Computational design is often seen as a juvenile field of research in the architectural discipline; a mere 20 years old. [1] However, the tools and techniques of digital design have a much longer legacy: modern CAD environments date back to 1963, with Ivan Sutherland’s pioneering computer program Sketchpad, and we have been questioning the creative potential of the computational medium since 1967, with Jasia Reichardt’s exhibition Cybernetic Serendipity. So before we dive into the current state computation and architecture, let’s revisit these two foundational moments to understand the historical context upon which the work contained in this book is built. By doing so, we can calibrate our understanding of computational design within a contemporary context, and, perhaps, rediscover potent ideas that still elude computer aided design today.

Early Human-Computer Interfaces

Although analog computers have existed for centuries, our relationship with modern day, digital computing developed from WWII era code breaking machines.[2] Early modern computers were primarily used as tools for automation by a specialized group of scientists and mathematicians. Machines, such as the ENIAC, were developed to rapidly calculate mathematical tasks in pattern recognition and physical simulations. These computers enabled lengthy repetitive tasks, such as encoding or decoding messages and calculating firing tables or missile trajectories, to be executed more reliably, precisely, and quicker than a human counterpart.[3] Though the mechanics and methods of computation developed rapidly in the subsequent decades, the perception of computers as automation machines remained engrained as a fundamental typology for the design and use of computers.

Figure 1 – Yershóv Diagram (1963) Director-agent model of human-machine interaction.

Figure 1 – Yershóv Diagram (1963)

Director-agent model of human-machine interaction.

In 1963, A.P. Yershóv described this automation typology as the director-agent model of interaction. In this one-way “closed chain of transfer and processing [of] information”, the human director commands a machine agent to execute a directive.[4] This is articulated in the Yershov Diagram, where we see the thinking human, sensing and sending information to the machine on the right (fig.1). The machine is a closed black-box with a portal for incoming and outgoing information. In the diagram, the onus is on the human to distill the creativity and imagination of the mind into simplified commands that the machine can process. Once processed, the results are sent back to the human, returning to the mind for creative interpretation.

Today, our primary means of interacting with CAD programs is still this director-agent model from the 1960s — the command-line, the mouse, and even the script editor are all input devices to give a computer explicit directions to execute. This command-based approach to human-computer interaction is very useful when a desired outcome is well defined, like in tasks of automation. When the human understands what information to send, the speed, accuracy, and robustness of the computational machine can be leveraged to execute complex tasks. However when an objective is ambiguous, like in the nebulous process of design, this communication model breaks down. If the human does not know what information to send to the machine, Yershóv’s ‘closed chain’ of communication stagnates. The scope of the machine’s usefulness, therefore, is limited by the mind of the human.


In contrast to Yershóv’s model of human-machine interaction, Ivan Sutherland’s Sketchpad (1963) made it possible for “man and a computer to converse rapidly through the medium of line drawings.”[5] Through Sketchpad, Sutherland translated the complexities of computation into an interactive and visual interface. Its interface was one of the first that separated programming a computer from using a computer, and it demonstrated how computing could become accessible to graphic-based professionals, such as artists, architects, and designers.

Sketchpad’s interface was composed of two separate components: the light pen and a dock of push buttons (fig.2). The light pen, a predecessor to the mouse, was used for coordinate input when drawing graphics and demonstrative input for pointing to and selecting graphics. The dock of push buttons were used to switch between drawing functions, geometric operations, and constraint behaviors. By coordinating light pen movements with push button commands, Sutherland was able to package the program’s complex code into hidden subroutines dynamically triggered by these tangible input devices. Therefore, a user could directly manipulate the contents of the program without writing code.

Figure 2 – Ivan Sutherland’s Sketchpad (1963)

Figure 2 – Ivan Sutherland’s Sketchpad (1963)

Sketchpad enabled a user to ‘sketch’ 2D and 3D graphics directly on the computer screen and is recognized as the first Computer-Aided Design (CAD) tool.[6] Its sketches are not simply digitized drawings, however. The graphic objects in a sketch are embedded with layers of topological intelligence — internal data structures that can store complex relationships. For example, rather than defining a line by two endpoints, it can also be defined by its relationship to other graphic objects: parallel to, perpendicular to, same length as, and so forth. These topological structures allow users to add design constraints within their sketch. Design constraints define a framework for what the user intends to create, rather than what they explicitly create in the program environment.

Communicating through these design constraints gave the computer an understanding of a user’s intentions, not just their actions. Sketchpad could interpret and readjust a graphic model to fit what the user intended to make, and not just what they directly input. Therefore, a user could begin with a vague idea of the final graphic object, then collaborate with the computer to layer in design constraints and iteratively refine the desired output. Figure 3, for example, shows Sutherland quickly drawing the arcs and lines of rivet-like graphic object (left). Rather than carefully drafting the graphic object, he layers a series of design constraints that communicate to the computer what he intended, but didn’t actually draw. Sketchpad then interprets and readjusts the graphic object to match Sutherland’s intention (right).
Figure 3 – Before (right) and After (left) applying design constraints to a graphic object in Sketchpad. Demonstration video shown here.

Figure 3 – Before (right) and After (left) applying design constraints to a graphic object in Sketchpad.

Demonstration video shown here.

Sketchpad illustrated a vision that anticipated and inspired the future of human-computer interaction, and many of its concepts have been adopted and improved on over the last half century. The Graphical User Interface (GUI), Object Oriented Programming (OOP), and Computer-Aided-Design (CAD) were all developed on foundations that Sutherland built through Sketchpad.[7] What has yet to be improved on, however, is Sutherland’s notion of “conferring with our computers”. Sketchpad proposed a way for the computer to extend the human imagination. Through its constraint system, it could differentiate between a user’s intentions and their actions: the program was able to internalize an abstract idea of what the user wanted to achieve, and it could interact and collaborate with the user to reach that goal. In a rudimentary way, Sketchpad was an environment where man and machine could have a shared state of mind. This connection between human and computer made it possible for an abstract idea to develop into something initially unforeseeable by the human.

Cybernetic Serendipity

The term cybernetic is derived from Greek for “steersman,” and was appropriated for the meta-science Cybernetics in the mid-twentieth century. Norbert Weiner formalized the field in 1948 as “the scientific study of control and communication in the animal and the machine.”[8] Cybernetics provided a model for understanding complex behaviors as simplified systems of regulatory information exchange. It reduced ‘the animal and the machine’ down to the messages and signals it sent through an environment. As Wiener writes in The Human Use of Human Beings:
When I give an order to a machine, the situation is not essentially different from that which arises when I give an order to a person. In other words, as far as my consciousness goes I am aware of the order that has gone out and of the signal of compliance that has come back. To me, personally, the fact that the signal in its intermediate stages has gone through a machine rather than through a person is irrelevant and does not in any case greatly change my relation to the signal. [9]
Here, Wiener describes a man-machine system, where communicative fields replace the physical or biological boundaries of each entity. Man and machine become a part of a single continuum in an environmental context, blurring the dichotomy between the natural and the artificial.[10]

This ambiguity was the subject matter of Cybernetic Serendipity, an exhibition organized by Jasia Reichardt twenty years later, in 1968. For the exhibition, Reichardt invited a hybrid group of artists, engineers, mathematicians, and architects to submit work that demonstrated the creative possibilities engendered by contemporary technology.[11] The collection of computers, electronics, art, poetry, and machines translated scientific advances in cybernetics into sociocultural speculations on nascent relationships between man and machine.

The exhibited artifacts proposed ways for computational machines to form dynamic and interdependent bonds with humans — not as tools for automation, but as a medium for connection. Chuck Csuri’s Sine Curve Man (1967), for example, was one of the first examples of an artist using the computational medium to extend there own creativity. The plotter drawing was generated by programming an IBM 7094 with the X and Y point coordinates of the outline of a man. The Y coordinates of the original outline were then successively altered by a sine function, while the X coordinates remained constant. The resulting oscillated portrait was then plotted with a Calcomp 563 drum plotter (fig.4).
Figure 4 – Sine Curve Man by Charles Csuri (1967) Created with an IBM 7094 and Calcomp 563 drum plotter.

Figure 4 – Sine Curve Man by Charles Csuri (1967)

Created with an IBM 7094 and Calcomp 563 drum plotter.

Unlike Sutherland’s Sketchpad, Sine Curve Man was generated using very conventional computer interfaces and programmed with standard punchcards. This portrait, however, is the artifact of a profound new relationship between artist and computational platform. The cybernetic serendipity of Sine Curve Man is derived from the artist and machine’s connection throughout the process of creative discovery. As Csuri writes:
The question is not one of hand skills versus machine skills. The problem is one of conception, wherein the mathematical transformations made possible by the computer present a new dimension to art. [12]
The connection between artist and machine occurs at an abstract level, through machine language. Csuri grafts the conceptual framework of the artist onto the logic of the plotter to create a shared state of mind between man and machine. By binding his creative output to the underlying abstractions of the plotter’s processes, both artist and computer/plotter collaborate with equal agency in informing the final image. The specific graphic output, be it Sine Curve Man or Cosine Curve Man, is somewhat arbitrary. What remains important is how artist and computer combine to extend the limits of human creativity.
Figure 5 – Reconfiguring the director-agent model to the agent-agent model of human-machine interaction.

Figure 5 – Reconfiguring the director-agent model to the agent-agent model of human-machine interaction.

Over the past half century, the development Computer-Aided Design has ineffaceably altered the design disciplines; from pedagogy to practice, CAD tools mediate how and what we create. Yet despite the relative ubiquity of digital design, our means of interacting with these computational tools have remained largely unchanged since the 1960s. We still operate, rather than collaborate, with our digital design environments through WIMP (Window Icon Menu Pointer) interfaces.

The 1960s gave us visionary prototypes. Both Sketchpad and Sine Curve Man demonstrate that it is possible for a computer to participate in the process of design, and not just be used to execute a design. They forego the director-agent model for a cybernetic model of human-computer interaction (fig.5). As a result, each are able to craft a shared state of mind between human and computer: Sketchpad provided the computer with a model of the user’s abstract intentions, whereas Csuri was able to infuse an archetype of artistic expression (the portrait) with the essence of computation (the algorithm) when creating Sine Curve Man.

As we continue to explore new frontiers in computational design, it is important to understand the historical context in which we create; not only to ensure that we are building new knowledge (and not just than reconstituting existing knowledge,) but to also challenge assumptions about how a technology should behave or perform.


  1. Mario Carpo. The Digital Turn in Architecture, 1992-2002. (Chichester: Wiley, 2013), p. 8.
  2. Jasia Reichardt. Robots: Fact, Fiction and Prediction. (New York: The Viking Press, 1978), p 35.
  3. Saul Rosen. 'Electronic Computers: A Historical Survey'ACM Computational Survey. 1, 1 (March 1969), 7-36. DOI=10.1145/356540.356543
  4. A. P. Yershóv. ‘One View of Man-Machine Interaction’, J. ACM12, 3 (July 1965), 315-325. DOI=10.1145/321281.321283
  5. Ivan E. Sutherland. ‘Sketchpad: a man-machine graphical communication system’, Proceedings of the May 21-23, 1963, spring joint computer conference (AFIPS ‘63 (Spring)). ACM, New York, NY, USA, 329-346. DOI=10.1145/1461551.1461591
  6. Gabe Johnson, et al. ‘Computational Support for Sketching in Design: A Review’, Foundations and Trends in Human-Computer Interaction. 2, 1 (January 2009), 1-93. DOI=10.1561/1100000013
  7. William Buxton, et al. ‘Interaction at Lincoln laboratory in the 1960's: looking forward — looking back.’ CHI ‘05 Extended Abstracts on Human Factors in Computing Systems (CHI EA ‘05). ACM, New York, NY, USA, 1162-1167. DOI=10.1145/1056808.1056864 
  8. Norbert Wiener. Cybernetics: Or,  Control and Communication in the Animal and the Machine. 2d ed. (New York: M.I.T. Press, 1961,) p.10. 
  9. Norbert Wiener.The Human Use of Human Beings: Cybernetics and Society. (Boston: Houghton Mifflin, 1950,) p. 16-17.  
  10. Katherine Hayles. How We Became Posthuman: Virtual Bodies in Cybernetics, Literature, and Informatics. (Chicago, Ill.: University of Chicago Press, 1999).  
  11. Jasia Reichardt.  Cybernetic Serendipity: The Computer and the Arts. (New York: Praeger, 1969).
  12. Charles Csuri and James Shaffer.  ‘Art, computers and mathematics.’ Proceedings of the December 9-11, 1968, fall joint computer conference, part II (AFIPS ‘68 (Fall, part II)). ACM, New York, NY, USA, 1293-1298. DOI=10.1145/1476706.1476759