent people from dressing and impostering in other social roles. The importance of this — now we're of course free to select and self-select into these conditions but it's more important than ever and we are spending huge amounts of money on our wardrobes. I don't know why the ladies spend more sometimes, but and makeup and self — this is absolutely essential. Architecture gets also decorated up and I don't want to go into detail but these are systems of signification, systems of differences. You cannot by yourself determine the meaning — it's what everybody else is doing. You fit into a system of distinctions and yourself to kind of self-calibrate into this with picking, affiliating with, distinguishing and affiliating. Different languages are quite differently dissecting. The world is not preconfigured.
Another thing we're learning — traffic sign systems. Can you imagine a traffic without any of this, all these kind of regulations and lines and demarcations and signage? These have already some grammar by the way. This becomes the example and something like this in a corrupted and incoherent way is operating in the built environment at all levels. This one is particularly important because you might run over by a truck, but in many ways it's even more important in the social realm to understand the situation. What does a kind of soft light mean? A carpet, a sofa versus a table and you sit and what you do — each comes with a protocol of interaction and an information about the situation, what you do and what you don't do and who is who and who raises a void in this. We're using all of that and tectonism is the much richer repertoire of articulation than modernism to have similarities and differences. You're always similar to many different things and different from other things to certain degrees. That's the way you self-calibrate or an institution and a place is calibrating itself into an iconography and we can do that systematically and that's an important part of our job. We're doing it intuitively but we don't do it systematically.
I'm working with students and in the office on these kind of systems of signification where you set up distinctions but parametrically. They're not the same traffic sign — they all share something. So they all on that level have this meaning but then you could also in between — this is for instance first of all open public space versus business space. Within business space, it's meeting versus work and you can make this distinction. You can in between and you can do a grammar of connections. The way meet and work can be connected, work to work, or you have overlap conditions — suddenly belongs to two workspaces but the overlap could also be a meeting, etc. It has a grammar which works. You test that out and then you can kind of lay it out and three-dimensionalize it. It could also be kinetically still work and be there and you get a gradient from open to closed, etc. You can then go into the next layer — this is the work desk, this is the meeting desk — so you reuse the same iconography and distinctions on different scales in different situations. You can put color onto it, etc. These kind of systematic work where you build up a vocabulary, you build up a grammar, and then you have to simulate the meaning. The meaning is actually what people do — they're actually there and meet and talk versus work concentrated in the other table. You need to — you can have props which structure situation. If you have nothing, you cannot structure situation. We couldn't have easily that seminar if it's totally contingent where I stand and where I start talking. The different territories have different rules of operation. You have also the two teams and the referee. We can also overlap these different — simultaneously have different games at play which different sets of people work towards. This is the kind of world where we're working on and I'm saying we're not only making setups — we have different categories of people and we are bringing the meaning into the model. Meaning is the behaviors which are structured, framed, and suggested, and therefore we can operationalize semiology.
We don't label it up because that's just few categories. We would have to have long text — this is the kind of space with some relation to XYZ and it's semi-informal. So no, bring it in, bring it on, show it, let it play out what the protocols are. And when you shift from one space to another here, you can speed up. If you speed up here, then you're kind of elephant in the porcelain shop. And that's the way things work. And we can work on this. That's social functionality research and development. And that's what we're doing. We're doing that on a larger scale. It doesn't only have to be kinetic — we're running literally hundreds and thousands of agents as a work of Tyson Hosma. I've done three PhDs on this and the research team to look at dwell times and counter frequencies between relevant groups across hierarchies, across teams, within teams. When you meet somebody close to a stand-up table, this then has a probability of converting into communication of a certain length. Next to a sit-down table, less likely. But when you're in longer communication, you can say what is desired, what is a successful place, and then you can run these agents and mutate the layouts. Then you can say to Google — this layout — and how otherwise do you distribute 150 meeting places and social zones and 50 entrances? We are absolutely agnostic and clueless. We have no clue. And how many layouts can you generate before you kind of get tired? You've done then tested five layouts within looking at them intuitively. That's bankrupt with respect to what we could do. And imagine the cost of a building are trivial compared to the salaries and the productivity they gain. We know that the value, the benefit of productivity enhancement and satisfying relations and place of work where you crack on and throw out these apps to billions of people a year earlier or before your competition — that's massive value and the architecture is kind of insignificant. If we can upgrade this and push this in high performance on the social level, what it's all for? Why does Google bring 25,000 people to a place? Not that they have a desk — that they kind of share knowledge, interaction, they find each other, collaborate, inter-awareness, inter-visibility. This is the kind of thing we're doing and we machine learning optimizing for certain hard criteria and then we're running the soft social criteria through agent models which can be tailored to particular companies. That's the kind of thing we're doing and that's kind of social — and again the parametricism helps when you have kind of certain degrees of fluidity and gradients and you don't have only five categories, you have everything in between conditions. And this is then also possible dynamically where you have interoperability, where these agents become intelligent, where they can reconfigure, where you can have responsiveness, where you can intervene and interact with the environment to keep — you have an essential need for flexibility in the corporate world. And you can do that and maintain legibility, perceptual tractability, and semiological coherence and kineticize and flexibilize. That's the vision.
We're bringing in the actors to know what we're doing. Particularly if you were innovative and abstract, you can stare at a form and you have no clue what it's for. You show me what it's for. How people are meant to interact. How is this productive? I'm saying an architectural model which is all about social functionality optimization without agents and simulating kind of interactions within it isn't an architectural model. Then you're just staring at a geometry. Architectural models are these kind of models and we're looking at these projects and the way they treat the end users as physical bodies, as sentient beings, as social communicators and socialized actors. That's what we do. And the engineers only look at physical bodies — physiological comfort, temperature, enough fresh air, and the floor strong enough to carry the 100 people. They're just looking at as physical bodies. Everything is physical for us — it's social, communicative, cultural. That's our core competency. And we need to do that because now it's super more complex and dynamic when we interact in the knowledge economy than it used to be. The architecture shifts to this condition where it's freely free market reconfiguration and so on. But the architects haven't yet made that legible. It is visual chaos. It's better than this which is monotonous kind of dysfunctionality. This has a potential of functionality but it's visual chaos.
We're also looking at the work environment shifting and it's very important. Work is really the driver. That's the prosperity engine. How are we more productive? That gives us the creature comforts and the freedom to you, all of us. None of some of you might have never worked in your life yet you have an app. You can book a flight now to Australia, take an Uber, sit there five movies and three meals later. Voila. Right? So where does it come from? Productivity, productivity, productivity. And that's the way we generate that in the knowledge economy. And the buildings need to be like this and need to be inter-visible, inter-aware. Every day you seat many people and you have to calibrate what you're working on with what everybody else is working on. You have to come to these conferences all the time and meet and see what you're doing, otherwise you're running down dead ends continuously, dead ending and finishing yourself up. So you need environments like this. This is what we've done in Beijing Galaxy and we have our own office there for a reason. There's lots of startup companies, lots of co-working places and restaurants and so on and so forth to make that hub going. And that's why Parametricism is epocal. It delivers this.
You can see that the different eras — by the way, architecture starts like with capitalism in the Renaissance, early capitalism. These are the different stages of capitalism and before capitalism there was just tradition-bound building, absolute — there was only medieval backwardness. But the architecture dramatically is changing, it has to, as the technology-driven societal dynamics changing and the conditions of productivity keep changing. And that's where we are. Even within Parametricism, in modernism we had already several subsidiary styles — this was 70, 80 years. We had the white modernism, then we had brutalism, then we had metabolism, then we had high-tech. But similarly in Parametricism we had foldism, blobism, swarmism, tectonism — but these all adhere to the basic heuristics and taboos of Parametricism and the older versions of modernism were very different. So that's what I'm looking for. In the urban, we have to — this we rejected, we are here, but I'm not happy with this. Nobody loves that because there's lack of identity, sense of place, sense of understanding, sense of orientation, sense of navigation. We're basically living in a garbage spill. Everything since 1980 on the urban scale is garbage spill — disorienting, menacing. That's why we came up with — we looked at this and said this is a spill. The only thing with shape, order, rhythm, direction is the river and maybe landscape formation. So that's the only thing we have — global order which gives you orientation and some structure, what gravitates to that. And so we're looking at models like this for inspiration, formal research, etc. Biomimetic conditions and we're starting to say the city needs to become like this — a global order which has kind of structure, could be extremely diverse and varied but still there's order, differentiation, conditionality, and these are different layers. So we're doing this kind of urbanism — parametric urbanism — and look at this, we can take two conditions and instead of modernist throwing three more different conditions at it, we can actually integrate and unify and redeem the existing pieces into a new unity which is highly differentiated in terms of the conditions. That's what we can deliver and we're doing this on many levels and we can do that as a single firm — not single hand because it's a large team generating a whole kind of million square meter which has kind of structure, order, this kind of explosion condition, variety and unity. I think it's quite potent but we couldn't do the whole city. We need the whole discipline to help and interlock and do a multi-author condition. These are multi-species ecologies in nature — the new species doesn't look like a piece of garbage floating around. It weaves itself in and a bird can navigate this because it's not a garbage heap because there's logic. Where's the water? Where's the rich fauna? Where's the flora of a certain type? Where is it getting cold? So there's logic and we said all scripts — nature has scripts. This is a parametric model where everything is correlated and scripted according to parameters relative to the landscape, topography, conditions of water. We need the city to do like this — when a stupid bird can navigate this, we can certainly navigate it too and we could navigate a city which was much more complex but not this kind of garbage spill disorder. We tried experiments where you have 12 different students picking up pieces, all working from the tectonism repertoire with their own logics and differences — lots of different but also a lot of unity and coherency. That's the kind of model I would love to offer. That means we have to stop this insane fragmentation of our discipline into minimalists and pragmatic modernists and neo-historicists and neo-postists and neo-deconstructivists. All of that has to converge otherwise we're going to keep delivering this garbage spill.
These are some experiments we're doing for London and of course we need the freedom, we need the true entrepreneurial freedom to weave these collocation synergies and come up with these typologies, plan us out of the way, preconceived, use different allocations out of the way, and us as a discipline upgrading to deliver. That's where we're failing — we're failing on the political front and we're failing on the disciplinary front. And I think this is what we should be doing. These are the kind of spaces and finally we're also working — our job as well, Parametricism is also working in the metaverse where we will have the internet transformed into real-time virtual cities, environment interaction spaces where you collaborate, communicate, conference. That's the kind of thing we have been doing and we do in digital twins of our offices and this I think is important. It could be interacting and interfacing with physical environments that will be co-designed. This is our job and it's going to be an avalanche of job for everybody here, in particular for young and tech-savvy architects. We're going to do metaverse environments. We will have double lives — half of our time we will be in virtual scenarios but they should be huge screen and lots of virtual participants participating and interacting. Every meeting is going to be hybrid. And so we've designed a number of conditions like schools of architecture, schools of design, office conditions and knowledge economy hubs and Mettopia which I just been flushing through is also actually an ecosystem for all the designers, architects, fashion designers, brands, conferences like this one, balancing coming into a virtual environment but using these kind of techniques of correlation, identity creation, paths for that. So this is where I would leave you with.
