1. Overview

As far as we know, this is the first published scholarly edition of a born-digital work—that is, a cultural artifact created for and made to be experienced primarily in digital form, as distinct from a digitized print or manuscript text. The work we represent in this edition—a website implemented in Flash, which simulates a DOS-style computer terminal—was published nearly twenty years ago as a companion to a popular video game, Portal.¹ As scholars of transmedia storytelling have demonstrated, any given paratext—that is, any text in an ancillary relationship to a primary text—may revise our understanding of a video game’s settings, characters, plots, and themes, often by providing key details not present in the games themselves.² Yet paratexts in any medium can also be among the most ephemeral kinds of texts, given their existence on the margins of more well-known cultural works, and we risk losing them to time and neglect without deliberate efforts to preserve them and understand how they fit into larger networks of meaning. This is the challenge that fans of Portal and scholars of video game history alike have faced with ApertureScience.com since the site first appeared on the web nearly twenty years ago—and then disappeared and reappeared in various forms in the intervening years. In this introduction, we will contextualize the ApertureScience.com website as an artifact of internet and video game history, describe its structure and narrative strategies, and detail how and why we undertook this experiment in scholarly editing.³

In 2006, Valve Software launched a website at http://www.aperturescience.com to promote their new video game Portal, which was published on October 7, 2007.⁴ Given Portal’s initial appearance as a stand-alone game supposedly unconnected to any of Valve’s well-known franchises Half-Life, Counter-Strike, or Team Fortress, and marketed as a minor entry in a compilation of more highly anticipated games, Portal was an unlikely success. Yet it went on to become a sleeper hit, winning numerous gaming industry awards soon after its release, and retrospectively included on several lists of the best video games of all time.⁵

As a first-person game focusing on puzzle-solving rather than combat, Portal offered an unexpected mix of two mostly separate genres: the puzzle-platformer and the first-person shooter.⁶ Players begin the game in a windowless testing facility with a computer voice guiding them through a succession of increasingly difficult tests using a “portal gun” to place the entrances and exits of small, localized wormholes upon various surfaces.⁷ As the game unfolds, the owner of the computer voice, a rogue artifical general intelligence named GLaDOS (short for “Genetic Lifeform and Disk Operating System”), emerges as nuanced antagonist in her own right—in what is arguably one of video game history’s great achievements in character-writing (by Erik Wolpaw and Chet Faliszek) and voice-acting (by Ellen McLain). In a comparablegagme design milestone, Portal’s realistic physics engine allowed its developers to exploit a simple but ingenious gameplay mechanic: the velocity of moving objects entering a portal—including players themselves—is preserved when they emerge from the exit portal; for example, a player jumping from a great height into a portal placed on the floor would then find themselves travelling horizontally at great speed if the exit portal were placed on a vertical wall.⁸ However, Portal also told a story—in oblique and surprising ways—that cemented the game’s reputation as one of the greatest video games of all time.

This critical edition uses the Twine platform to offer a playable reproduction of the now-defunct ApertureScience.com website, with added commentary and navigational tools, as a canonical paratext and interactive storytelling platform for one of the early twenty-first century’s most important video games.⁹ ApertureScience.com was updated occasionally since it was first posted in 2006, resulting in four distinct versions that we have been able to identify (detailed below in part 3.a). Our edition is based on the second version of ApertureScience.com, numbered “v1.07” using the site’s internal versioning system (explained below in part 3.a). Our edition mostly presents ApertureScience.com as it would have appeared to visitors between approximately October, 2007 and December, 2009. As our editorial policy explains in detail in part 3.b below, we have reproduced the original site’s text and functionality with only a few exceptions. The changes we have made are mainly to compensate for the original site’s lack of navigational tools, incompatibility with accessibility supports like screen-reading software, and overall user-unfriendliness—the latter being a deliberate part of the site’s aesthetic and narrative design, as we explain below. Our changes to the original site are noted in this introduction and in the annotations on specific screens.

ApertureScience.com was no ordinary website. Using an inventive narrative strategy, the site positioned visitors as though they were using a computer terminal in Aperture Science Laboratories, the fictional setting for Portal. The site used Flash to simulate a DOS-style command-prompt terminal interface, typical of computers prior to the era of windowed graphical user interfaces. Visitors to the first two versions of ApertureScience.com, between late 2006 and late 2009, were greeted only with a blinking green cursor on a black screen, and had to figure out what commands to enter in order to proceed.

Logging in as a guest allowed users to interact with a simulated file system, including a long and complex application form to become a test subject at Aperture Science Laboratories. Observant players of Portal also noticed a login name and password hidden in one of the levels of the game, which, when used at ApertureScience.com, gave users admin-level access, including access to a fictional history of Aperture Science Laboratories. In its time, ApertureScience.com has thus been several things at once: a website, a video game paratext, an artifact of transmedia storytelling, and a work of interactive fiction.¹⁰ With this edition, it also becomes—to the best of our knowledge—the first born-digital work to be reproduced in the form of a scholarly edition.¹¹

ApertureScience.com nonetheless embodies several specific qualities which traditionally have made certain texts good candidates for scholarly editions. First, the text that is difficult to access in its original forms: ApertureScience.com was implemented using Flash, a program that major web browsers ceased to support in early 2021.¹² It is possible to download and run instances of the website’s original Flash files outside of a web browser, but it is not easy (see part 3.a, below). Second, the text exists in multiple versions, with differences that are not obvious but are worth knowing about. ApertureScience.com was updated several times while it was accessible and exists in four distinct versions, with variants between the first two versions that aren’t clearly apparent (see this edition’s annotations on the first two screens of ApertureScience.com, and part 3.a, below). Third, the text can be complex and challenging to understand, but rewards close analysis. As a simulation of the GLaDOS computer system (explained below), ApertureScience.com deliberately employs an aesthetic of user-unfriendliness, making it a difficult text that benefits from commentary to help readers make sense of it.¹³ Finally, the form of the text is innovative and interesting, but not straightforward to reproduce. ApertureScience.com offers a fascinating example of innovative narrative form, as a web-based simulation of an earlier form of computing, and it employs some creative and subtle signifying strategies.¹⁴ The demise of Flash, and the necessity of relying on second-order representations of the original website, such as emulations and screenshots, only adds to the complexity of ApertureScience.com as a text with a layered history of transmission.¹⁵

Like many complex literary and historical texts, any representation of ApertureScience.com thus requires careful attention to the interplay of form, meaning, and interface. Our edition seeks to make all of these facets of ApertureScience.com available to readers and players alike, including players of the Portal games, and scholars of electronic literature and transmedia storytelling.

2. The Structure of ApertureScience.com

A. Mapping ApertureScience.com in Twine

Before we consider ApertureScience.com within its broader contexts, it is worth examining the structure of the site itself and the narrative experiences it enables. How is it possible to tell a story in the form of a simulated computer system? We can understand how the designers of ApertureScience.com approached this question by examining the different sections of the site, which are shown on the map below (Figure 2).

To build this edition, we have recreated ApertureScience.com in the open-source, multilinear storytelling platform Twine, which enables each of the original website’s screens to be plotted as passages in the Twine app. The grouping, arrangement, and names of passages in Figure 2 were chosen by the editors to clarify the internal structure of ApertureScience.com and to make it visible as a whole. Throughout this edition, when we use the term screens, we mean the basic textual units in ApertureScience.com, as experienced by users of the site (such as the screen shown in Figure 1 above). When we use the term passages, we are using a Twine-specific term for the building blocks in a Twine story. Each of the boxes in Figure 2 represents a Twine passage, and each of those passages represents a screen in ApertureScience.com. Conversely, each of the screens that readers will encounter in our edition of ApertureScience.com corresponds to a Twine passage somewhere on this map. (Readers can download our edition and open it in the Twine app to experiment with it first-hand.¹⁶)

Structurally, ApertureScience.com can be described as a set of 86 distinguishable textual units. For example, the opening login screen (Figure 1) is one such unit, the “DIR” screen reachable via the regular command prompt is another, and the “DIR” screen reachable via the admin prompt (showing the secret file “NOTES.EXE”) is yet another. In hypertext fiction, these units are typically called lexia. In this edition, we refer to them as screens when discussing them as units experienced by readers (drawing the term from computing history; see OED screen n.1 V.27.b), and as passages when referring to them in the context of Twine, drawing the term from Twine’s technical documentation.¹⁷

There are five identifiable groups of screens in ApertureScience.com:

1. the login sequence: 5 mostly linear screens, marked in purple in the upper left of the Twine map;

2. the command prompt screens: 20 non-linear screens, marked in yellow and arranged in two circles on the map (one circle of 10 screens for admin-level logins, the other circle for regular logins, with slight differences between the two sets);

3. the Enrichment Center test subject application: 54 linear screens, marked in blue on the right;

4. the secret message: 2 screens, marked in amber and placed between the two sets of command-prompt screens;

5. the notes sequence: 4 linear screens, marked in red on the lower-right.

B. The login sequence

These are the first screens you encounter, beginning with a simple, blinking cursor with no other context or explanation other than the domain name ApertureScience.com itself. The first two versions of the site (v1.07 and v1.07a) displayed only the command prompt on the first screen, but a third version (v1.09) added a looped, Christmas-themed video (Figure 3). The video appears to be from a security camera feed from somewhere inside Aperture Science at Christmas time. The Aperture Laboratories logo is positioned in the upper-right, and visible in the background are objects from Portal (a turret, cake, Weighted Companion Cube, and safety sign) and Half-Life 2 (crates and a garden gnome). A Christmas version of the end-credits song from Portal, “Still Alive” by Jonathan Coulson, plays in the background as the camera pans back and forth. Randomly, the gnome’s face appears ominously close to the camera. As the camera pans back to the right, the words “HAPPY <HOLIDAY NAME HERE>” appear as a typed overlay, continuing a running joke in which supposedly sincere messages from Aperture Science are revealed to be impersonal computer-generated forms.

The login sequence also includes a help screen with a curiously specific, even alarming, set of instructions, as if they were written in response to a plea for help with an injury, rather than a request for assistance with technical difficulties with the computer system.

The most unusual feature of the login sequence is the option to log in as an ordinary user (with any username and the password PORTAL or PORTALS) or as Aperture Science’s CEO and founder, Cave Johnson (with the username CJOHNSON and the password TIER3). Logging in as Johnson provides administrator-level access to the simulated computer system, including some significant extra material (detailed below). The CJOHNSON/TIER3 login information was originally found in the game Portal itself, scrawled on a wall in a normally inaccessible backstage area of one of the test chambers (Figure 4). The implication is that Johnson’s login credentials were stolen and shared by Doug Rattmann, a scientist who survived GLaDOS’s lethal purge of Aperture Science’s employees and has been eking out a fugitive existence in the forgotten corners of the facility.¹⁸ This direct connection between the game and the website is an example of what video game players call an Easter egg, which are objects and other in-game references that signal links between different parts of a transmedia story, or point to other recognizable works.¹⁹

C. The command prompt screens

Once logged in, you will encounter one of two types of command prompt, depending on whether you log in as a regular user or as an administrator. Note the slight differences between the two prompts in Figures 5 and 6 below, including the operating system versions, and their similarities with original forms of the MS-DOS prompt shown in Figures 7 and 8.²⁰

The form of the command prompt is a clear reference to MS-DOS, the operating system which became the basis for Microsoft Windows. In the regular prompt shown in Figure 5, the “B:\>” prompt indicates that the user is accessing a disk drive labelled “B.” Comparison with the original MS-DOS prompt also helps to highlight that the name of Portal’s primary antagonist, GLaDOS (short for “Genetic Lifeform and Disk Operating System,” spelled out on the final page of the notes sequence), is a reference to MS-DOS. The implication is that users of the computer system are interacting with GLaDOS herself via the simulated computer system at ApertureScience.com. The pattern of psychological manipulation and disregard for research ethics that GLaDOS displays throughout Portal is reflected in subtle ways throughout ApertureScience.com, especially in the application sequence described in the next section below.

Like most command-line interfaces, the GLaDOS prompt includes a set of standard commands—though not all of the commands available on ApertureScience.com are MS-DOS commands. For example, LIB is not a standard MS-DOS command, but entering LIB or HELP in GLaDOS produces a list of available commands. The resulting screen (Figure 9) is similar to that of the HELP command in MS-DOS, but without the inline explanations of what the commands do.

Entering most of these commands will result in an error screen informing you that the disk is write-protected or you do not have authorization. However, some of the commands produce unusual results that can help players understand how GLaDOS and Aperture Science are characterized in Portal. For example, the incongruous command INTERROGATE (which was never part of MS-DOS) can be followed by a space and a second string, presumably someone’s name. Regular users will receive an error message that says “ERROR 01 [Illegal attempt to initiate disciplinary action],” but users logged in as Cave Johnson will receive “ERROR 07 [Unknown Employee].” The sinister joke is that Johnson or other Aperture Science administrators would apparently initiate interrogations of employees, and could do so with a casually entered command in the company’s computer system.

The command PLAY PORTAL also produces an unexpected result, triggering a sound file in which a female voice—presumably voice actor Ellen McLain in character as GLaDOS—simply says “goodbye,” followed by an automatic redirection of the user’s web browser to a YouTube video which has since been removed.²¹ The reference to playing the game Portal and the unusual response could be an example of the designers breaking the fourth wall—a theatrical term for moments where the performance breaks its fictional frame and acknowledges the presence of the audience and real world. Similarly, ApertureScience.com also breaks the fourth wall by redirecting the web browser to the user’s default starting page or to the Valve store at SteamPowered.com when the user enters LOGOUT, LOGOFF, BYE, or VALVE.

The commands DIR, DIRECTORY, CAT, CATALOG, LIST, and LS all bring up a screen that provides information about the disk, including executable (.exe) files which you can open. For regular users, this screen only displays one file, “APPLY.EXE” (Figure 10). However, when a user logs in with Johnson’s admin account, this screen displays the additional file “NOTES.EXE” (Figure 11). These two simulated files contain some of the most interesting material on ApertureScience.com: the Enrichment Center test subject application and the secret message, as detailed below and in the edition’s annotations for these screens.

D. The Enrichment Center test subject application

Entering the command APPLY initiates a series of screens that lead you through an application to become a test subject in the Aperture Science Enrichment Centre. The application screens form the longest sequence in the site, and could be considered the centerpiece of ApertureScience.com as a world-building paratext for Portal. They provide an indirect backstory for the game’s opening moments, in which the player-character awakens from stasis as a test subject in the Enrichment Centre. On the original site, once users began the application sequence, they were forced to proceed through each of the 54 screens and could not go back or quit using commands available in ApertureScience.com. Our edition, however, provides additional navigational tools, including a back button and the ability to jump to any other part of the site, enabling you to revisit specific screens or sequences without having to traverse the entire application sequence.

The application sequence can be understood as a parody of the kinds of personality tests whose use became widespread in military and corporate workplaces over the twentieth century, such as the the Humm-Wadsworth Temperament Scale (1934) and the Myers-Briggs Type Indicator test (1944). These and other personality tests have been criticized for their lack of objectivity, validity, and reliability, for their alleged origins in pseudoscience, and for their susceptibility to misuse by employers.²²

As psychometric managerial tools, these tests share the premise that individuals’ personalities can be sorted into scientifically definable types using methods such as self-report questionnaires, like the one in the ApertureScience.com application sequence. The ApertureScience.com Enrichment Center test subject application describes itself on the first screen as a test of personality and general knowledge, apparently the first part of a larger set of tests (though only part 1 is included on the ApertureScience.com site). The questions that follow tend to imitate those found in standard personality tests, but with frequent absurd or satirical touches. Some questions seem impossible to answer. Other questions come with an improbably exhaustive set of options or seem to reflect a flawed understanding of human experience, including what counts as general knowledge (such as a question about ancient mathematicians’ home cities), and hint at threatening consequences for the person taking the test. These characteristics, along with several questions fixated on cake—a running joke in Portal, explained below—may indicate that GLaDOS herself designed the test, or that GLaDOS and the test alike are both symptoms of the same pathologies present in Aperture Science’s corporate culture since its founding (see below). Several questions seem designed to erode the applicant’s self-esteem by implying that they are socially isolated, untrustworthy, and paranoid. Not coincidentally, all of these attributes characterize GLaDOS herself, at least as she is depicted in the first Portal game.²³

This edition includes annotations for most of the individual screens in the application sequence, but certain screens are worth a more detailed discussion. The most potentially confusing screen in the sequence is the second one. This screen gives users the impossible task of memorizing their own “Unique Identity Number (Plus Letters),” or UIN(+L) as it is called in abbreviated form on several screens. This unnecessarily complex name, especially the parenthetical “Plus Letters,” is typical of Aperture Science’s technocratic language. In the same vein, the UIN(+L) is composed of an absurdly long 64-character string, composed of random letters and numbers, all of which blink rapidly at the same time—and also change randomly, one character at a time, between blinks (Figure 12).²⁴ On the original site, the UIN(+L) was apparently randomly generated by a server-side PHP script named "gdtx.php." According to a 2006 post by AndrewNeo on the Valve Developer Community wiki, the PHP script's interactions with the site's Flash file included sending answers for almost every question back to the server, accompanied by the visitor's UIN(+L)—though we have been unable to verify this claim.²⁵

The first UIN(+L) screen sets up one of the application sequence’s running jokes. Much later, page 33 of the test asks you to name a crime that only you know you have committed, as a security question to allow you to recover your UIN(+L). The penultimate screen in the application sequence asks you to enter your UIN(+L) to complete the test. It is impossible to enter the original sequence of characters correctly, first because the Flash file seems designed not to accept and verify the user’s input on this screen, and second because there never was a stable UIN(+L) to memorize, due to the random changes between blinks when it was first displayed at the beginning of the test.²⁶ The final screen in the application sequence, along with the absence of any alternate endings to the application sequence in the recovered Flash code, indicates that the impossible UIN(+L) is not an error in the site’s design, but a narrative strategy to bring the application sequence to an unsatisfying conclusion—one that underscores the irrationality of this supposedly rational testing process. This edition reproduces the behavior of the UIN(+L) as it appeared on the original versions of ApertureScience.com, including the random mutations when the UIN(+L) is first displayed in the application sequence, and the impossibility of entering it correctly at the end.²⁷

Another subtle joke takes the form of slowly pulsing letters on thirteen of the screens at various points in the application sequence (all of which are flagged in the annotations). Each of these screens has a single pulsing letter; for example, see the t in the word “to” on page 1 of the application. The pulsing is subtle enough that it could easily be missed or mistaken for a visual glitch. However, writing down all of the pulsing letters in sequence spells out the phrase “thecakeisalie”—one of Portal’s most recognizable catchphrases, and a reference to GLaDOS’s dubious promise of cake as a reward for subjects who complete the Enrichment Center tests.²⁸ Entering THECAKEISALIE at the GLaDOS command prompt leads to a secret message, described below.

Related to the pulsing letters, another easy-to-miss phenomenon in the application sequence is the occasional flashing of an image of a slice of cake on the screen. The cake image is composed of alphanumeric characters with upper- and lower-case English Latin letters arranged to create a recognizable shape, in a form commonly known as ASCII art (Figure 13). The cake will appear only within the application sequence, and only on certain numbered application question pages. In the original versions of the site, cake flashes are very infrequent and very brief (approximately 400-500 milliseconds), such that it is possible to progress through the original site’s application sequence without noticing one. We have attempted to reconstruct the behavior of the cake flashes in the original site, based on details gleaned from the decompiled Flash code along with painstaking observation, but this edition also changes the functionality of the original site by giving you control over the behavior of cake flashes, including the ability to turn them off entirely. This added level of control improves accessibility, as the cake flash may not function well with some screen readers, and also helps readers who wish to study the cake more closely.

There are many other in-jokes and allusions placed throughout the application sequence, including impossible-to-answer questions about your favorite color, an absurdly long list of animals you might wish to domesticate, and an oddly placed data-use consent screen. For more details about the application sequence, see the annotations on individual passages.

E. The secret message

As noted above, several of the application screens contain individual pulsing letters which, put together, spell out “thecakeisalie.” Entering this phrase (without spaces) at the GLaDOS command prompt brings up what appears to be a message hidden by another Aperture Science employee on the computer system; presumably the same person caused the letters to pulse in the application as a clue to the hidden message. The message contains a warning to any other surviving Aperture Science employees, and speculates about the purpose of a “relaxation Vault” in which the player-character awakens at the beginning of Portal (Figure 14). Fans have speculated that this secret message, like the Cave Johnson login credentials, may have been left by Doug Rattmann (mentioned above).

The message concludes with the lines “I don’t think going home is part of our job description anymore. If a supervisor walks by, press return!” Pressing any key causes the screen to switch to a spreadsheet to conceal the message, and pressing any key again will switch it back.²⁹

F. The notes sequence

The notes sequence is only four screens long, but it provides background on the origins and history of Aperture Science that cannot be found anywhere else in the Portal transmedia franchise. As noted above, these screens are part of the (simulated) file “NOTES.EXE,” which can be accessed only by users who have logged in with an administrator account. Doing so requires the username and password of Cave Johnson, Aperture Science’s founder and CEO, and this login information can only be found in a hidden area in one of Portal’s levels (see Figure 4 above). The fact that users can only access these screens through Cave Johnson’s login information makes the notes sequence one of the most obscure Easter eggs in the Portal franchise, but it is also one of the most revealing sources of information about the enigmatic Cave Johnson and the backstory to Portal.

As Figure 15 reveals, the notes sequence takes the form of a timeline of Aperture Science’s development, beginning in 1953 during the postwar corporate research boom in the United States, extending through the Cold War years to 1996. The notes include details on Johnson’s convoluted revenge plot against the House Naval Appropriations Committee, his inadvertent mercury poisoning and its physical and psychological effects, his increasing eccentricity and obsession with dubious research projects (the “Heimlich Counter-Maneuver” and the “Take-A-Wish Foundation”), and the development of portal technology and GLaDOS. A final, undated entry from “Several Years Later [after 1996]” indicates that GLaDOS’s (untested) AI capabilities were activated on Aperture Science’s first bring-your-daughter-to-work day. This last detail hints that the Portal games’ player-character, Chell, may be the daughter of an Aperture Science employee, and was placed in stasis by GLaDOS when she activated and took control of the facility.³⁰

The information hidden in the notes sequence is important context for understanding the Portal franchise as a transmedia work. In addition to hinting at a backstory for the otherwise minimally drawn player-character, the notes also definitively place the events of Portal within the same universe as the Half-Life games, via the reference to rival defense contractor Black Mesa on the third notes screen.³¹ The notes also do more than any other information source in the Portal franchise to place Aperture Science’s origins in the historical context of what American President Dwight Eisenhower famously called the “military-industrial complex” in his 1961 farewell address. This term describes the network of relationships that formed between American military agencies and private industry in the years after World War II and through the Cold War. Aperture Science's initial specialization in shower curtains makes it an eccentric outlier in this emerging defense industry. By the end of the notes sequence, however, the company’s research projects become more outlandish and dangerous in their pursuit of government contracts, culminating in the activation of an untested artificial general intelligence on bring-your-daughter-to-work day.

Many of these themes are elaborated in the sequel Portal 2, particularly when the player explores a set of disused test chambers that date to some of the periods described in the notes sequence (accompanied by audio instructions from Johnson, steadily deteriorating from his mercury poisoning, who made the recordings when those parts of the facilities were built). Overall, this short timeline helps to place Portal among a thematically connected set of works of popular culture, including Fail-Safe (the 1962 novel and 1964 film), Dr. Strangelove (1964), WarGames (1983), and The Terminator (1984) and its sequels, all of which deal with the dangers of military-corporate research and the ceding of control functions to automated or artificially intelligent systems.

3. Notes on the Text

The primary purpose of this edition is to make ApertureScience.com available to readers in a form that is likely to keep working into the future, and is accompanied by information that will help readers understand and contextualize ApertureScience.com as a paratext from video game history. Portal’s influence has remained strong within the video game canon. It has also been the subject of a substantial amount of academic writing in game studies and adjacent fields. However, ApertureScience.com itself has apparently gone entirely unmentioned in academic publications, despite its recognition as an important official paratext among fan communities.³² This edition therefore serves as a recovery project, making a rare and ephemeral digital document available for future study and enjoyment.

A secondary purpose of this edition is to explore the possibilities for creating scholarly editions of born-digital works. Surprisingly, very little scholarly work has been done on this front, even by 2024.³³ The centuries-old tradition of scholarly editing has naturally tended to focus on print and manuscript works, and digital scholarly editing in the digital humanities has tended to focus on applying digital tools to non-digital texts. As far as we can determine, this is the first scholarly edition of a text that is born-digital. Born-digital texts are created to be experienced only in digital form and have features, such as animated text and interactive questions, that can only be represented accurately in a digital environment. Our edition uses the affordances of web browsers, Twine, and other digital technologies to recreate the text as accurately as possible while also adding commentary and instructions to benefit readers. It is our hope that this editorial experiment may inspire others, and serve as a model and source of reusable code.

This edition is therefore both a resource and an experiment. In this section we will focus on explaining the decisions we made in the creation of this site as a resource, leaving its experimental aspects as a topic for follow-up publications.

A. ApertureScience.com as an artifact of video game history

The primary source documents for our edition are Flash SWF (ShockWave Flash) files. For each version of the site, a single SWF file contains the text and functionality that the ApetureScience.com site used to simulate an Aperture Science Laboratories computer terminal. Although ApertureScience.com was a website, visitors were primarily interacting with Flash files via a web browser. Through collation of decompiled code in SWF files, we have identified four distinct versions of the ApertureScience.com website over time. Those distinct versions correspond to the versions listed on the Valve Archive and documented on the Combine OverWiki.³⁴ Like these websites, we follow the version numbering system found in the GLaDOS command prompts in each iteration of the site (for examples, see Figures 5, 6, 8, and 9 above):

1. v1.07: the version published prior to Portal’s release, posted at http://www.aperturescience.com from approximately October 5, 2006 to at least September 30, 2007;

2. v1.07a: an updated version with additional material (for example, the admin-level command prompt and “NOTES.EXE” file), posted on the website as early as October 11, 2007 until at least February 16, 2009;

3. v1.09: identical to v1.07a but with the addition of a Christmas-themed security camera video that auto-played on a loop on the initial screen, posted as early as March 8, 2010 (the date of this version’s earliest Internet Archive snapshot), but probably earlier, in December 2009, and until at least December 12, 2010;

4. vH2010 (Holiday 2010 version): this version removes the command prompt entirely and replaces the v1.09 video with a slightly different one, posted as early as February 10, 2011 (the date of this version’s earliest Internet Archive snapshot), but probably earlier, in December, 2010; this presently still available on https://www.aperturescience.com/, though it will be non-functional in most browsers.³⁵ The Half-Life wiki and other fan resources indicate that the video was altered on April 1, 2010, to include a potato beneath the Christmas tree, as a tie-in to an ARG (Alternate Reality Game) called PotatoFoolsDay to promote Portal 2.³⁶ Unfortunately the Internet Archive does not record a snapshot of this sub-variant of vH2010.

Each version replaced the prior version on the website. Except for the final version—which is the least interactive—we have used the Internet Archive’s Wayback Machine to download Flash files for the different versions.

Although most of the ApertureScience.com versions’ functionality is text-based, there are three short looped video files: the two holiday videos mentioned above, and one short audio clip in which GLaDOS’s voice actor, Ellen McLain, says “goodbye” when the user logs out. We have recovered the video and audio files from the Flash files and use copies of these original files in our edition. They are in .mp4 and .mp3 formats, respectively, and thus straightforward to integrate into Twine for posting on the web.

As with other well-loved video games, Portal’s fans have been creating resources to document ApertureScience.com for some time. Detailed descriptions of ApertureScience.com can be found on fan-created wikis that bring together remarkable amounts of research by fan communities and have provided an important resource in the creation of this edition.³⁷ In 2020, the Valve Archive, a fan-created repository of hard-to-find materials from games published by Valve Software, published emulated versions of ApertureScience.com after the only other accessible copies of the site, preserved by the Internet Archive’s Wayback Machine, became inoperable due to the discontinuation of Flash support by web browsers.³⁸ The Valve Archive’s work on this project is especially valuable for video game history because they have documented and emulated all four versions of the ApertureScience.com site. This approach has made the Valve Archive’s emulations the most easily accessible version of ApertureScience.com since the site first appeared on its original domain in 2006. Arguably, the Valve Archive emulations may be more useful than the original site for the purposes of video game history because the Valve Archive offers all four versions together as a set and notes the differences between them.

B. Editorial policies

Our project complements the Valve Archive but does not duplicate it. To use terms from the world of scholarly editing, the Valve Archive emulations could be considered facsimiles, which reproduce primary source documents as closely as possible, usually without conflating versions or correcting apparent errors. For example, a facsimile of an historical printed book might take the form of a set of photographs of each page from a given copy, usually with minimal commentary by the facsimile’s editors. By contrast, we have created a critical edition of ApertureScience.com, which, like a facsimile, reproduces a primary text, but which also provides layers of editorial commentary, notes points of variation between the versions, and gives readers navigational tools to move among ApertureScience.com’s mostly linear screens. As we can attest, having to move sequentially through the “APPLY.EXE” sequence to get to a specific screen is not a convenient way to navigate.

Our edition reproduces the second version of ApertureScience.com, numbered v1.07a on the Valve Archive. We have chosen this version as our base text because it includes screens not present in v1.07 but unlike v1.09 it does not include the Christmas video on its landing page—ApertureScience.com can be confusing enough without it. Although v1.09 also provides a command prompt and the same sequences of screens as v1.07a, it does not include any new material aside from the login screen video, and is otherwise identical to v1.07a. We have indicated any significant points of variation between our edition’s text of ApertureScience.com and the other versions in the commentary notes.

Generally, our policy has been to make minimal interventions in the text, but we do intervene in cases where the original site’s functionality is an obstacle to access, such as providing a back button and a navigable table of contents. We have mostly refrained from correcting apparent errors or altering the text of ApertureScience.com, with some exceptions noted below and in our annotations.

For example, on page 49 of the application sequence, which asks the user to pick their favorite kind of cake, the name of the cake in option 48 is spelled “St. Honore Cake” without an acute accent over the e in “Honore”. Similarly, the name of the Croatian cake in option 37 is normally spelled “Orehnjača” with a circumflex accent over the c, but the original versions of the site display “Orehenjac(a” in what appears to be an encoding error. We are unable to determine whether the encoding error happened when the designers of ApertureScience.com attempted to display a circumflex-c in Flash, or whether the error was intentional—perhaps as a reference to the Anglo-centrism of older text encoding systems and software, which sometimes did not handle accented letters well. Another example of editorial intervention is the number of options mentioned as possible answers to the question on page 21 of the application, which asks the user to select the wild animal they would most like to domesticate. The prompt indicates that there are 2,314 choices, but the final choice is numbered 2,313 (“Zorilla”), and there are no errors in the numbering that would explain the discrepancy.

In all of these cases, and in more ordinary cases of inconsistent capitalization or spacing, we have let the apparent errors stand as they appear in the original versions. Our rationale is that minor errors like these are not unusual in the forms generated by bureaucratic organizations, and such errors may offer hints about labor conditions and workplace cultures. Judging by the Enrichment Center application form, Aperture Science seems to prioritize copy editing about as much as research ethics and employee safety.

On the other hand, we have changed the functionality on page 21 of the application form to correct what seems to be a bug in the original Flash site’s design, again with the intention to improve the accessibility of our edition. In the original Flash versions of the site, pressing PgUp or PgDn would skip large spans of options. When the user first reached this screen, the last option shown in the bottom right was 0057, but pressing PgDn to advance the list meant that the next option shown in the upper left was 0105, not 0058 as expected. Each advancement of the list would skip exactly 59 options. For this edition, we have recovered all of the animal names from the original site’s decompiled Flash code. The “animals” screen provides a setting that allows you to choose between displaying all of the recovered options or preserving the option-skipping behavior of the original site.

Our explanatory notes—which include lists of possible commands—appear alongside each screen, though readers will have the option to hide the notes if they wish to experience ApertureScience.com in something close to its original form. Our annotation strategy is to note anything unique, problematic, or interesting in ApertureScience.com, with a particular focus on the site’s transmedia connections to the Portal franchise, and the historical background of the site’s (and games’) parody of Cold-War-era corporate research cultures. Our annotations are based on our original research and the remarkable amount of documentation collected on fan wikis and message boards over the years since ApertureScience.com and Portal were released. Our intention with this edition is to build upon and supplement these community-built resources, not to replace them.³⁹ However, by reproducing ApertureScience.com in the form of a critical edition, our goal is also to take advantage of a very old form of scholarly publication which has proven useful for representing literary, historical, musical, scientific, and other kinds of texts—and which we believe holds promise for representing historical video games and interactive digital works like ApertureScience.com.

C. Twine and accessibility

Our edition makes ApertureScience.com available to readers using Twine, a browser-based, open-source, community-supported platform normally used to create hypertext fiction and games. The Twine navigation user interface has been re-coded in order to fit the design and house style requirements of Scholarly Editing. Using our own customization of the Twine platform, our edition reproduces the look and feel of the original Flash website, including details such as the teletype printer animation effects.⁴⁰ The Twine interface also enables us to provide readers with the traditional editorial apparatus, namely an introduction, commentary such as explanations of commands users can enter at specific points that accompanies the text, indications of textual variants, and navigational aids. As a platform designed for writing and publishing multilinear stories and games, the Twine app also provides readers with an interactive map of the screens that make up the ApertureScience.com website, making visible the site’s Easter eggs and hidden areas (see Figure 2 and the discussion of Twine above).

The names used in this edition to identify specific screens—such as “preLoginCursor,” “adminPrompt,” and “memorize”—were created by the editors while adapting ApertureScience.com for Twine, and do not derive from the original site.

Since we are, as far as we know, the first to produce a scholarly edition of a born-digital work, it is important to set the precedent that such editions should be accessible to as many users as possible, including those who use screen-readers and other assistive technology. [Note to reviewers: the editors will complete the subsection on accessibility features after they have been implemented in the edition’s interface. The plan is to complete the accessibility work simultaneously with peer review, and then to conduct an accessibility audit to ensure the features work as intended. Our plans include ensuring navigation via keyboard (which is largely already implemented) and screen-reader compatibility (yet to be tested).]

D. Text encoding

Although Twine serves as the primary interface layer for the edition—both for reading and for our own editorial work—we have also encoded the entire edition in XML (eXtensible Markup Language), following the P5 Guidelines of the Text Encoding Initiative (TEI). Using a single XSLT (eXtensible Stylesheet Language Transformations) template, supplemented by separate files for Javascript/CSS libraries, we transform the edition’s TEI-XML into a single Twine HTML file which may be opened in a web browser, and which links to copies of the media files used in the original ApertureScience.com site (specifically, the looped videos and the “goodbye” audio file, mentioned above). Following the normal practice for micro-editions published in Scholarly Editing, we have made these files and assets available for downloaded; to do so, click on the TEI button on the bottom right of the edition to access the code repository. Readers are welcome to access our edition directly through the website of the journal Scholarly Editing, or download the Twine HTML and media files to work with them in the Twine desktop app (see Figure 2), or download the XML, XSLT, and library files files and carry out the Twine transformation themselves.

Surprisingly, given the decades-long history of hypertext fiction in digital form, as of 2024 there are presently no parts of the TEI P5 Guidelines that deal specifically with the encoding of born-digital texts like ApertureScience.com.⁴¹ Even more surprisingly, perhaps, we were nonetheless able to adapt the TEI Guidelines to represent ApertureScience.com without creating new tags to extend the TEI P5 schema.⁴² Although more work on representing born-digital texts of various kinds in TEI would be welcome, one conclusion we have drawn from our work on this project is that even the current TEI P5 schema—in its most unrestricted form—may be sufficient for encoding certain kinds of born-digital texts. In this section, we will explain some of the choices we made in adapting the TEI Guidelines to a born-digital text.

As noted above in part 2.a, ApertureScience.com may be understood as a set of 86 textual units, which we refer to as screens. Each screen in ApertureScience.com maps more or less straightforwardly to a Twine passage, to use the name of Twine’s basic narrative building blocks. To represent these screens/passages in TEI, we have used the generic <div> element, adding unique @xml:id attributes with our own shorthand names for each unit. Transcription of each screen’s text was relatively straightforward using standard TEI tags. However, an interesting feature of ApertureScience.com as a born-digital text is its use of animation, especially the teletypewriter (TTY) effect used to display some screens one letter or word at a time, evoking early forms of computing and fictional representations such as The Matrix.⁴³ We have used @rend attributes on standard TEI tags to categorize different kinds of animation effects, including the pulsing letters that spell out “thecakeisalie” in the application sequence described above. These attributes, which accompany the elements through the XSLT transformation into a Twine HTML file, then tell the customized JavaScript how to animate the text. Although animation is a typographical feature that is almost certainly unique to digital texts, it was relatively simple to describe it in TEI-XML; the challenge, rather, was in its implementation in the interface layer and adaptation for screen-reading software.

The encoding of screens/passages also raises the question of their relationships with each other as narrative building blocks in a text. In some places ApertureScience.com’s structure is strictly linear, such as the application and notes sequences, and in other places the structure is non-linear, such as the various GLaDOS prompt commands and resulting screens. In the TEI-XML, @corresp attributes indicate which screens are directly accessible from any given screen, and @ana attributes flag screens as members of the larger narrative units described above, such as the login or application sequences. The terms used within @ana attributes are encoded and described as a small-scale taxonomy included in the TEI-XML. The multi-linearity of ApertureScience.com’s screens also raises the question of the ordering of the corresponding <div> elements in the TEI-XML file itself. Given that XML files are both linear and human-readable (for example, in an XML editor) and simultaneously non-linear and machine-readable (for example, via XPath queries), we have arranged the <div> elements for linear passages in sequence in the XML, but treat the ordering of non-linear screens or <div>s as entirely arbitrary. To understand which is which, please consult the Twine map, whose arrows indicate possible sequences of reading.⁴⁴ In the XML we have used TEI’s @org=“composite” attribute to indicate <div> elements corresponding to screens that are in non-sequential relationships with each other.

The most unusual encoding decision we have made is probably the decision to undertake the project at all. By reproducing ApertureScience.com in the form of a scholarly edition, using heavily customized Twine as the main interface layer and minimally customized TEI P5 XML—without extension—as the main encoding layer, we have demonstrated that the TEI Guidelines are presently capable of representing a moderately complex born-digital text such as ApertureScience.com. Interestingly, this has proved to be the case even though the TEI P5 Guidelines (at the time of writing, in late 2024) offer relatively few features to represent born-digital forms specifically. Should this outcome be surprising? With digital texts especially, there is no such thing as raw data; any act of encoding is always an act of translation from (at least) one set of encoding conventions into another—even when printed or handwritten documents are encoded with XML.⁴⁵ Considered in this light, it is not such a stretch to attempt to encode a born-digital text like ApertureScience.com using TEI-XML. As this and other Scholarly Editing micro-editions demonstrate, TEI encoding is particularly suited to representing texts that are marginal, understudied, and interesting—to use three adjectives drawn from the journal’s statement of purpose—because the TEI Guidelines strive to be adaptable to unusual materials. Unlike GLaDOS, the TEI Guidelines do not embody a positivistic worldview, in which every dimension of humanity is potentially measurable, classifiable, and controllable.

Although TEI has provided a mostly stable basis for digital scholarly editions for decades, what makes our project unusual is the choice to represent an ephemeral born-digital text as a scholarly edition. In this respect, we have intentionally departed from the normal practices of digital preservation, which have tended to focus on migration and emulation as primary strategies.⁴⁶ In 1995, Jeff Rothenberg called public attention to the need to preserve born-digital materials, and to the difficulty of doing so, arguing that “the significance of many digital documents—those we consider too unimportant to archive—may become apparent only long after they become unreadable.”⁴⁷ In this light, whether or not digital objects are truly ephemeral depends, at least in part, on the choices we make about them, through action or inaction. Rothenberg’s use of the word “document” suggests he was not thinking of video games or interactive fiction, much less video game paratexts like ApertureScience.com. But in the decades since Rothenberg’s call to preserve objects of born-digital heritage, the efforts of video game fan communities have resulted in vital preservation resources like the Valve Archive, The Cutting Room Floor, and Unseen64.net, and have expanded what counts as important and worth archiving. In the same spirit, our project seeks to add the centuries-old form of the scholarly edition to the toolbox of strategies for preserving born-digital works from the past. Although scholarly editions are resource-intensive to create, and not suitable for mass preservation efforts, they are a viable form in cases where researchers and communities care enough to invest the work—and have the resources to do so.

Such work is an exercise in determination. By design, the players of Portal embody Chell’s character-defining tenacity while coming to understand—and ultimately best—each in-game puzzle chamber. We hope that readers of this scholarly edition find themselves invested with the same qualities of curiosity and persistence as they, too, navigate the passages and command lines within. The rewards at the end are real.