Exocog: A case study of a new genre in storytelling

Jim Miller
Miramontes Computing
November, 2004
jmiller@miramontes.com

The evolution of storytelling: Now underway

One of the oldest known human activities is that of telling stories. It's an important part of how we educate ourselves, pass down culture across generations, and entertain each other. Throughout the ages, storytellers have adapted their art to take advantage of changes in technology -- moving from cave walls to stone tablets to papyrus to sheaves of paper to the printing press. It should be no surprise, then, that some storytellers are looking at computers and the Internet with interest. This is not simply a matter of how they might use the Web as a publishing or distribution source for their stories, but how the special characteristics of the Internet can affect and change the nature of creating, telling, and experiencing stories.

What follows is a detailed look at one experiment in this evolution of storytelling. My partners and I wrote and released a story on the Internet not as narrative text, but as a set of Web sites whose content evolved over five weeks. The story was conveyed by the changes that occurred in these Web sites and in the events that those changes implied in the minds of the readers. As a result, the experience of watching this story unfold was perhaps more like playing a game than reading a book. To me, what is significant in this experiment is the balance between the opportunities that arise from the new technologies and the things that stay the same. This evolutionary process is still underway, but enough has happened to make this a reasonable time to look around and think about where this combination of technology and the creative arts is today, and where it might go tomorrow.

Early experiments with storytelling using computers and the Internet

The first experiments with computers and storytelling took two forms. In the mid-1980s, some adventurous authors created stories as hypertext or hypermedia structures. They broke a story into a collection of passages, and used a hypertext system to connect the passages into a web that readers could explore however they chose [1]. These stories were first built with experimental software or commercial packages such as HyperCard, but the future transition of these experiments to the Web was inevitable. From a literary perspective, these authors were challenging the traditional notion of narrative: the unfolding of the story's events was less a matter of the author's design than of the reader's own interests. At the same time, computer-based stories emerged that were similar to the childhood "Choose Your Own Adventure" books. Here, the author still defined the flow and narrative structure of the story, but allowed the reader to make decisions on behalf of the protagonist -- "Should Sir Gawain slay the dragon, or try to reason with it?" Story fragments corresponding to both alternatives were provided by the author, thus producing a tree (an acylic graph, actually) of stories, with each path through the tree amounting to a separate narrative.

These were worthwhile experiments, but ultimately they were not successful. Hypertext hasn't worked as a literary device beyond an experimental niche: It would seem that there really is value to an author building and unrolling the narrative structure of a story. Similarly, the Choose Your Own Adventure idea failed to become much more than an interesting novelty for children: It's hard enough to write one good story around a plot and set of characters, let alone many. So the search for an interesting convergence of storytelling and computers moved on.

A new approach to storytelling, from an unlikely source: Product marketing

From its beginning, product marketing groups saw great value in the Web as a marketing device, none more so than those who were promoting movies and other entertainment products. The first uses of the Web for entertainment marketing tended to take the form of what might be called "press-kit sites:" Web-based versions of the kinds of information traditionally given out to the press to publicize a movie, such as a synopsis of the movie, biographies, photographs, and interviews with the cast and crew. Given the Web's fluency with media, sites built around this content grew to include downloads of movie trailers, a broader collection of photographs and video clips, and computer-based giveaways like movie-themed desktop wallpaper and screensavers. Some sites also began to offer simple Flash-based games themed around the movie: a spy movie site might have a target practice game and offer a special video clip or picture as the prize for scoring above a certain level.

The use of the Web as a promotional device took a significant step forward with the faux Web site built for the 1995 film The Blair Witch Project. The marketing team working with the filmmakers constructed a site that used a variety of Web-based media to tell the purported history of the witch depicted in the movie. In keeping with the pseudo-documentary style of the movie, the site was presented in a completely factual manner -- there was nothing on the site to suggest that the witch and her history were anything but real. The site caught the interest of its audience, logging over 75 million hits throughout its release, and is widely recognized as the basis for the movie's success in the marketplace [2]. The Blair Witch site was hardly the first movie-oriented website, nor the first faux site to appear on the Web, but it was a landmark in showing how the Web could be used to tell stories in a way that was new, different, and appealing to large numbers of people.

Blair Witch was a relatively static site: It was presented as a Web-based publication of an investigation of the witch, and did not change much after it was posted. But, of course, many Web sites in the real world change on a regular basis. Could a story be told by gradually evolving one or more Web sites over time, with the story growing out of the changes in the sites and, perhaps more importantly, the visitors' interpretations of these changes? This challenge was taken up by a team from Microsoft Game Studios, who created a Web-based project, known informally as The Beast to promote Steven Spielberg's movie Artificial Intelligence (2001). (The sites that made up The Beast are no longer live, but archives of them are available at the Cloudmakers site.) This project was built around a carefully-plotted story related to the movie's plotline of a future time in which humans coexist with human-like, artificially intelligent creations. The story evolved over the 12 weeks leading up to the release of the movie, and was told through the creation and frequent updating of a collection of 35 Web sites augmented with mailing lists that visitors could join, voice mail, faxes, clues in print ads and TV commercials, and in-person events. This approach was designed to create an immersive experience, placing the visitor directly into an Internet-based version of the world envisioned by its creators. All the project content was designed to fit together and change in sync with the real-time evolution of the story, just as if it were really happening. Like the Blair Witch site before it, this project created a huge amount of publicity for the movie, drawing about two million unique visitors over five months [10]. The Beast was a unique experiment in marketing of any form, and was a significant step forward in our understanding of the nature of stories and games, and how they influence -- and are influenced by -- the medium of the Internet.

Meanwhile, parallel changes have begun taking place in the videogame world, where more and more story content is being incorporated into games and game-related books and Web content to provide motivation for the games' action sequences. These two trends -- games as stories and stories as games -- recently came together in the campaign created by 42 Entertainment (a company formed by the creators of The Beast) for Bungie's video game Halo 2, via the I Love Bees site. Starting as a simple Web site for a Napa Valley honey company, the site evolved over three months to tell a story of how an AI entity had crash-landed on Earth, took control of the I Love Bees site as it attempted to reconstruct itself after the crash, and began to interact with -- even threaten -- the innocent owners of the site and the broader community of people experiencing the story. As players attempted to understand the site's ongoing evolution by working through ambiguous and sometimes confusing content, puzzles, telephone calls from the game to phones throughout North America, and other game-related activities, they were rewarded with downloadable audio files that, when properly ordered, created a ten-hour audio play related to the story underlying Halo 2 [9]. And so the merge, or cross-connection, of these genres continues.

Design principles for Web-based storytelling

Other projects have followed in the footsteps of those above, some more successful than others. But in our current pre-theoretic state, all have contributed valuable information in helping us understand what does and doesn't work in this use of Web sites as storytelling devices. Looking over a broad collection of these experiments, a rough set of design principles for this style of storytelling has begun to emerge:

  • Plot and character dominates. As with any form of storytelling, a good Web-based story depends on a compelling plot and interesting characters. Nothing about this critical part of storytelling has changed.
  • Build your world as a set of Web sites. Create the world of your story by building one or more Web sites corresponding to the characters and organizations featured in the story. The visual designs of the sites are critical: They must look appropriate for the roles they play in the story. Characters in the world may have other forms of Internet presence, such as e-mail, just like real people.
  • This is a game... The interactivity inherent in The Beast and other events like it have blurred the notions of story and game to the point where it often seems more natural to speak of these events as games, and the participants as players. This is a linguistic tendency that I'm likely to adopt in this report.
  • ...but it's not a game. Release the sites on the Web as if they had been created by real people and organizations. The site design should minimize any evidence that the sites and the events depicted in them are fictional [5, 6].
  • Tell the story by evolving the sites' content, in much the same way that real-world Web site and blogs change with time and reflect events taking place in the real world. Instead of simply telling the story, show the results of actions taking place in the story's world and leave it to the players to figure out what those actions were and why they occurred. In a very real sense, this is one big exercise in inductive reasoning. Participants are challenged to actively understand what's going on in the story world, so that they can properly interpret past events and predict future ones.
  • Monitor and adapt. Like any Web-based property, it's important to watch the event carefully and make sure that your story is getting across. If it's not, it might be possible to improve the flow of the game by adjusting the design of the Web properties and the ways in which they convey the story to the players, or, as a last resort, revising the story itself.

Design principles in action: The creation of Exocog

On a personal basis, I found myself tracking the release and evolution of these projects, and getting more and more interested in them. As an interaction design consultant who often works on Web site design and development, they struck me as being a new and creative use of the technologies I work with every day. I was also spending more of my professional time on the use of the Web as a marketing tool, and these events brought a new perspective to many of the promotional issues with which I was concerned. They also raised the more pragmatic question, always in the mind of a consultant, of whether there was a business opportunity worth pursuing here.

But how to pursue it? I had no connections to movie studios or their marketing and advertising agencies, and nobody was going to hire me to promote their movie, sight unseen. My solution was to adopt a guerilla strategy: I would just pick a movie and act as if I had gotten the job of promoting it. Better to ask forgiveness than permission.

There were obviously some things to be wary of in this strategy. I would need to be careful to not tread on copyright and other legal issues. In fact, this wasn't hard. From a promotional perspective, the goal of a project like this isn't to tell the story being told by a movie, but simply to inhabit a space thematically close to the movie. As Blair Witch and The Beast suggested, telling a back-story or side-story related to the movie, but with a completely different plot and characters, would be a reasonable way in which to proceed.

So, in early 2002, I began to look for a movie that would be a reasonable subject for a self-funded, four-to-six-week event. I chose Minority Report; like The Beast's target of AI, another Spielberg movie, this time based on a short story by Philip K. Dick [3] about a future society in which "precogs" -- humans with the ability to foresee the future -- were used to detect and prevent murders about to happen. This made sense from a number of perspectives. Its science-fiction focus would, I presumed, be a good match to an Internet-based audience, and the story had a conspiratorial tone that promised the kinds of suspense and conflict needed for a good story. I expected the shooting script of Minority Report to be rather different than the original story, but that was fine: My concern was with the overall world depicted in the story, and whether I could find an appropriate space nearby in which to experiment.

So, I got the project started: I formed a six-person team of writers and visual designers (all volunteers), started developing the story and the Web content, and got the event ready for release. Before going into detail about how this was done, and how the design principles discussed above influenced the event's design, it's probably best to provide an overview of the story and how this played out on the Internet.

The story behind Exocog

In the interest of staying away from the plot of Minority Report, I chose to avoid the crime-fighting orientation of the movie and instead wrote about a near-future world in which precogs were becoming the basis for commercial activity. This gave me a viable set of heroes and villains on which to base the story, while not risking any collisions with the content of the forthcoming movie.

Briefly, the story is:

A company called Exocog has announced their intentions to provide precognitive services -- product forecasts, competitive analyses, and the like -- to any client with an interest and ability to pay. However, Peter Glenstone, Exocog's unscrupulous president, has a secret plan to use precogs to collect government secrets that he can then sell to foreign agents or organized crime. He has begun a collaboration with the Institute for Precognitive Studies, a university-affiliated research institute, and has bribed Richard Dexforth, the director of the Institute, to find ways of getting maximum results out of a team of precogs, even at risk to their health. But these experiments are not working, no matter how hard he pushes. Peter concludes that his only viable way out is to bilk his investors out of as much money as he can, and leave the country. If bad things happen to the precogs along the way, so be it.

Meanwhile, a young woman, Sarah Ames, is trying to find out why she has lost contact with her precog boyfriend, Jason O'Reilly. Sarah eventually discovers that Jason had volunteered for an experiment being run by Exocog, and is now being forced to participate in Peter's experiments. Sarah is contacted by other precogs (unofficially referred to as the Precog Chorus) and by WeTheFuture, a secretive precog activist organization led by Seth Rydel. They, with the help of Richard, who comes to realize the wrongness of his actions, break into Exocog's offices, free Jason, and expose Peter's plans, sending him to jail one step ahead of his angry investors.

While I was happy with this story as a starting point for the project, what's most relevant to the current discussion is the process by which the story was turned from a traditional narrative into a set of Internet-based events that was played out over a five-week period.

Defining the sites and the characters

Our team began the design of the sites for Exocog and the other organizations with a basic outline of the sites' requirements, and proceeded through much the same design process as we would for any Web site. After all, that's exactly what we were doing in the context of the event -- building Web sites for clients. Of course, these designs also needed to be informed by the plot of the story, so that all the appropriate sections of the sites would be in place and ready to reveal whatever bits of information the story called for.

The sites used in Exocog were:

  • Exocog (www.exocog.com): A typical corporate Web site of about 25 pages, with descriptions of its products and services, office biographies, press releases, and an entry point to its corporate Intranet. Game participants eventually obtained "stolen" logins and passwords for the Intranet, and were able to read Peter's e-mail and private documents. Visitors could also join a mailing list and obtain reports about upcoming Exocog product announcements.
  • Institute for Precognitive Studies (www.precogstudies.org): Quite the opposite of the Exocog site, this was an intentionally simple, even primitive collection of about 25 pages. The Institute also had a mailing list and Intranet, and accepted inquiries into research partnerships and visiting scientist program.
  • Sarah's blog (blog.personalfusion.com/sarahblogger): A typical blog (built, in fact, with the commercial Blogger tool), this was the announced entry point to the game. It featured frequent commentaries from Sarah and occasional messages from the Precog Chorus who had insights into what might be happening with Jason. She also had a mailing list, and invited interested people to "join her blog ring."
  • WeTheFuture (www.wethefuture.com): Stylistically edgy, the participants behind WeTheFuture were prone to manifesto-like statements about the state of precogs in the world. They were also highly skilled at finding and publishing embarrassing documents about Exocog.

We made every effort to create the illusion that these sites were real, with real people and organizations behind them. The sites were registered and taken live with appropriate domain names. Characters identified in the game had functional e-mail addresses, which were either stated on the sites or were easily guessable. The Exocog and Institute sites also offered addresses for typical corporate functions, such as jobs@exocog.com. Mail sent to most of these personal and corporate addresses activated some sort of autoresponder, which returned an appropriate but noncommittal message to the sender.

Together, these sites formed the foundation on which the Exocog story would be told. But how the story would come together and be told through the evolution of these sites' content is another matter altogether.

The writing process

At its most basic, there was nothing unusual about the story's creation. As it evolved, I worried about the same things any author worries about -- whether the plot made sense, whether the story's structure held together, whether the conflicts among the characters properly motivated the action. I gave some thought to how the story would ultimately translate into a set of Web properties, but not a lot -- my primary concern was to create a viable story and a set of characters that players could care about.

Once I got to a reasonably detailed "treatment" of the story -- a two-to-three page summary -- I abandoned the usual narrative form, and began to specify how the story would play out on the Web: How do the Web sites and related Internet properties change over time, and how do these sequences of events tell the story? My first attempts to go directly from the treatment to specific events didn't work -- the jump in detail was too great. So I broke the story down into a set of five week-long "episode summaries," which described how the overall story was advanced during that week of the event. For example, the episode summary for Week 2 was:

Traffic heats up on the WeTheFuture page. One of the precog chorus announces receiving an unusual vision or three. Sarah and Seth both discuss, in their blogs, a meeting they had. Jason sends an email through some means -- something of a cry for help.

At the end of the episode:

  • We know and are in touch with some precogs with real powers, through Sarah's blog and/or WTF.
  • We have evidence that precogs are being mistreated by Exocog.
  • We hear from Jason through a third channel, which means we might have someone inside Exocog helping him (and us). He implies that, if he's lucky, he might be able to get these out on a regular basis.
  • We've gotten a couple of clues about Exocog's shady business deals.
  • The collaboration between Exocog and the Institute is clear -- Richard is in Peter's hip pocket. Exactly why, we're not sure, but a combination of money and blackmail seems likely.
  • WeTheFuture has joined the fray, but there are tensions between their "big picture" concern for precogs and Sarah's primary concern for Jason. Nevertheless, they're a good source of hacking skills, and passwords might show up there from time to time.

Evolving sites tell the story

These episodic outlines provided the right level of detail for me to plan exactly what would happen on the sites to convey this part of the story. Here we encounter the critical question of the project: How do you tell a story by manipulating the content of a set of seemingly real Web sites? Sometimes, of course, simply laying out a piece of information for the players to read and act on was the simplest and most direct way of moving the story forward. This was the primary technique used in the WeTheFuture site and on Sarah's blog: these were good places to report on Sarah's investigation of Jason's disappearance, or to reveal stolen e-mail or documents that advanced the plot in some way.

However, in real life, chains of story-like events don't simply appear in print for us to read and understand. Most of the time -- and especially on the Web -- we don't see people doing things; rather, we see the results of what they have done. We only know what has happened by collecting the bits of information available to us, and inferring what must have happened in the world to cause the observed events to have occurred. So it is in the game world: Most of the significant events that transpired in the Exocog story were conveyed in this indirect manner. We used several techniques [7] to lead our players through the story:

  • Content ambiguity: Give the players some information that's ambiguous at the time of its release, but that provides some indication of where the story is going and that may be resolved by future events. For instance, an e-mail message slipped out of Exocog by a friendly informer might refer to a previously unknown and potentially dangerous experiment. The uncertain contents of this information will give players some new things to look for as they move ahead, and perhaps lead them to re-interpret some things from the past that hadn't made sense at the time.
  • Cross-event integration: The payoff of content ambiguity: put several individually ambiguous events together, and the meaning of all of them becomes clear.
  • Storytelling by puzzles. Suppose Sarah receives a text message from her boyfriend on her cell phone:
    38 37 104 45
    Is that a telephone number that she (and the players) should call? The IP address of a Web site she should visit? Latitude-longitude coordinates? There's an implicit puzzle posed here to not just Sarah, but to the players as well. The impact of such information on the story is generally not the content of the puzzle itself, but the information that players get from the solution to the puzzle (e.g., the existence of a high-security Air Force base at those lat-long coordinates).
  • Information hiding. Use HTML or stylesheet coding tricks to make text invisible on a Web page as rendered in a browser, but visible if one inspects the source for the page. Or, use stegographic or other techniques to hide messages inside the GIF or JPEG images that make up a Web site.
  • Out-game property manipulation. Maybe Peter has an account at Amazon -- it could be useful to know he was interested in a book called Hide Your Assets and Disappear.

These techniques, combined with the week-sized level of detail in the episode summaries, allowed me to define what would happen on the sites to convey the various parts of the story, down to the level of what would happen on a particular day and which team member was responsible for it. For instance, the Week 2 episode summary above expanded out into the following set of events for Tuesday of that week:

LIZ & JIM: PM: Image of a cellphone message goes up on Sarah's blog: the Air Force base coordinates. "lv u 34 45 104 45", which we're going to use as the location of the government group of precogs about to be unveiled (partly) by Precog1/Anders. (This will become known as the "Cheyenne wiretapping" experiment inside Exocog.)

LIZ & STEF: Noon: We are introduced (via Sarah's blog) to Precog 2, who has information about something totally different (he/she sees armed men at a pier, waiting, staked out for something, but nothing's happening, sometime in August?) This makes no sense, on the outset; he/she will try harder to localize it. Sarah: "I'm putting this up in case it makes sense to anyone."

LIZ & JANET: Sarah and Richard trade mail via Sarah's blog about the WeTheFuture mail: Richard says that this was a reference to damaged servers, not injured precogs – "I'm offended, stop listening to these people who willfully misinterpret everything they see (and tolerate theft as well), etc." This exchange is summarized on Sarah's blog ("Even though I sent the original message out privately, I think I owe it to Richard to post his response more publicly.") However, Sarah wants to know more about WeTheFuture; she will meet with them and will report back, one way or another.

Note that these specifications are still rather open-ended; coming up with the ultimate content for the sites was left to the people responsible for the various characters. This gave the writers the creative room to make the characters their own and give each of them a particular and distinct voice. Note also the design challenge and the risk inherent in this approach: The site updates needed to be interesting enough to engage the audience, but simple enough that most visitors would solve them. Otherwise, the game would become either a boring affair that no one would care about or a puzzling muddle that no one could understand. In practice, achieving this balance can be difficult, and I'll have more to say about this challenge later.

Taking the game live

Once the sites were built and their evolutions outlined, it was time to launch the game. To do this, I posted messages to two Internet groups devoted to exploring game-like events like The Beast. We knew this limited announcement would not have the impact of the publicity behind these better-known games, but we were happy to keep this first project small, the better to watch it and learn from what was going on. In this message, I passed along the URL for Sarah's blog, said that I had gotten this from a friend, and wondered if anyone on the group knew what was it was about. We then sat back and waited to see if anyone took the bait. They did. The visitors were sophisticated in both these kinds of online games and in their knowledge about the Internet; they understood that someone was launching a game, and that they were being invited to play.

We were now live, and our primary job as puppetmasters, to use the players' term for the people running an event like this, was to stick to our plan and roll out the changes to the sites, unveiling the details of the story as we went. But, obviously, we couldn't do this running blind. The players' interaction with the sites needed to be monitored and evaluated, just as with any Web site. But how to do this? The level of detail in server logs was far too fine-grained and abstract, and didn't give useful information about whether players were able to follow the game and its story. Were they mostly able to stay on top of the story, or were they in need of some help?

Monitoring the game at a useful and informative level would have been almost impossible, were it not the case that Internet-based forums often spring up as a meeting place for players involved in these games. Too many things go on in the games for anyone to do everything by himself or herself. If players are to be successful -- and, of course, if they're to have fun -- they need a clearinghouse for reporting discoveries, debating theories, and generally socializing with each other, in and out of the context of the game. The Cloudmakers site emerged to serve this purpose for The Beast, and we were fortunate that a similar group formed at the Alternate Reality Gaming Network, or ARGN, the successor to Cloudmakers, to provide a place for Exocog players to congregate. (More recently, the Unfiction site has taken on many of the services originally provided by ARGN, e.g., the Unfiction forum for I Love Bees. Deaddrop is also a valuable resource for these events.)

These discussion boards are an invaluable resource to people behind events like Exocog, as they provide a very clear picture of what is and isn't working in the game. They also allow puppetmasters to be riskier in the design of their game: They can plant a clue in a relatively obscure part of the game world, and know that if just one player finds it, it will soon be publicized on the boards and many more players can add it to their understanding of the game. But, beyond their role as a feedback mechanism, it's fair to say that these boards are the real locus of the gaming activity. As the creators of Exocog, we may have been providing the fodder for the players, but the true activity of the game was the social interaction of the players as they worked together to work through the clues, and collectively understand the story. We came to realize that what we were really doing, through our Web site and e-mail manipulations, was creating a social event. The people who came to the ARGN board may have gone there in search of information, but if they stayed, it was because they valued the sense of community that grew out of the players' shared interactions on the board. As we proceeded through the game, more and more of our attention was devoted to nurturing this sense of community, through our own properties as well as the ARGN board. There's much more to be said about the relationships between games and these discussion groups; some further thoughts on how these groups co-existed with Exocog can be found in [7].

While we didn't have the budget or the outside marketing connections to get the visibility needed to attract large numbers of players, we were pleased by the response. We received over 150,000 visits to all our sites, and were able to identify over 2600 unique visitors by an analysis of server log IP addresses; once the multiplying effects of proxy servers were considered, the visitor count would surely be larger. A more detailed analysis of the logs identified 520 visitors by clearly unique host names and/or IP addresses, and we were able to evaluate in some detail how and when they came into the game, and to what extent they participated in it. Of these people, 38 percent came to a game site on more than one day, and, of them, 50 percent returned to a game site during the last two days of the game's activity, a date that had been suggested as end-of-game by headlines on the Exocog site and by the impending theatrical release of Minority Report.

Figure 1 (right) shows a plot of the number of people who entered the game during the days of its run. There were clearly spikes in participant acquisition, and we have never been able to learn what outside events led to these spikes. However, this graph also confirms that we continued to gain participants all the time the game was active. All told, we have good reason to believe that we were successful at creating an event that, throughout its run, continually attracted new participants and maintained the attention of those who got involved in it.

Looking back: What's hard about this?

Running the Exocog game was a very intense experience, growing out of the demands of keeping to a rigorous schedule and satisfying an audience with high creative standards. Looking back, there were three particular areas where the challenges were the greatest. None of these are simply problems to which solutions can be applied; rather, they are issues that must be managed.

Maintaining continuity

In events like Exocog, players scan the content of in-game Web sites with amazing care, in search of clues about how the game will proceed. If there are continuity errors on the sites -- incorrect dates, mixing up the names of characters, creating evidence that a character was in two places at once -- the players will find them, and both the realism and the credibility of the game will go down. Planning is critical, and an essential part of a game plan is a clear timeline of all the important events in the game, showing each game character's involvement and the character's involvement in them. The event schedules we created served us well; once the game went live, charts like Figure 2 (right) were helpful in keeping track of when the various events were scheduled to take place, and which ones had or hadn't already been taken live. In this figure, each note represents a specific event, with annotations recording its status in the game.

The corollary to having a good plan is to avoid making any changes to the plan in the hopes of making things somehow "better," unless the changes are absolutely necessary. It's hard to reconstruct all the aspects of the plan and check a new proposal against them while the game is underway -- it's far too easy for continuity problems to slip in. For similar reasons, we chose to not respond to the e-mail messages that players were sending to game characters. Aside from the fact that our ability to offer individual responses wouldn't have scaled to more than a small number of messages, the content of these messages would have been a ready source of continuity errors. (On the other hand, I Love Bees found ways to build player-generated e-mail, even individual telephone calls to players, into the game [9].) Some real-time tweaking of these kinds of games is probably inevitable -- even the best plan can have problems or even outright errors at execution time -- but it's wise to tread carefully here.

Keeping the game going

We alluded earlier to the risks inherent in using ambiguity as a tool in telling the game's story and motivating the players to keep exploring the game world. What happens if, despite their efforts, some critical piece of information isn't found, and an equally critical inference about the state of the game can't be made? The players' progress is halted, but the timeline is demanding that the game move on.

When this happens, several alternatives exist, none of which are good:

  • Put the game's progress on hold until the critical information is found by players. This runs the risk that the game goes stale and the players begin to wonder, privately and on the discussion boards, why nothing seems to be happening.
  • Keep things running per schedule, running the risk that the game's progress gets so far ahead of the players' understanding of it that nobody knows what's going on.
  • Patch the game in real-time on the fly: perhaps provide another route to the information.

Roadblocks like these occurred twice during the running of Exocog, although in both cases we were lucky and players broke through the blocks just as the game needed to move on. In general, this concern tends to push game design in a conservative direction, but this is better than finding a game stalled or confusing. There's nothing wrong with having a few parts of the game that only a few motivated people will find -- it's a good reward for their dedication. But it's not a good idea to put high-risk events on the main path of the story.

Getting enough content

From a puppetmaster perspective, there is simply never enough content for these games. Do a major site update, and, an hour later, the most devoted players have absorbed it and are begging for more. This strikes fear into the puppetmaster heart: You become convinced that these people, who have the greatest interest in the game, are about to get bored and abandon it. But coming up with more content on the fly isn't the answer. It risks introducing continuity errors into the game and ultimately just whets the appetites of devoted players: More content leads to a desire for still more.

The Beast used puzzles (sometimes remarkably complex ones) as a way of keeping players busy between updates. I Love Bees had their players chasing down calls to pay telephones so as to "activate the axons" of one of the game's AIs [9]. In Exocog, we came upon the idea of getting the players to generate their own content, which we could then introduce into the game in a controlled way. For instance:

  • The Institute for Precognitive Studies initiated a "visiting scientist program," and invited players to submit research proposals for the experiments they, as potential visiting scientists, hoped to carry out during their stay at the Institute. We collected these for a week, and then published them on the Institute Web site for "peer review" -- other players were encouraged to read the proposals and submit reviews of them. Over the three weeks that this program was in place, we received nine proposals and several reviews, all of which generated considerable game site and discussion board traffic.
  • The Institute also offered "Web-based precognitive screening tests," through which the Institute claimed to be able to identify people with significant precognitive abilities who might be suitable for future work with the Institute. Interested players were invited to submit a form on which they had made predictions about random events in the Institute's world (e.g., whether the Institute's director would gain or lose weight on his diet) in the hopes of receiving a high precognitive ability score and appearing on an Institute Web page that honored promising new precogs. Like the visiting scientist program, this testing was pure game play -- we made up the scores at random, and gave everyone who participated a score that indicated high precognitive ability -- but over 40 of our players took part in each of the three weeks of "testing" we carried out.
  • Thirteen of our players took Sarah up on the offer to join her blog ring, and sent us links to precog-themed Web sites that they had created as a side effect of their participation with Exocog. The sites, depicting such things as a precog-based healthcare provider, a detective agency, and a competing research organization, were created for no reason other than their creators' affinity with the Exocog game and the chance of finding themselves listed on Sarah's blog.

These activities and others like them paid off for everyone: We got new content for the site that enriched our game world, kept our players busy and interested, and reduced the pressure on us to generate new content of our own. They also provided something special to the players who put in this additional effort: The activities were way for them to gain a visible, personal presence inside the Exocog world. The players clearly wanted some way to get inside the game, and these techniques gave us a way to let them in, but in a controlled and managed way that didn't overwhelm us or threaten the game's stability.

Three issues for the future of immersive games

As I look back over the Exocog project, and look forward at the opportunities that lie ahead, there are three issues I think are central to these kinds of events, and that need some careful attention.

How do you build an audience?

Reliable data about the sizes and natures of the audiences for these kinds of events are hard to come by. Most of the events we've discussed here are marketing-oriented, and businesses generally don't release that kind of information, whether good or bad. The few statistics that are publicly available include reports that The Beast got two million unique visitors over its run [5], and I Love Bees reported receiving one million visitors in its first three months [11]. While these are impressive numbers, especially for Web sites that receive little if any formal advertising, a statistic I'd very much like to see is some measure of the number of committed players in these games -- perhaps the number of people who came to a game site three or more days per week -- and how this statistic changed over the run of a game. The "80/20 rule" will almost certainly apply here: Most of the activity in a game will come from a relatively small percentage of the participants, and there's nothing wrong with that. But, speaking from my own experience, it's easy to find yourself evolving a game in ways that will primarily benefit your more vocal and dedicated players a more enjoyable experience, even if those differences make the game less accessible to casual players. Is it possible to keep the dedicated players happy while still reaching out to a broader audience?

Building an audience is hard in any medium, and it's perhaps especially hard in these kinds of events, where the Web sites evolve as the game proceeds. It's hard for a new player to join a game like Exocog in the middle, because they can't experience the evolution of the game sites from their initial state to their current form. It's not like opening a book in the middle, getting interested, and backing up to start at the beginning. Volunteers on ARGN-type discussion boards often provide summaries of "the game so far" as a starting point for new players, but there's considerable risk in betting the success of your game on the chance that some unknown person will do this (and keep it up to date, and do a good job, and so on).

As much as we hoped that a discussion board would evolve for Exocog, we also tried to address these concerns in the game's design. This was another part of the reason for our taking an episodic approach to our story. While we inevitably built the game around the telling of a single story, we tried to structure the weekly episodes so that players could enter the game at the beginning of any episode and need only a simple summary of the preceding episodes to follow the new activity, and, ultimately, the outcome of the main story. Providing this history was the purpose of the weekly summaries on Sarah's blog -- to get all the players, including the newer ones, onto an equal footing before moving on. (Dana's blog served a similar purpose in I Love Bees.)

Of course, not all games are, or should be, alike. There's nothing wrong with a game that caters to people willing to make a commitment to their participation. Modern video games require their players to make a substantial commitment in terms of time, software, hardware, and sometimes monthly fees, and that hasn't stood in the way of their selling millions of units. In other cases, long-term player involvement with a Web event may not be essential to meeting the event's business goals -- impulse purchase decisions might only require a brief interaction with a game. Ultimately, these issues are different facets of a much broader set of questions about the structure of these games and the relationship between a game and its audience.

How should you situate your event in the real world?

One of the main bits of conventional wisdom for these kinds of events is that you must portray their Web presences as being completely real -- any evidence that this is "only" a game is to be avoided as it breaks the illusion of reality you are working so hard to create. We followed this rule with Exocog; the only explicit indications that the game was not real was that the pages had a copyright date of 2013. While maintaining this illusion wasn't particularly difficult, I'm no longer so sure about the need for purity here.

What must happen for a game experience to be successful is that players must be able to act as if the experience is real, regardless of what they know to be literally true. Players experiencing The Beast obviously knew that it was not the year 2142 and that robots with human-like intelligence were not walking the streets among us. Similarly, it was clear from the Exocog discussions on the ARGN site that our that players were quite able to think about their participation in the game in different ways, treating the game as either a reality in which they were interacting ("We have to warn Richard that Peter wants to kill him!") or as a subject for discussion ("Is there going to be another site update today?"). It has always been possible for people to enjoy movies, books, and the theater while knowing that these experiences are staged creations: In all these cases, the audiences choose to and are able to consciously take a stance of reality towards these events, even though they know them to be fictional. Web-based events should be no different.

Further, it's possible for an insistence on faux-ness to work against a project. For instance, part of the promotion of the movie Godsend included a faux Web site for the Godsend Institute, a fictional research institute that, in the movie, was able to create human clones. While visitors could reach the Godsend Institute site from the movie's more traditional press-kit site, many visitors to the faux site arrived there as a result of doing a Google search for information on cloning, since the marketing organization had purchased "cloning" as a Google AdWord. Since there was no explicit connection between the institute site and the movie, and since the content of the site was not significantly different than real pro-cloning sites that can be found through similar search techniques, it was quite possible for visitors to mistakenly assume that the Godsend Institute was real. Not only did this absence of a connection fail to connect the visitors to the movie -- the ultimate point of a movie's marketing campaign -- but the controversial and emotionally-laden nature of the site and the movie content ultimately led to a considerable amount of bad publicity for the movie, once the site was determined to be just a part of a movie promotion [4].

What determines when or whether you can relax the rules of faux-ness? Our recent experience with these sites suggests that there are several factors in making this decision:

  • The extent to which a game's premise is obviously non-real. As with The Beast, the description of a world of robots and intelligent houses should be sufficiently non-real that it's probably safe to portray the Web sites as real. But since claims about dramatic advances in cloning technology have not been uncommon in recent years, more caution with the faux-ness of the Godsend Institute site would have been wise.
  • The way in which visitors discover the game. If visitors arrive at the game from a game-oriented Web site, a TV commercial for the promotional target, or press coverage of the game itself, it's reasonable to assume that they know they're in a game world, and you can proceed with faux-ness. But if they arrive through a Web search on a not-uncommon term like "cloning," that assumption is less reasonable.
  • The business intent of the site. If the site has been put in place as a promotional device, its faux-ness might be relaxed to be sure that the promotional connection between the site and the subject of the promotion is evident. (See the Stepford Wives site for an example of a site that uses faux content in a way that is clearly associated with the movie it's promoting.)
  • The subject matter of the site. The controversial and emotional aspects of cloning expressed on the Godsend Institute site, when portrayed as reality, clearly upset some of its visitors. These concerns might have been lessened if the promotional nature of the site had been made clearer. Or, pushing this issue to the extreme, the faux site technique simply may not have been an appropriate way to promote this movie's content.

These are only rules of thumb, and they can interact with the real world in unexpected ways. Orson Welles presumably thought that the idea of Martians landing in New Jersey would be sufficiently non-real that he could proceed with his broadcast without continual reminders to the audience that it was just a radio play. But the quality of the work and the tendency at the time to believe anything presented on the radio in a realistic way led to a quite different outcome -- in some cases, literal panic in the streets.

Are these games working as promotional devices?

It would be a mistake to rush too quickly into business-level thinking about these games. We're still in the pre-theoretical stages of their development, and we should be thinking broadly about the games and what practical roles they might play. But we can't ignore this discussion, either. Even now, these games take time and money, and if they're going to find their place among other forms of entertainment, sooner or later they're going to have to pay for themselves.

So far, there have been no significant Beast-like games where players have been willing to pay to play. Electronic Arts' Majestik (2001) tried flat-fee and per-month models; the game was ultimately unsuccessful, although it's not clear to what extent this was due to its non-free status. Experiments with alternative models are underway -- sponsorship and product placement, or purchasing in-game accessories similar to how There allows participants to customize the appearance of their avatars by purchasing clothing and hairstyles -- but the viability of these approaches are as yet unclear.

As a result, the primary business justification for Web-based events has been that discussed here: a promotional device, especially for the entertainment industry. This is a challenging role for Web-based events to play. They can be built relatively inexpensively, and they have the potential to reach a large number of people, but they are competing against forms of promotion and advertising with much better understood business models and a greater ability to predict their return on investment. So let me offer a few questions about the promotional aspects of these games that might help focus their design as effective promotional devices:

  • How much is too much? This is really a return-on-investment question: At what point does further development in a game stop returning value? Would a one-month version of I Love Bees have motivated the sale of just as many copies of Halo 2 as the three-month version? This is also a question about the degree of game complexity that potential players are willing to tolerate. If your goal is to attract a large number of possibly casual participants, simple is better than complex. But if you're looking to build a relationship with a smaller but dedicated group of players willing to commit time and effort to the game, complexity may not be a barrier to success. Both The Beast and I Love Bees attracted very large numbers of visitors, and complex, time-intensive multi-player Internet-based games like EverQuest have signed up hundreds of thousands of players on a regular basis, even at $10 or more per month.
  • What are you promoting? There are specific and general components to any promotion: I Love Bees is as much a promotion for Bungie's brand as it is for the Halo 2 game itself. Thus, extra effort on I Love Bees could be justified as a way of building a stronger connection between gamers and Bungie, even if the direct impact of this additional effort on Halo 2 sales is limited.
  • Don't count on free publicity. Don't confuse the publicity created by an event with publicity generated about the event. The properties promoted by the Blair Witch, The Beast, and I Love Bees sites gained a lot of attention simply because their use of the Internet was novel enough to attract attention to itself, and thus to the properties. As these events become more common, this secondary publicity will become less common.
  • What's the scope of the promotion? Is the event aiming for a short- or long-term payoff? Since these games generally play out over an extended period of time, they may be better suited for building an audience over time (as was the case with the Blair Witch site) than to simply motivate a quick purchase decision (e.g., "What movie should we see tonight?") Nevertheless, some of the techniques from these games, such as faux Web sites, may be effective ways of drawing short-term interest to a property.

A place to play

I believe we're seeing something significant in the ways in which games like The Beast, I Love Bees, and Exocog have made use of the Internet as a game and storytelling medium. The success of these events validates the notion of the Internet as a place to simply play. And this is a form of play that can make complex demands on its participants, keep players active and involved over weeks if not months, and serve as the birthplace of rich personal relationships. Most importantly, this sense of play is one that, by calling upon traditional notions of plot and character, bridges the gap between game and story.

We're at a very early stage in the development of this genre. The power and opportunities introduced by its technological base are still coming to be understood, and we're still learning the parameters of success and failure. Every experiment teaches us something useful, and I'm hopeful that my experiences with Exocog can provide some insights into the strengths and weaknesses of the genre, and where the genre may someday go. Ultimately, this is a story that is far from over.

Acknowledgements

Thanks are due to my collaborators in this project: Philip Andrews, Stefanie Jones, Sameer Ketar, Janet Miller, and Liz Miller. I am also indebted to the team at ARGN and the many people who participated in the game, especially those whose comments on the ARGN boards kept us on our toes, and to Chuck Clanton and Sean Stacey for helpful comments on an earlier draft of this paper.

References

1.   Coover, R. The End of Books. New York Times Review of Books, June 21, 1992.

2.   DeLong, D. F. Sex sells on the net. http://www.newsfactor.com/perl/story/8909.html. (Originally published April 12, 2001.)

3.   Dick, P. K. The Minority Report. In The Philip K. Dick Reader, 1987, Citadel Press, pp 323-354.

4.   Duncan, D. E. Hollywood takes a look at cloning -- and opens up a can of worms. San Francisco Chronicle, April 19, 2004. (Available at http://sfgate.com/cgi-bin/article.cgi?f=/c/a/2004/04/19/BUGH865P511.DTL)

5.   Fabulich, D. What not to do: A lecture by Elan Lee. http://pantheon.yale.edu/~dgf4/notagame/

6.   McGonigal, J. "This is not a game": Immersive aesthetics and collective play. Proceedings of MelbourneDAC 2003.

7.   Miller, J. R. Exocog: An experiment in games and storytelling. Proceedings of the CHI2004|ICSID Forum, 2004, Vienna, Austria.

8.   Sieberg, D. Reality blurs, hype builds with "A.I." game. http://www.cnn.com/SPECIALS/2001/coming.attractions/stories/aibuzz.html. (Originally published June 19, 2001.)

9.   Sci-Fi fans are called into an alternate reality. New York Times, November 4, 2004.

10. Stewart, S. The A.I. Web Game. http://www.seanstewart.org/beast/intro

11. Terdiman, D. I Love Bees game a surprise hit. Wired News: http://www.wired.com/news/culture/0,1284,65365,00.html. (Originally published October 18, 2004)

Notes

Links to external sites may go stale, as the sites are repositioned and repurposed by their owners.