This is the text of the talk I gave at the MLA Convention in Philadelphia on 28 December 2004. An early version of this talk was given at the History of Material Texts workshop, and a version directed at electronic artists was given at ISEA. This talk mentions additional specific ways in which print interfaces may have affected the experience of Eliza/Doctor and Adventure, and it argues that an understanding of the interface and the material nature of electronic literature must be joined with a formal understanding of how computer programs work. If you'd like to correspond about the topic and correct or inform me about the use of print-based interfaces, please contact me: nickm at this domain.
I want to begin by thanking Peter Stallybrass and the Penn History of Material Texts seminar for the chance to present and discuss this work at an earlier stage, back in February.
The literary works I'm going to talk about today differ in some significant ways from the texts that have been discussed by the other panelists. For one thing, I'm going to discuss works that are "born digital," that is, developed originally for experience on the computer. They are not facsimilies of printed texts or electronic books, any more than Don Quixote is the print version of an oral story. Additionally, the two works I'm going to focus on are complex "born digital" pieces that are best understood as interactive computer programs, not as e-books, as chunks of text that are hooked together with links, or as multimedia presentations. But, for reasons that I hope will become clear, I do not think these works will be fully understood using only disciplinary approaches from computer science. I believe they have important literary aspects that call for humanistic approaches, including some from textual studies, and this is why I've presumed to present some of my investigations to you today, and why I hope to discuss them with you after my talk.
When scholars consider electronic literature, the screen is often portrayed as an essential aspect of all creative and communicative computing — a fixture, perhaps even a basis, for new media. The screen is relatively new on the scene, however. Early interaction with computers happened largely on paper: on paper tape, on punchcards, and on print terminals and teletypewriters, with their scroll-like supplies of continuous paper for printing output and input both.
By looking back to early new media and examining the role of paper (in both punning senses) we can correct the "screen essentialist" assumption about computing and understand better the materiality of the computer text. While our understanding of "materiality" may not be limited to the physical substance on which the text appears, that substance is certainly part of a work's material nature, so it makes sense to comment on that substance.
There were important screen-based systems early on — Spacewar, the first modern video game, developed at MIT in 1962; Ivan Sutherland's Sketchpad, also developed at MIT in 1962; Doug Englebart's NLS (oNLine System), developed at SRI and shown in the "mother of all demos" in 1968; and Grail, developed at the RAND Corporation in 1969 — but these were the high-budget exceptions to the rule. Most computer users, including those who were developing early electronic literature, did not have access to screens until at least the mid-1970s. I'll describe two early computer programs that were developed and experienced using an ink-and-paper interface:
The Eliza system was programmed by Joseph Weizenbaum in the mid-1960s. It was written in a language called MAD (Michigan Algorithm Decoder), using a package called SLIP (Symmetric LIst Processor) which Weizenbaum originally developed in 1963. Weizenbaum had an IBM 1050 in his office, a print terminal which featured a Selectric typewriter ball. He programmed Eliza for the IBM 7094, running the Compatible Time-Sharing System that was developed at MIT's Project MAC. This operating system was one of the first successful time-sharing systems, one of the first to offer email, and, apparently, the very first to offer text formatting capabilities.
Eliza was a general-purpose system for conversation, but the Doctor script has become inextricably associated with it. When running this script, Eliza impersonated a Rogerian psychotherapist. While Weizenbaum called this a "parody" of a psychoterapist — hinting that this academic research system had some sort of artistic or literary nature — Eliza could play the Doctor role well, if the user was willing to play along. The system had a tremendous impact on natural language processing, artificial intelligence, the relationship of computing and psychotherapy, and the ethical uses of computers. Janet Murray identifies Eliza/Doctor as the "moment in the history of the computer that demonstrated its representational and narrative power with the same startling immediacy as the Lumières' train did for the motion picture camera," and she names Weizenbaum "the earliest, and still perhaps the premier, literary artist in the computer medium."
Sherry Turkle offered an important discussion of the Eliza/Doctor phenomenon. She quoted two users who related to the system in different ways — one trying to trick it, another actually seeking help in sorting through her feelings — and presented the points of view of Weizenbaum (who thought the system should never be used for real psychothereputic purposes) and psychiatrist Kenneth Colby (who disagreed). However, Turkle did not discuss the material experience of the system in the early days, and the title of her book, Life on the Screen, suggests that Eliza started off as a screen-based program. Turkle states that the all-uppercase output of the program was an attempt to imitate a Teletype, when, in fact, people initially interacted with Eliza using Teletypes and other print terminals. Tom Van Vleck, who was at MIT with Weizenbaum when he programmed Eliza, wrote: "I think mixed case output was seen as a 'decoration' in those days.. it made data files twice as big and had other mode switching problems with CTSS, and I think it would have actually detracted from the point Joe was trying to make about computer interaction."
Many people have interacted with Eliza/Doctor on screens. But the fact that early output was printed more or less permanently on paper, rather than transiently appearing on the screen, influences how we understand the system as literary and as psychotherepeutic. A session that leaves a printed record, like a diary, may be experienced differently than a transient on-screen encounter. A printout of the interaction may leave people less likely to genuinely confide, or it may have different effects. Although Turkle doesn't mention the print-based nature of early interaction with Eliza, she does offer the diary as a metaphor to help us understand how a "dumb" computer may be able to provide therepeutic benefit. The actual material aspects of the original Eliza/Doctor interface only strengthen Turkle's argument and make the system seem more like a diary. There are other aspects of the early interface that may be important as well. When interacting on a slow-moving Teletype, it was possible to read the beginning of Eliza's statement and still have time to guess what the conclusion of that utterance would be, as is the case in human conversation. The versions of Eliza that are available today, even if they are authentic re-implementations from a formal, computational perspective, are not authentic to that aspect of interface, nor do they leave a paper trail of the session by default.
About a decade after Eliza/Doctor, Will Crowther programmed Adventure in FORTRAN on Bolt, Beranek, and Newman's Cambridge-based DEC PDP-10. The first version was probably completed in 1975. The version of Adventure that became widely known was an expanded and modified version of this original, released by Don Woods in April 1976. He worked on it across the country, at Stanford. The result was a simulated cave with irregularly connected "rooms" — rooms of the sort cavers discover, not the sort that architects design. Some of these formed two mazes of twisty little passages; others were easier to navigate. When compared to earlier computer games, it's easy to see that not only the regularity of space (which was often laid out on a simple grid) but also the regularity of the overall, governing system had been broken in an interesting way. There were numerous different opponents and obstacles to overcome, areas to map, and pieces to fit together. While Adventure wasn't a literary marvel, it did show that the experience of reading and of puzzle-solving could be integrated, and that textual output could do more than display instructions and pure status information. This would later lead to the interactive fiction works that functioned like literary riddles, combining strange systematic workings with a lattice of evocative language.
By 1975, screens were becoming more common, at least at companies like BBN and universities like MIT, where, a few years later, the programmers of Zork would have access to powerful Imlac terminals, screens included. But Crowther apparently programmed Adventure from home, using an ASR-33 Teletype — a rather old apparatus, even at the time. It's not clear whether he wrote the program offline, for the most part, or interactively, but it was written in FORTRAN, all in uppercase, and Crowther probably went through printouts to debug it. Perhaps he even used the practice of checking off lines as he verified them to be correct, as some did at the time; whatever the case, debugging and revising Adventure is likely to have involved not just computational reasoning, but proofreader's marks.
Players also used paper in their adventuring. Tracy Kidder noted that one Adventure player's desk held "roughly drawn maps. They consisted of circles, inside of which were scrawled names such as Dirty Passage, Hall of Mists, Hall of the Mountain King ... Webs of lines connected the circles, and each line was labeled, some with points of the compass, some with the words up and down. Here and there on the maps were notations — 'water here,' 'oil here,' and 'damn that pirate!'"
The experience of interactive fiction as a continuous series of textual exchanges, and the requirement that the interactor look back carefully over the text to puzzle out how to proceed, is certainly consistent with the more permanent and uninterrupted printed output that early terminals produced. There are also details that relate later, home computer interactive fiction to print-based minicomputer experiences. For instance, the commercial games of Infocom — which popularized the term "interactive fiction" and made up some of the best-selling entertainment software of the 1980s — did include a "script" command that would allow players to print out the input and output text as the game progressed, just as if they were using a print terminal to play. While Adventure would describe each location only once, Infocom's interactive fiction included a "verbose" mode in which the "room description" would be output every time a room was revisited. This would have been wasteful when paper was being consumed, but it was practical for home computers, since it was impossible to literally "scroll" back through the printed transcript and it would have been difficult to enable virtual scrolling, which later become common.
These print-based programs were written by people who had good access to computers to begin with; it should come as no surprise that other early experimenters were also unable to use screens. For instance:
Brion Gysin and Ian Sommerville, who most famously collaborated on The Dreammachine in 1959, also did a computer collaboration around 1960. Sommerville programmed a computer to produce every permutation of the phrase "I AM THAT I AM." Gysin read the list on a BBC broadcast, almost certainly without those permuted phrases ever appearing on a screen.
Italo Calvino was invited by IBM to write a story using a computer in 1973. Calvino had the protagonist of "The Burning of the Abominable House," the story which resulted, use punchcards to feed data into the computer. But according to Calvino's wife, the limited computer access in Paris, where they were living at the time, meant that Calvino worked by "carrying out all the operations the computer was supposed to do himself," apparently using paper to do all the computing, not just to interact with the computer.
Clearly, some significant creative work has been done on the computer using paper interfaces. This does not, of course, refute the idea that computer users today, and in the recent past, have a "life on the screen." It should, instead, urge us to better understand the "life on the scroll" that the users of print terminals had — as recently as the early 1980s — and to also understand how punchcard and paper tape interfaces were important to computer creativity. This topic seems particularly relevant as we hurtle past the flat panel into a life of mobile phones and hypertext-enabled MP3 players, while, at the same time, we continue to send more and more plain text emails, sometimes even printing these emails out — just as, originally, emails were printed out, rather than being displayed on a screen.
Since the physical substance that is involved in computer interfaces is important to understanding the cultural and literary dimensions of computing, there seems to be a clear role in digital media studies for an approach analogous to that of textual studies. Similarly, it is difficult to see how Eliza/Doctor can be fully understood without recourse to psychoanalysis, or how Adventure can be understood without studying the textual description of space. My own understanding of Adventure and of interactive fiction has also benefited from an approach that is analogous to narratology, and from the use of the literary riddle to explain how something can both be solvable and literary. So, even when we look at complex computer programs developed by researchers and programmmers, humanistic approaches can be essential.
It also seems clear that in studying such programs, we have to know something about a topic I didn't really delve into: how they actually work — what they do in reponse to input, what rules they follow, functions they evaluate, procedures they carry out. Computer science may not be important for understanding every sort of digital literature — e-texts and link-and-node hypertext novels, for instance, may be usefully understood as something other than computer programs. But when we consider complex computer literary works — the ones that I, at least, find the richest and most interesting — it seems important to not just consider the interface but also reach into the formal workings of these programs, as some scholars and some students in interdisciplinary programs are already beginning to do. Only when we are able to join this sort of understanding of creative computing to an understanding of the material level and the cultural situation will we be able to engage computer literature in the way that we have been able to engage the film of the Lumières' train and the rest of cinema, the way that we have been able to engage the Quixote and the rest of literature. We can then, perhaps, begin to understand digital literary work on its own terms.