We can summarise the history of computation in architectural design in terms of five eras: the 2D drafting era, the 3D modelling era, the building information modelling (BIM) era; the design computation (algorithmic) era; and more recently, the machine learning era. These eras are recognisable but overlap in practice and represent a fundamentally different way of thinking.
All tools modify the gestures of their users, and in the design professions this feedback often leaves a visible trace: when these traces become consistent and pervasive across objects, technologies, cultures, people, and places, they coalesce into the style of an age and express the spirit of a time.1
Era 1: 2D drafting
The first era of computation mimicked drafting, documenting pen drawings, sketches, and blueprints. The original Computer-Aided Design (CAD) system was Sketchpad, developed by Ivan Sutherland in 1963.2 Sketchpad was a constraint-based model – the manipulation of one geometric element in turn modified other geometric elements. However, it would take a further 20 years to make this technology affordable and accessible to a broader audience with AutoCAD released in 1982 – continuing the practice of representing buildings as multiple 2D drawings. As Aish describes, 2D drafting is a travesty of Sutherland’s original intentions.3
Era 2: 3D modelling
The second era of computation, also known as the first digital turn, emerged in the 1990s and duplicated model-making. The era was characterised by ‘blob’ architecture, most notably by architects such as Greg Lynn, NOX (Lars Spuybroek), and Frank Gehry.
However, by the early 2000s, design professionals started to lambaste the digital blob as the most conspicuous symbol of an age of excess, waste, and technical delusion.4 The era coincides with the release of such software as 3D Studio Max, Rhinoceros and Maya.
Era 3: Building Information Modelling (BIM)
The third era of computation added physical properties to 3D models. It may be surprising to note that the BIM era started in the 1980s, before the 2D drafting era.5 The term ‘building model’ was first used in the mid-1980s: In a 1985 paper by Simon Ruffle,6 and later in a 1986 paper by Robert Aish 7– then at GMW Computers Ltd, developer of RUCAPS (Really Universal Computer Aided Production System) software – referring to the software’s use at London’s Heathrow Airport.8
The term ‘Building Information Model’ first appeared in a 1992 paper by G.A. van Nederveen and F. P. Tolman.9 However, the terms, including its acronym ‘BIM’, did not become popularly used until some ten years later when Autodesk released a white paper entitled ‘Building Information Modeling’10 in 2002. Prior to that, different software vendors used differing terminology for what is now accepted as BIM. Graphisoft, for example, used ‘Virtual Building’, Bentley Systems used ‘Integrated Project Models’, while Autodesk and Vectorworks used ‘Building Information Modeling’.11
Limitations of BIM
It is important to note that even with the emergence of 2D CAD, 3D modelling and BIM, the computer didn’t aid design, they aided documentation.’12 Dr Robert Aish elaborates on BIM’s limitations 13 14, claiming:
- BIM assumes that buildings are assemblies of components, but that does not necessarily imply that a designer conceives of a building in terms of such assemblies. This ‘component’ assumption forces the designer to think about micro-ideas (the components) before macro-ideas (the building form).
- BIM models require precise coordinates and dimensions. As a representation, BIM forces the user to be too precise too early in the design process…Effectively BIM confuses precision with certainty.
- The hard code functionality of the built-in system components is orientated towards conventional construction. This gives a productivity advantage if this approach is accepted, but to circumvent this approach requires additional effort, potentially inhibiting more experimental or unorthodox design exploration.
- BIM as a representation of the building concept may be too detailed for some forms of analysis, which require simple volumetric description and because of the level of detail the BIM model may be difficult to edit, and therefore inhibit design exploration.
- BIM is a technology, it is a methodology, but it is not a philosophy of design.
Era 4: Design computation (algorithmic)
The design computation era, which began in the late 1980s and early 1990s, saw architects designing not the specific shape of the building, but a set of principles encoded digitally as a sequence of parametric equations. The designer no longer directly models the building: instead, they develop a graph or script whose execution generates the model.
As John Frazer describes 15, when it came to parametric architecture, the concept and use of the term predate the feasibility of using actual computational processes and appears to have originated from the Italian architect Luigi Moretti in the 1940s.16 Other authors support the claim that algorithms have been used implicitly in architecture for centuries prior to the digital age. Terzidis 17 for example postulates that while the algorithm is often associated with computer science, the use of instructions, commands or rules in architecture are, in essence, algorithms. Hersey and Freedam 18 for instance, were able to ‘detect, extract, and formulate rigorous geometric rules’19 by which Palladio conceived his villas. The work of Antonio Gaudi is also essentially parametric.
While the early explorations of the (digital) design computation era date back to the 1960s and 1970s with the early pioneering works of Nicholas Negroponte and Chuck Eastman among others, it was not until 2003 with the release of Generative Components and later in 2007, with the release of Grasshopper, did design computational take off.
Two projects completed in the early 1990s heralded the new possibilities of the design computation era: The Vila Olimpica Complex in Barcelona (1992) by Frank Gehry which used CATIA, and the International Terminal at Waterloo Station in London (1993) by Grimshaw which used I/EMS mechanical modelling software.20
Era 5: Machine learning
The fifth age of computation has begun – not coincidentally alongside the dawn of the fourth industrial revolution. It is the era of machine learning which combines a range of algorithms, pattern recognition, neural networks, generative design, artificial intelligence, and distributed computation to change how we make things.21
|1st||C18th & C19th||Used water and steam power to mechanise production.|
|2nd||1870 – 1914||Used electric power to create mass production.|
|3rd||1980s –||Used electronics and information technology to automate production.|
|4th||2010s –||Characterised by a fusion of technologies that is blurring the lines between the physical, digital, and biological spheres.|
Whereas previous computation eras used new technology to implement the old science we knew, now, to the contrary, we are learning that computers can work better and faster when we let them follow a different, nonhuman, postscientific method.22 Machine learning follows three main steps: training, analysis and application.
Computation is the only resource that has consistently dropped in price and vaulted in quantity and quality. As computation accelerates, algorithms will become more sophisticated and pack more intelligence into every action. Digital tools will harness more complexity and simplify work into higher-order activities. Computation will increase options but reduce complexity. Designers will need to educate themselves continually, and they will need to cultivate a new literacy of thinking in systems.
1 Carpo, M. (2017). The Second Digital Turn: Design Beyond Intelligence. The MIT Press, Cambridge, p. 55.
2 Sutherland, I. (1963). Sketchpad: A Man-Machine Graphical Communication Systems, AFIPS Conference Proceedings, vol 23, pp. 232-238.
3 Aish, R. (2013). First Build Your Tools. In Inside Smartgeometry: Expanding the Architectural Possibilities of Computational Design, Peters, B. & Peters, T. (eds). Wiley, Chichester, p. 41.
4 Carpo, M. (2017). The Second Digital Turn: Design Beyond Intelligence. The MIT Press, Cambridge, pp. 4-5.
5 Aish, R. (2013). First Build Your Tools. In Inside Smartgeometry: Expanding the Architectural Possibilities of Computational Design, Peters, B. & Peters, T. (eds), Wiley, Chichester, p. 43.
6 Ruffle, S. (1986). Architectural Design Exposed: From Computer-Aided-Drawing to Computer-Aided-Design. In Environments and Planning B: Planning and Design, March 7, pp. 385-389.
7 Aish, R. (1986). Building Modelling: The Key to Integrated Construction CAD. CIB 5th International Symposium on the Use of Computers for Environmental Engineering related to Building, 7–9 July.
8 Eastman, C. et al. (2008). BIM Handbook: A Guide to Building Information Modeling for Owners, Managers, Designers, Engineers, and Contractors. John Wiley, New Jersey.
9 Van Nederveen, G. & Tolman, F. (1992). Modelling Multiple Views on Buildings. Automation in Construction. 1 (3), pp. 215–224.
10 Autodesk (2002). Building Information Modeling. Autodesk, Inc., San Rafael.
11 Quirk, V. (2012). A brief history of BIM.
12 Wujec, T. (2017). The Future of Making. Melcher Media, London, p. 88.
13 Aish, R. UK Dynamo User Group.
14 Aish, R. (2013). First Build Your Tools. In Inside Smartgeometry: Expanding the Architectural Possibilities of Computational Design, Peters, B. & Peters, T. (eds). Wiley, Chichester, p. 43.
15 Frazer, J. (2016). Parametric computation: History and future. In AD: Parametricism 2.0. John Wiley &Sons, London, pp.18-23.
16 Bucci, F. et al. (2002). Luigi Moretti: Workings and writing. Princeton Architectural Press, New York.
17 Terzidis, K. (2006). Algorithmic Architecture. Elsevier, Oxford, p. 39.
18 Hersey, G., & Freedman, R. (1992). Possible Palladian Villas: (Plus a few instructively impossible ones). MIT Press, Cambridge.
19 Terzidis, K. (2006). Algorithmic Architecture. Elsevier, Oxford, p. 21.
20 Kolarevic, B. (2013). Parametric Evolution. In Inside Smartgeometry: Expanding the Architectural Possibilities of Computational design, Peters, B. & Peters, T. (eds). Wiley, Chichester, pp. 50-59.
21 Wujec, T. (2017). The Future of Making. Melcher Media, London, p. 88.
22 Carpo, M. (2017). The Second Digital Turn: Design Beyond Intelligence. The MIT Press, Cambridge, p. 7.