FOUR CHAIRS and all the others
A Three Dimensional Narrative
The project Four Chairs and all the others opens up the possibility of an alternative understanding of design. Rather than offering yet another thesis in support of linear design development, it emphasizes design’s polysemantic nature by understanding its processes in terms of an open field of possibilities. Design processes not only explore physical limitations of space, but also react to contemporary social and cultural phenomena. In order to explain the idea, specific techniques are used to replace simple design concepts with a series of parallel narratives, thus provoking new and unexpected situations. The primary interest of this project is to explore the intersection of different domains of human insight, especially regarding design, culture and information sciences.
The prefix Eigen is commonly used in linear algebra, in compounds such as eigenfunction, eigenstate, eigenvector. It comes from the German word eigen which means “one’s own, proper”. The basic tool for the design of the population of chairs to be investigated in such a way – i.e. “all the others” – is the Principal Component Analysis algorithm. It is a standard tool for contemporary data analysis that has been adapted in various applications according to diverse needs, from neuroscience to computer graphics, and now begins to be applied in the field of design. Principal Component Analysis reduces a given data set to a set of principal components, i.e. eigenvectors. The key feature of this algorithm is the intersection and interconnection of all data, whose result adapts and changes according to the required point of view, i.e. according to interpretation attributed to the problem.
EigenChair is a concept that results from the effort to design a chair that continues the genealogical orders of designed chairs, and yet is carrying a potentiality of all the chairs that might be created in the future. EigenChair is not an ideal chair in the sense of pureness or prototypicality. It is real (and not ideal) in the sense that it has a history, it originates and becomes, it must be regarded in the context of populations of chairs from which it evolves, and in the sense that it can be modeled by empirical experimentation (by observing and testing). So it is a real chair, and yet it is an abstract chair!
The interest of this project is to show strategies and concepts for designing with the use of information technologies. How can we engage with objects once they take an abstractly modular form, and their manifest materiality is diffused into a set of data? What are the potentials of data driven design?
Alternative Understanding of Design
Design Approach
Radical views of the world and of society are today mediated through advanced technological systems. Thanks to – or perhaps due to – such circumstances, design seeks new ways of thinking and conceptualizing, as well as of producing objects and inciting feasible. The ‘information-alization’ of the society, and the scope of applicability of computer-aided design tools are opening up a whole range of new manners of how to perceive the temporality and spatiality we inhabit. Algorithmic design is based on new parameters: design of ideas, narratives, procedures, populations, digital production, and new understandings of materiality. Generative design methods drive us to create and modify rules and systems, such that we are generating abstract machines: the products are not items of a set, but instances of a population that are one in kind, that of an abstract object. The designer therefore does not manipulate the “artifact” itself, but rather the rules and systems that allow for generating and producing it. The emphasis is no longer on the creation of physical objects, but on conceiving meta-objects in the possibility space of symbolic forms.
Recycling Information
The postmodern condition equips us with a set of critical, strategic and discursive practices which, as their main tools, use concepts such as difference, repetition, simulacrum, hyperreality in order to destabilize modernist concepts such as identity, linear progress of history, or unambiguity. In contrast to such a reactive point of view, an emerging condition which we call ‘pre-specific’ ceases to focus on the representation or identification of existing ‘truths’, and instead guides its interest to the filtration of attractive and promising approaches out of the plenitude of information. In order to avoid postmodernist tautological nihilism, the ‘pre-specific’ paradigm approaches the abundance of information in an active manner. This paradigm also operates within the field of design. But it puts no longer the object into the focus of its investigation and research, but an object’s characteristics, features, relations, ratios, structures, and its indexical context. The information age enables a redefinition of postmodern techniques such as collage, assemblage or bricolage, all of which define an object by collecting and reassembling various aspects and fractal components. The newly created abstract object is now a fusion of different objects’ constitutive data, but it is also completely unique and independent in the forms it can take from any one object in particular. The project Four Chairs and all the others – EigenChair is an example of digital recycling: it brings information and data of chairs into new manners of circulating, accumulating, integrating.
Elitism And Exceptionalism Of Singular Object Vs. Individual Populism Of Generic Objects
So far, design understood its practices as dealing with individual objects, their typicality and their specificity. Design was interested in the invention of “ideal objects”, which are to be original, and yet specific. Such an approach was closely related to the modernist paradigm. Today, however, the emphasis is moving from designing ideal objects to designing the ideality of real objects – the ideality in reference to which an object can be designed as singular and generic instead of original and specific. The new paradigm changes the designer’s relation to an ideal, static and original object, by putting an emphasis on conceptualization, interaction of the components, systems, and processes within the referential framework of an object’s ideality. What was once the design of a perfect, unique object featuring specific materiality is today the design of a population of objects featuring any materiality. Instead of a specific object, the designer creates an algorithm. Elitism and exceptionalism contained in the idea of a singular object is replaced by “individual populism” of generic objects. The key role in design is taken over by generative systems (syntaxes and grammars) that offer methodologies and theoretical world views as frameworks that distribute processes by multiplication, rather than by unification. The design process becomes an abstract definition of algorithms. Hence in this project, the focus was not on designing a “perfect” chair, but on engendering a whole population of chairs. Instead of creating a parametric master model, indexes of all objects are correlated to a framework of a possibility space – to a Pre-Specific model.
Imposed Materiality
In generative object design, the particular materiality of an object is not a precondition for its final manifestation. The choice of materials to work with has so far served as the basis for determining the design process, defining the expected execution of details, connections and textures. Today, generative system design enables the imposition of materiality to each instance of an abstract object. The form, not anymore complimentary to certain material, can now be attached to it by mere use of intellectual control. Therefore, the objects, previously described by fixed geometries, can now be variously described by relative geometries that can be rendered into reality in any materiality via 3D printing.
Designing Narratives
By rethinking the notion of “good design”, one comes to the conclusion that design is just a tangible fragment of reality, which narrates one of the many stories that surround us. Design never appears in silence. What we call “good design” nowadays is imbued with a series of narratives constructed by different discourses: formal, ideological, psychological, theoretical… It is only one part of the design process that is constituted by object’s material and formal aspects, while most of it is built upon stories that describe the object, and upon the individuals who transfer the stories or identify with them. Therefore, besides designing an object, it is also necessary to design a narrative that defines the objects ambition in terms of how it will become meaningful.
The research focus of the project Four Chairs and all the others is the design of a chair that does not carry on the heritage of originally iconic or functional pieces of furniture, but a generic heritage that cultivates information about “all chairs ever created“. For this, the term EigenChair is used – to describe a sum of ideas, and the ‘realities’ of particular chair designs that are elected as actors in the design narrative. The algorithm database contains a large amount of “other chairs”. Their fusions enable an infinite variety of possible results. In order to achieve a certain control over the results, out of “all other chairs” we have chosen four particular chairs that will provide the basis of recognizability in the dramatization of the object in the particular narrations. Fusions of characteristic parts of those four chairs with all the others are defined by user made maps that define the transformations, upgrade the performance of the Principal Component Analysis tool, and enable the control of the result. The project Four Chairs and all the others has elected four iconic chairs: Thonet’s Chair No.14, Wire Chair by Charles and Ray Eames, Panton Chair, and Ghery’s Wiggle Side Chair. Their main mutual link is specificity and uniqueness of the materials, and the respective technological innovation in the context of which they had been designed. It is the richness of meaning and historical references of these examples that are responsible for enabling us the further creation of analogies, stories and narratives, which, in turn, fertilize the viewer’s active participation in the process of visual representation.
Multidimensional Vector
Technical Approach
The project Four Chairs and all the others deals with options of manipulating data, and thereby engenders new objects. It takes a whole library of chairs as its starting point, that is, their geometric and spatial characteristics along with their historical importance and their narratives. By using open source 3D models of chairs from Google warehouse, their geometry is appropriated through a set of algorithms, on which the Principal Component Analysis algorithm is applied to calculate fusions, mergings and manipulations from the input information, from which new objects can be generated and produced. The result is a population of objects that are over-coding cultural and historical space-time relations through the imposition of logistic networks. The final objects are entirely a product of mathematical and logical thinking, designated according to a particular aesthetic sensibility (mine). The identity of the object is engendered by pure intellect, and contingently rooted in historical and cultural legacies. The main algorithm, which technically organizes the whole project, is the Principal Component Analysis algorithm.
Logical steps
The initial step was to normalize and prepare the data of all the chairs. In this case study, due to computational limitations, a total of 12 chairs were used as a testing data set. All data had to fit in the same bounding box, and mesh vertices were equally distributed throughout the mesh.
The whole procedure consists of three main parts. The first part is the Algorithm for Voxelizing Polygon Meshes. This algorithm transforms each mesh into a voxel-based object defined by a one-dimensional numerical array list, i.e. a multidimensional vector. In case of the highest resolution, each chair is represented with 2,788,875 values. Each value marks the distance between the given voxel and the closest mesh vertex. Values for each chair are exported as separate txt files, in order to reduce computing time of the main application.
The second part is the Algorithm for Morphing Chairs. The base of this algorithm consists in the multidimensional vectors generated by the Principal Component Analysis. The goals of Principal Component Analysis are (1) to extract the most important informational aspects from the data set, (2) to compress the size of the data set by keeping only the important informational aspects, (3) to simplify the description of the data set, and (4) to analyze the structure of the observations and the variables. In order to achieve these goals, Principal Component Analysis computes new variables, called principal components or Eigenvectors, which are obtained as linear combinations of the original variables. The first principal component is required to have the largest possible variance. The second component is computed under the constraint of being orthogonal to the first component, and thus needs to have the second largest possible variance. The other components are computed likewise.
According to the size of the initial bounding box, a voxel-based space is created. Each voxel receives values from txt files exported in the first step. With the use of Principal Component Analysis we can represent each chair by using only a set of so-called eigenweights, e.g. (-5673, -85184, 50, -25533, 31594). By changing the values of the principle components, i.e. the eigenweights, we are able to achieve linear transformations between all the chairs.
The third part is the Algorithm for Mapped Morphing. It is an upgrade from linear Principal Component Analysis transformations to mappings of nonlinear transformations. An RGB map, in which each color represents a particular chair, is projected onto the voxel-based space. This enables us to define and control the nonlinear transformations and fusions of three different chairs into a new one. Thus created, chairs can be used again as input chairs for the second step, and achieve a new nonlinear variability.
The rest of the algorithms served to prepare the data for Principal Component Analysis and to help with their final visualization. Furthermore, an important role was played by a series of open source libraries, especially the Marching Cubes Algorithm, responsible for generating watertight mesh objects ready for 3D printing. All codes were written in Java programming language.
Having in mind the thoughts presented in an earlier part of this text regarding referentiality and recycling, it is important to note that the algorithms used in the project, e.g. the Principal Component Analysis algorithm and Marching Cubes Algorithm, are already and widely in practice. They are thoroughly adapted, functionally redirected, recycled, to fit the needs of design in this particular project.
Articulating Indexes
Theoretical Approach
Information
The key term that best describes and corresponds to what characterizes, overall, contemporary society and science is information. Information technologies are entering all spheres of society: from the ways in which we organize our everyday life, to the ways in which we think about natural sciences and humanities. This view suggests the inadequacy of understanding human environments in pre-dominantly material terms and physical relations between energy and matter; in order to create a more comprehensive world view, analysis must take into consideration also information as a quasi-material category. At the same time, being surrounded by excessive amounts of information, any analysis requires a stable environment, which enables its observations and uses.
Reflections On The Real
It is impossible to comprehend or examine exhaustively what is to be considered as “real”, because such consideration depends upon the quantization and formalization of ideas. Hierarchies and the relations between originals and their copies, which is the key concern of materially oriented societies, have become almost completely irrelevant in an age in which virtual realities dominate human lives. Depending on the ways of our understanding and capacities of accepting the “unfamiliar”, we comprehend and legitimate what is to be considered as real. Brian Massumi is perceptive to this in a multifaceted way, by comparing Baudrillard’s interpretation of the reality-simulation, in which there is no division between the real and the virtual, with Deleuze’s and Guattari’s negation of the linear approach to the real. Such a non-linear approach to reality is supported by the vanishing of boundaries, and the influence of the virtual on the real.
„Baudrillard sidesteps the question of whether simulation replaces a real that did indeed exist, or if simulation is all there has ever been. Deleuze and Guattari say yes to both. The alternative is a false one because simulation is a process that produces the real, or, more precisely, more real (a more-than-real) on the basis of the real. “It carries the real beyond its principle to the point where it is effectively produced.” Every simulation takes as its point of departure a regularized world comprising apparently stable identities or territories. But these “real” entities are in fact undercover simulacra that have consented to feign being copies.” (Massumi, 1987).
Abstraction
The ‘Internet age’ is exactly such a condition, in which immaterial information is part of what we call reality. In other words, there is a peculiar reality proper to models, even if they are, necessarily, idealizations. In such a condition, the only way of dealing with information is abstraction, and it can be adequately used only by those who are, in a mass of information, able to define their contexts as flexible, adjustable fields of possibilities with polyvalent, and ultimately undecidable, meaningfulness. The project Four Chairs and all the others considers the creation of abstractions of objects to a degree that multiplies the manners in which objects can be manipulated beyond any definite bounds, and by this, it considers how new meanings can be provoked from the abundance of information. If objects – chairs, or entire populations of objects – are assigned an abstract expression, as multidimensional vectors (i.e. as a series of numbers in a line, as linearized indexes) they become very potent and can be manipulated in manifold manners . Such abstract objects, which consist in nothing else but indexes, are placed in a meta-space that contains the summation of the potentials of all the objects which are constitutive for this meta-space. Governed by the Principal Component Analysis algorithm, meta-space is able to correlate indexes of all objects, creating thus an open logistic network, an abstract possibility space. This marks the level of how objects might be articulated as an abstractly engendered kind, and it allows for the generation of entire populations of singularly-particular objects which all belong to the same generic “kind”. By looking at objects through the levels of their abstractions, we realize the potency of information (in meaning and shapes, with which we can work), but at the same time we realize the sheer emptiness that is proper to abstraction, when we regard it on the symbolical level of indexes alone.
Meaning, Context And Narrative
Post-traditional societies (societies that embrace modernization) offer new perspectives on old concepts to which new meanings are attributed, or which are judged critically, by negotiating their discursive contexts. The mass of information shapes our world: text, visual representation, music, money. However, the idea offered by the information theory pioneer Claude Shannon, namely that “information does not itself carry meaning but transmits messages”, has become rather liberating in the academic discourse: in carrying no meaning, information offers unlimited freedom of manipulation. It is important to emphasize that contextualization and the successive creation of narratives inevitably “fill in” the void of information (its constitutive meaninglessness). Contextualization and narration gain power by carefully gathering evidence (real data) for what they are meant to comprehend. At the same time, they take care that the larger contextualizations and stories in which they claim to be embedded, rely on the collective reality and memory of culture and history. It is also important to note that the process of contextualizing generic instances, by composing their proper narratives is happening before they are actually generated and produced. There is a whole world of possibilities from which one actualizes only a fraction. Yet the effects of such ‘reductionism’ are not to impoverish, but to maintain open the potential for novelty and for the unexpected. This project shows that design is able to manipulate predetermined potentials, while filling them, at time same time, with narratives. Design is not a part of the endless evolutionary process aimed at creating the next new ideal object, but a part of a defined context with chosen references, and their respective genealogies.
EigenChair – Data Driven Design
By using information manipulation and various spatial conceptions, algorithmic design approaches an object in a completely abstract manner, distancing it thus from its own immediate “reality”. In making the object extremely flexible for different interpretations and contextualizations, algorithmic design also contributes to the instability of its design process: lacking the resistance of material constraints, designing an object could easily be reduced to a formalistic geometry exercise. Therefore, a key feature of such an understanding of design is not only the definition of algorithms, but also the construction of parallel narratives around the object. It seems therefore inviting to re/turn to the postulates of the pre-Socratic philosopher Empedocles, who claimed that “nothing comes out of nothing and nothing disappears into nothing”. Such philosophical re/turn marks an effort to observe context and processes as more important factors for defining the object than those implicit in the Objectivism. The advantage of procedural design in our contemporary world is its ability to refer to partial summations of global knowledge, and to use it effectively.
This project tries to show – by conceiving and shaping the idea of a chair (object) for the early 21st century – the necessity of perceiving design through three equally important, interdependent aspects: design, theory and technology. Design is now data driven.
Bibliography:
Abdi, H. and Williams, LJ. (2010), ‘Principal component analysis’, Wiley Interdisciplinary Reviews: Computational Statistics, 2(4), pp. 433-459.
Aylesworth, G. (2013), ‘Postmodernism’ in EN Zalta (ed), The Stanford Encyclopedia of Philosophy.
Lorensen, WE. and Cline, HE. (1987), ‘Marching cubes: A high resolution 3D surface construction algorithm’, SIGGRAPH Comput. Graph., 21(4), pp. 163-169.
Massumi, B. (1987), ‘Realer than real’, Copyright no.1, pp. 90-97.
Sirovich, L. and Kirby, M. (1987), ‘Low-dimensional procedure for the characterization of human faces’, JOSA A, 4(3), pp. 519-524.
Terzidis, K. (2012), Algorithmic Architecture, Taylor & Francis.
Turk, M. and Pentland, A. (1991), ‘Eigenfaces for recognition’, Journal of cognitive neuroscience, 3(1), pp. 71-86.
Vegesack, A., Dunas, P. and Schwartz-Clauss, M. (1996), 100 Masterpieces from the Vitra Design Museum Collection, The Museum.
This project was produced as a part of a research on Data Driven Design at the Chair for Computer Aided Architectural Design ETH in Zurich.