The drawn or written error must be erased. I have rubbed my mistakes out on paper, scratched them off trace and Mylar and deleted them from the screen: each a methodological register of the representational act of forgetting.1 In 1987, in perhaps nothing more than an extended witty aside, Umberto Eco wrote Ars Oblivionalis: Sulla Difficoltà di Construire un’ Ars Oblivionalis.2 In the opening of this bracing tonic of an article, he parodies the Ars Memoriae of Cicero and Quintilian, among others: when you want to ensure you will remember something, associate it with a “monstrous bleeding image,” a mutilated body for instance, and take it up the stairs to the second floor of an elaborate memory palace and leave it in the third room on the right.3 To forget or erase this memory, imagine a man going into that same room and throwing the image out the window. However, this usually only makes you remember that you tried, very hard, to forget that thing which you now persist in remembering. Given that mnemotechnics is semiotic in its performance—the mutilated body (or sign) stands for that which is to be remembered—how, asks Eco, can the absence of a sign, or rather the presence of another sign, reverse this logic and cancel the memory it was wired to evoke? By definition, semiotics makes present what is otherwise absent, not the reverse. Much like the phenomena of remanence in magnetic memory in which data is never truly deleted, only overwritten, Eco concludes: “one forgets not by cancellation but by superimposition, not by producing absence but by multiplying presences.”4 Forgetting, or l’Oubli—there is no noun of equivalent weight in English, unless we consider the “oblivion” of oblivionalis or even more obliquely, “ignorance, as Beckett did in his translation of l’oubli—is not the performative opposite of Memory.5 Like Eco’s thwarted efforts to forget by throwing gory images out of windows, we are doomed to fail to forget. Alternately, we are doomed to never forget. As Aristotle distinguishes in On Memory and Reminiscence, our minds are either that of the quick-witted ilk, which although great at thinking, and by corollary targeted recall, are like hard stucco walls or running water (yes, Lethe is there) and thus incapable of securing an imprint of large data sets, or our minds are like that of the slow-witted, poor conversationalist who, though able to store vast arrays of data eternally, is cursed with Funes-like total recall and thus zero ability to generalize or truly think for that matter.6 We each know which we are.

Military operative using the Semi-Automatic Ground Environment (SAGE) light-gun to input information into, or delete hostile bombers from, the onscreen representation of US Cold war airspace in 1961. Image: The MITRE Corporation.

Military operative using the Semi-Automatic Ground Environment (SAGE) light-gun to input information into, or delete hostile bombers from, the onscreen representation of US Cold war airspace in 1961. Image: The MITRE Corporation.

But what of our drawings? How the architectural drawing forgets (or not), or how we install the act of forgetting in the production and amendment or revision of the architectural drawing is very instructive of what architecture would prefer to forget, but despite itself, persists in remembering. The representational error is revised, amended and corrected, with the rub of the eraser, the bite of the razor blade or, for the first time in 1961, the trigger of a gun. The Semi-Automatic Ground Environment (SAGE), which involved by far the largest computer program at the time and resources that dwarfed the Manhattan Project, was designed to detect incoming hostile bombers. The airspace defense system used a light-sensing gun to input information into real time onscreen representations of United States airspace. The electronic drawings on the monitor could be amended by the “light gun,” which when pointed at the offending blip and fired with the console’s “FIRE” button, removed Soviet atomic missiles from the real and virtual skies of the Cold War—correcting the “picture” back to what it should be. While the SAGE system, backed with unlimited resources of finance and fear in the face of potential nuclear attack, never produced a militarily effective system, it did produce a vast metaphoric dome of paranoia and a hermetic architecture of control, as well as invent, among other things: video display, magnetic core memory, graphic display and simulation techniques, networks, vastly accelerated processing speed (rather necessary given the brief) and the light gun.7 Well before video games, in SAGE Direction Centers dotted across the land, the erasing power of the gun met the nascent electronic drawing: “dim blue light from the consoles illuminated their interiors, known as ‘bluerooms,’ where operators used light guns to connect blips on video displays… the eerie calm of battle as an automated process for rational managers.”8

 SAGE’s eponymous “Blue rooms.” Image: The MITRE Corporation.

SAGE’s eponymous “Blue rooms.” Image: The MITRE Corporation.

The very next year the light gun reappeared in civilian-dress as the “light pen” in Ivan Sutherland’s still extraordinarily prescient and elegant program Sketchpad: A Man-Machine Graphical Communication System. A program, which inherited so much from the Whirlwind/SAGE research also developed at Lincoln Labs, and from which we have inherited all that is actually useful in electronic drafting. Sutherland, who did not waste his time with rendering, categorically stipulated that: “it is only worthwhile to make drawings on the computer if you get something more out of the drawing than just a drawing.”9

Operative (possibly Timothy Johnson) using a light-pen in to delete a line in Ivan Sutherland’s Sketchpad program run on a TX2 Console at MIT in 1962. Photo: The MIT Museum.

Operative (possibly Timothy Johnson) using a light-pen in to delete a line in Ivan Sutherland’s Sketchpad program run on a TX2 Console at MIT in 1962. Photo: The MIT Museum.

The light pen’s passage across the screen is tracked by a blinking light. Sutherland is at pains to explain that “light appears to be coming out of the light pen, like a flashlight, but it’s not, it’s coming up into it.”10 This pen is an eye.11 But the gun trope dies hard: during each complete display cycle a table exhibits “all the picture parts that could even be remotely considered as being aimed at,” and the cursor, which Sutherland describes carefully designing, is a cross-hair.12 The cursor on the screen—initially described as a “scope” in the 1964 MIT Science Reporter demo, then later, with exquisite hesitation, and possibly for the first time, as a “window”13 —faithfully tracks the movement of the pen but is curiously attracted to the light-lines or points as it approaches them. If it gets close enough it abandons the pen and latches on to them with a sudden leap. Paradoxically the metaphor of choice in this gravity-free but magnetized realm is the attractive force of gravity, not magnetism: “the bright dot will jump onto the line as I get close to it… it is much like a gravity field. At the end point there is even a high gravity field.”14 This gravity helps us not to make mistakes. As Sutherland explains several decades later:

I did a thing that I called ‘pseudo-gravity’ which said that around lines or around the intersection between lines that the light pen would pick up light from those lines and then the computer would do a computation saying: Gee, the position of the light pen now is pretty close to this line, I think what the guy wants is to be exactly on this line, and so what would happen is that if you moved the pen near to an existing part or near to an existing line it would jump and be exactly at the right coordinates.15

That is, the software “corrects” through the assumptions of approximation.  The TX2 console (1956), which first ran Sketchpad, just wants to help. This is made clear in the opening discussion of the demo film:

“John we are going to show you a man actually talking to a computer in a way far different than it has ever been done before.

“Surely not with his voice?”

“No he’s going to be talking graphically, he’s going to be drawing and the computer is going to understand his drawings and the man will be using a graphical language that we call Sketchpad…. You will see a designer effectively solving a problem step by step by step and he will not at the outset know precisely what his problem is, nor will he know exactly how to solve it. But little by little he will begin to investigate ideas and the computer and he will be in cooperation, in the fullest cooperation, in this work.” 16

However, should you make a mistake, despite the cooperative efforts of Sketchpad, you simply point the light pen at the line in question, the cross hairs lock-on, you press the ERASE button and “it’s gone.” Or, more cunningly, you undraw the line: “if you go backwards you erase it!”17

However, an existential problem presents itself that recalls Eco’s conundrum of the sign that is able to evoke, but not cancel, presence: how to delete things that are absent, that do not register their presence on screen? Sutherland’s struggle with this is poignant: “everything you put in a drawing had to have a representation on the screen because otherwise you couldn’t erase it. In order to erase something you had to point at it. And so the issue arose of what do things look like that otherwise don’t have an appearance—what does a constraint look like?”17 As the logic is extruded, a veritable comedy of errors emerges, of lines without end and rogue constraints adrift without variables:

If a thing upon which other things depend is deleted, the dependent thing must be deleted also. For example, if a point is to be deleted, all lines which terminate on the point must also be deleted. Otherwise, since the component elements for lines contain no positional information, where would these lines end? Similarly, deletion of a variable requires deletion of all constraints on that variable; a constraint must have variable to act on.18

Deletion too, it seems, must be supervised and corrected.

But it is very difficult to make a mistake in this drawing as the computer does not expect us to know the answer, not just yet. In the MIT Science Reporter demo, Timothy Johnson explains what it is to be a designer:

Say I am beginning a design. I have a very nebulous idea of what I have in mind. As I draw my part let’s say, on the scope, it reinforces what I have in mind; this is in general part of the design process. And as I apply design criteria stresses and so on, eventually I will know what the exact shape of this part is. I shouldn’t be required to draw the exact shape at the beginning. I really don’t know what it is. 19

Sketchpad redesigned not only the designer but the design process too: “construction of a drawing with Sketchpad is itself a model of the design process,” a “design process” whose legacy is such that it now, finally, no longer needs a designer.20 Nor does it need input exactitude: with Sketchpad “you don’t have to draw exactly at the beginning like you have to do in drafting.”20 Which is not to say the product is rough, approximate or in any sense a “sketch.” On the contrary, the corrective nature of what is essentially the beginning of predictive programing allows you to be “quite sloppy” when drawing and “get a precision drawing out at the same time.”20

1963 sketch by Ivan Sutherland explaining how to “draw” a line or a circle in Sketchpad with the light-pen showing how the program’s constraint logic ignores the erratic radius, accepting only the radius of the first point on the circumference and inputting only the degree of arc consequently drawn. Sutherland, Sketchpad, a Man-machine Graphical Communication System (Cambridge: MIT, 1963, p14).

1963 sketch by Ivan Sutherland explaining how to “draw” a line or a circle in Sketchpad with the light-pen showing how the program’s constraint logic ignores the erratic radius, accepting only the radius of the first point on the circumference and inputting only the degree of arc consequently drawn. Sutherland, Sketchpad, a Man-machine Graphical Communication System (Cambridge: MIT, 1963, p14).

Pseudo gravity, light pens and even the constraint satisfaction routine, which moved “whatever it had to move in order to satisfy the conditions that you set up as you did the drawing” had all been figured out, when Sutherland’s thesis supervisor (Sketchpad was his Phd research at MIT) said: “do circles.” The supervisor was none other than Claude Shannon. Shannon had agreed to take on the role, something he only ever agreed to for two other candidates (one being Sutherland’s brother), because he remembered that Sutherland had paid him a visit some ten years earlier as a high school kid in 1951. As Shannon had no doubt anticipated, circles proved difficult. Eventually a solution was found:

To draw a circle we place the light pen where the center is to be and press the button ‘circle center,’ leaving behind a center point. Now, choosing a point on the circle (which fixes the radius) we press the button ‘draw’ again, this time getting a circle arc whose angular length only [i.e. you don’t have to be able to draw a circle to draw a circle] is controlled by light pen position.21

Sketchpad will ignore our drunken radius, and use the light pen location only to decide how much arc to draw.

Figure from Alberti’s 1435 de Pictura explaining how to draw a circle in perspectival projection. Leon Battista Alberti, Della pittura e della statua di Leonbatista Alberti (Milan: Società tipografica de'Classici italiani, 1804, p 179).

Figure from Alberti’s 1435 de Pictura explaining how to draw a circle in perspectival projection. Leon Battista Alberti, Della pittura e della statua di Leonbatista Alberti (Milan: Società tipografica de’Classici italiani, 1804, p 179).

In the algorithm that is Alberti’s de Pictura, one also does not have to be able to draw a circle to draw a circle. Recognizing their inherent resistance to his system, even with the mediating use of the velo, Alberti offers his draughtsman largesse, a shortcut and license to fudge, in the case of “circular surfaces.” Should you choose to, you could of course always follow his instruction and patiently map an orthogonal projection of the curvature onto a perspectival grid:

But as it would be an immense labor to cut the whole circle at many places with an almost infinite number of small parallels until the outline of the circle were continuously marked with a numerous succession of points, when I have noted eight or some other suitable number of intersections, I use my judgment to set down the circumference of the circle in the painting in accordance with these indications. Perhaps a quicker way would be to draw this outline from a shadow cast by a light.22

The circular surface is the way-out. It is the analogue escape hatch from the pyramid of Alberti’s monocular constraining system—his perspectiva artificialis that transformed Erwin Panovksy’s “psychophysiological” space into mathematical space—a hatch that Sutherland’s own constraining system finally locked shut.23 Yet again we find the computer has inherited so much more from Alberti than simply the fenestra aperta inserted between our hand and the drawing surface: “the discovery was that there was in effect a computer between what I held in my hand and the drafting surface. And the idea that I could put computing power between where I held my hand and what the picture was was the powerful idea.”24 Sutherland installed (perhaps unconsciously) the software update: a constraining system with newly augmented corrective power between eye/hand and drawing—continuing Alberti’s project to disembody visuality. In doing so, what Sketchpad ultimately invented was a way to finally erase the one thing that needed no representation on the screen to be erased: our corporeal trace, our drunken radius, us.

  1. 1. I am indebted to Gergely Kovács and to our AA students with whom I have shared numerous conversations regarding erasure and deletion in architectural representation. ^
  2. 2. “Ars Oblivionalis: Sull difficoltà di construire un’ Ars Oblivionalis,” Kos 3, 1987, pp40-53. English version: trans Marilyn Migiel, PMLA, Vol.103, No.3, May 1988, pp254-261. ^
  3. 3. Ibid, 254. ^
  4. 4. Ibid, 260. Remanence refers to a magnetic palimpsest, the residual magnetism left in a ferrous material after a magnetizing field is removed. In order to effectively wipe magnetic memory Gutmann estimates we need to overwrite it, “cancellation by superimposition,” no less than 35 times for effective “data deletion.”  See: Gutmann, Peter, “Secure Deletion of Data from Magnetic and Solid-State Memory,” Department of Computer Science, University of Auckland, 1996.
 ^
  5. 5. Exactly how Beckett chose to translate his own work is always deeply instructive of the sphere of influence he deemed a word to hold within its native culture. In his translation of “Que Ferais-je Sans ce Monde? “ (“What Would I do Without this World?”), l’oubli becomes quite different in English, “ignorance,” as if while France looses itself to forgetting, the English may have nothing to forget in the first place. Samuel Beckett, Collected Poems in English and French (London: John Calder Publishers, 1977). ^
  6. 6. For readers yet to encounter Funes: following an accident, the protagonist of Jorge Luis Borges’ 1942 “Funes el Memorioso,” is physically unable to forget. Thus, as he involuntarily recalls every detail, he cannot ignore what doesn’t matter in order to generalize: a dog seen in profile at 3.14pm is for him utterly different from a dog seen front-on at 3.15pm. Without the approximating function of partial forgetting he cannot abstract and thus think. That is, Funes’ processing eye both remembered and saw like a computer. Borges’ ever prescient story is acquiring further resonance in contemporary conversations regarding Google’s struggle with the performance of generalization in its facial recognition algorithms, and most recently its work with Deep Dream. ^
  7. 7. As Paul N. Edwards argues: They must have known it was all a bluff, why else would the SAGE Direction Centers not have been installed in underground bunkers? Plus, the system was easily jammable and the test results frequently fudged. See his excellent The Closed World: Computers and the Politics of Discourse in Cold War America (Cambridge: MIT Press, 1996). ^
  8. 8. Ibid, 106. ^
  9. 9. Ivan Sutherland, “Sketchpad: A Man-Machine Graphical Communication System” (Technical Report 574, University of Cambridge Computer Laboratory,,2003), 110. http://www.cl.cam.ac.uk/TechReports/. ^
  10. 10. Sketchpad, lecture by Ivan Sutherland, 3 March 1994, Bay Area Computer Perspectives, No. 102639871. US: Sun Microsystems, 1994. ^
  11. 11. This eye with its photo cell is analogue and consequently, unlike that of Funes, able to selectively ignore: the input function of the light pen “uses the light pen optics as a sort of analogue computer to remove from consideration all but very few picture parts which happen to fall within its field of view, saving considerable program time.” Ivan Sutherland, Sketchpad(1963). In The New Media Reader, ed. Wardrip-Fruin and Montford (Cambridge: MIT Press, 2003), 115. ^
  12. 12. Ibid, 115. My emphasis. ^
  13. 13. MIT Science Reporter: Computer Sketchpad, John Fitch, Dir. Russell Morash, WGBH-TV, Boston (21 mins), 1964. Not only does this “window” have a “2 mile wide” blank piece of “paper” behind it (the design plane to come) we are reassured by the operative, Timothy Johnson, there are several pieces of paper available: does this mean we can make lots of mistakes? Why else would we need so much paper? ^
  14. 14. Ibid. ^
  15. 15. Sketchpad, lecture by Ivan Sutherland, 3 March 1994. Sutherland’s gravity system with its “latching” became the “snapping” now employed in almost all CAD software, whose metaphoric force of attraction, given contemporary GUI’s use of a magnet symbol, is, presumably, now magnetic. ^
  16. 16. MIT Science Reporter: Computer Sketchpad, 1964. There is so much more to say about this extraordinary description of man/computer cooperation: that the designer is to behave like an algorithm, acting step by step, never seeing (nor needing to even know) the big picture, that thus began the strange abdication of authorship which has become so pervasive on contemporary computational design software. ^
  17. 17. Ibid.  And as Sutherland, whose father was a draftsman, likes to point out: “of course you can erase lines and one of the nice things was you don’t get crummy bits of dust like you do on paper with a pen and pencil. You can erase nice and clean and tidy.” Sketchpad, lecture by Ivan Sutherland, 1994. Erasure via the undrawn line became in the 1980’s the baroque architecture of multilevel undo commands, the undostack which could reverse a whole strings of actions. ^
  18. 18. Sketchpad lines being Sutherland’s pioneering “rubber-band lines” – i.e. only defined by their end points. This erasing of the invisible is the origination of todays ‘purging’, we purge drawings ‘clean’ files and get rid of all the (invisible) no longer necessary deleted residuals. ^
  19. 19. MIT Science Reporter: Computer Sketchpad, 1964. ^
  20. 20. Sutherland, New Media Reader, 113. ^
  21. 21. Sutherland, New Media Reader, 111. ^
  22. 22. Leon Battista Alberti, On Painting, ed. Martin Kemp (London: Penguin Classics, 1991), 71. My emphasis. ^
  23. 23. “In order to guarantee a fully rational -that is, infinite, unchanging and homogeneous- space, this “central perspective” makes two tacit but essential assumptions: first, that we see with a single and immobile eye, and second, that the planar cross section of the visual pyramid can pass for an adequate reproduction of our optical image. In fact these two premises are rather bold abstractions from reality, if by “reality” we mean the actual subjective optical impression. For the structure of an infinite, unchanging and homogeneous space—in short, a purely mathematical space—is quite unlike the structure of psychophysiological space”. See Erwin Panovsky, Perspective as Symbolic Form, trans. Christopher S. Wood (New York: Zone Books, 1991), 29-30. ^
  24. 24. Sketchpad, lecture by Ivan Sutherland, 1994. ^

Architect, educator and writer, Francesca Hughes teaches at The Architectural Association and The Bartlett School of Architecture in London. She is editor of The Architect: Reconstructing her Practice (MIT Press, 1996), Drawings that Count (AA publications, 2013) and most recently, author of The Architecture of Error: Matter, Measure and the Misadventures of Precision (MIT Press, 2014). The work of Hughes Meyer Studio, an art/architecture practice, has been published by AA Files, AR, ANY, Art Forum, Harvard Design Magazine, Routledge, Monacelli and Wiley.