The Loose-Fishery

Frequently, I find myself wanting to link to or post an excerpt from Elan Lee’s seminal 2002 GDC talk about the A.I. Game. But for some reason, I can never find it online. So here it is. —jb

————————————-

THIS IS NOT A GAME

A Discussion of the Creation of the AI Web Experience
by Elan Lee

A Story

It’s Thursday night. It’s raining out. You sit amongst the remnants of a cheap Chinese take-out dinner watching M.A.S.H. reruns. The knowledge that this experience will eventually be interrupted by an inconsiderate bladder is your only care in the world. A commercial comes on. Oh right, that new Spielberg flick with the kid from the Sixth Sense. You might catch that one… maybe not. You’re engaged in an absent-minded hunt for fortune cookie shrapnel gone MIA, so when a secret message burned into the film credits flashes on the screen, you almost miss it. Wait, what was that exactly?

Saturday. You’re strolling through the mall in search of the fabled convenience cart where the guy will replace the LEDs in your cell phone if you slip him an extra twenty. Warranties be damned! Oh hey, there’s a poster for that A.I. film again. Wasn’t there something bugging you about this? Something you were supposed to remember? A quick peek at the credits: Director, Producer, Composer… yup everything seems to be in order here. Names you recognize, names you don’t, Warner Bros., DreamWorks, Gaffer (what’s a gaffer?), Sentient Machine Therapist, Makeup. You really hate doing double takes, because you always feel like a cartoon character, but what the hell is a Sentient Machine Therapist!?!

Later, back at the batcave. (Note: Batcave refers to the corner of your apartment containing the outdated conglomeration of machines you’ve inherited from friends, that strange liquid dripping from the ceiling, and the imaginary butler with the self-esteem problem.) You find yourself on a quest for answers. As is the case with so many of your life’s adventures, Google will save you. A search for “Sentient Machine Therapist” yields the name “Jeanine Salla”. You already knew that. That’s the name you saw on the posters and on the commercial. But there’s one more juicy tidbit here… she’s got a web page.

http://www.familiasalla-es.ro/ seems pretty straightforward. Just more reinforcement for your belief that, like reproducing, people should have to pass a test before being allowed to create content for the web. You notice that the year on the site is listed as 2142, and that Jeanine Salla claims to have worked on the movie A.I. more than 100 years ago. Is this just a marketing campaign for that damn movie?

There’s about a dozen links here that point to universities where she went to school, a company where she works, and some friends’ web pages. All of which are dated 2142, and none of which even mention a film.

OK, here’s something — Jeanine Salla’s phone number. You’ll call the number and hear a commercial for A.I., or maybe get into some contest for free tickets. So you give her a call, only to hear that she’s had to go out of town for personal reasons and won’t be available for office hours.

You’ve been hitting sites for the last ten minutes. Everything you try seems real. You’re finding web pages, phone numbers, email addresses, and they all work. You’re receiving faxes, getting e-mail from real people, and stumbling across a world that seems entirely unaware of the fact that it’s just a game build for a movie promotion. It is just a game, right?

But you’ve probably done enough surfing for one day. It’s time to turn it off. Walk away. You shut down your browser, only to watch it turn itself back on, display a really creepy image of a corpse, tell you to GET OUT, and then turn back off again.

OK, fine, ten more minutes. (Now that Napster’s gone, you don’t have any big plans for the evening anyway.) You go back to the site, and pay a bit more attention this time. Turns out Jeanine Salla had a friend that died recently. Before you know it, you’ve hacked into a coroner’s web page and are reading details on the autopsy of a thermal engineer named Evan Chan.

The next week of your job is filled with cracking codes, tracing e-mail and reconstructing the events leading to a murder. You’re forced to stop for office meetings, but at least it gives you a chance to sleep. You’re not sure exactly what you’ve found, but it’s started phoning you, sending you email and urging you to show up at odd locations in far away cities. You’ve teamed up with thousands of others who’ve stumbled across this bizarre world and are working together to figure out who Evan Chan was, why he was alone with a custom built sex-robot, and what he know that would make his death so important.

Another Story (except not about you this time)

We built a life for Evan Chan to make his death look more convincing. We should ignore the existential ramifications of that statement for a second so that we can hone in on the point here. Why and how did we build this?

The why part is pretty simple, but people tend to frown on my explanation of “cuz it’s really cool!” so I’ll try to be a bit more descriptive. A few years ago, Jordan Weisman introduced me to the concept of ‘Internet Archaeology’. He said, “The Internet is like a big ugly pile of dirt. You spend your time sifting through the filth to find the gem.” Actually he probably said something a lot more eloquent, but I have a really short memory. OK, so we got dirt, gems, and a whole lot of unpaid archaeologists. That’s where we started.

What Not To Do

We envisioned a game that would blend into your life. Something that would page you in the middle of business meetings, send you email from bizarre sources, and relay secret messages encrypted in cell phone ring tones. People spend all day surrounded by a system that *could* make them feel like secret agents, world-class detectives, or the coolest kid on the block. Someone just had to go and turn the system on.

We wanted to build a game out of real life. It’s a really strange moment when you realize you don’t have to design a game engine because evolution has spent the last 4.5 billion years designing one for you.

But where do you go next? How can you build a board game that uses your life as its board? We were pretty sure that our “to do” lists would be pretty straightforward. Write a story, cut it up into pieces, deliver it in small bits through web pages, telephones, actors, etc. Much more important would be our list of “What not to do”.

Here’s what we came up with:

Don’t tell anyone.
Don’t define a game space.
Don’t build a game.
Don’t Tell Anyone

Here’s a quick tip to remember next time you find yourself in a pitch meeting with high level executives: NEVER SAY “OK, so I need you guys to commit a full budget, staff, office space, and considerable company resources to a project that I can’t tell you anything about. And oh yeah — once we’re done we’re going to work very hard to ensure we never tell anyone about it.” Strangely that scene in your head where everyone yells “Let’s do it! Ready… Break!” and then runs off in different directions never quite materializes.

It takes a while, but eventually everyone comes around to drink the ‘silence policy’ Kool-aid. It’s very effective for this kind of project, but it’s a really strange way to live. We couldn’t tell the press, we couldn’t tell our co-workers, we couldn’t tell our friends, and we couldn’t tell our families. There were only a handful of people at Microsoft that knew this thing existed. Friends of mine would say, “Look, I know you’re really busy with your secret project, but you should take a few seconds to check out this crazy new game going on around the A.I. movie.” My usual response was that I didn’t have time for silly games.

We had several good reasons to keep the cat in the bag. One of the really nice things about the web is that it’s big enough to provide a personal experience for everyone. You find the things that interest you and then you share them with your friends. That’s exactly what we wanted for your project. We needed each person to think of this as his/her own, so that they could claim personal responsibility for its discovery, and a sense of ownership. The effect would be lost if they found a banner ad pointing them to a game.

Another reason for the silence policy was to create a secret. (I can’t think of a more compelling reason to crave information than to believe that it’s a secret.) We denied knowledge of the project wherever we could. We had our denials memorized. Everyone on the project learned to mouth the words “no comment” in their sleep. It took about a week, but our silence was louder than any shout we could have mustered. We started to see curious snippets in the media, participants started to speculate, and the buzz machine for the mysterious A.I. web phenomenon was put in motion.

Don’t Define a Game Space

We wanted to build something that was not limited to the confines of the computer. This had to jump in your face and scream the electronic equivalent of “Booga Booga!!” Our mission statement went something like, “The instant you click on a link, your phone should start to ring, your car should only drive in reverse, and none of your friends should remember your name.”

To deliver an experience of that scope, we realized that we should be using whatever electronic gadgetry we could think of. I remember one of our earliest meetings was spent coming up with a list of every possible way to deliver a message. You know, email, sky writing, text messaging, singing telegrams, carbon dating rust samples off the hull of the Titanic; we didn’t miss a thing.

So after building a great way to deliver content, a great story, and great puzzles, we had everything in place to build a truly kick ass game. Well, everything except for that stupid third rule we were supposed to worry so much about…

Don’t Build a Game

This was the one that makes a game designer’s life a bit more surreal. “We are not building a game” became our mantra. We were building a page of real life. It had to look, smell, and taste real. Web pages had to appear to have been there forever, actors had to think of their actions and interactions as real events, and everything with the Microsoft name on it had to be tossed out the window.

We had to scour HTML source to ensure that nothing identifying was present. We had to register websites using fictitious names with functioning email addresses. We had to ensure that each website had a different look and f eel so that no one would guess they were created by the same person. We had to register phone numbers in the area codes that matched their fictional locations. We even went so far as to enter some of the characters into online dating services. OK, that last one took it a bit far, but it was important for us to understand the scope of the world we were creating and the importance of ensuring that like life, there were no rules.

To build an experience without the benefit of a concrete set of rules meant that we had to predict every action a user might take and have a solution in place. We had to be prepared for anything and everything so that the life simulation would be flawless.

I know we’d all like to think that we did a pretty good job at it, but we were far from flawless. I remember watching in horror as a random phone number emerged from a bit of binary, and the poor owner of the number answered countless calls assuring strangers that he knew nothing of any custom built sex-bots.

There were just so many random events that we couldn’t predict. Sometimes web sites were found before they were ready to go live, sometimes extra data was released in HTML source, and sometimes our actors did a bit too much improvising. We learned as we went, and the participants leaned with us. They became quite good at determining what was “in game” and what was not, and were (usually) good about backing off dead ends without (major) incident (mostly).

Production

When we finally sat down to start production, we had a very simple plan. We were going to work very hard for two months to create enough content for six months. During that extra time, when we’d have nothing else to do, we could create the remaining content for the rest of the year. (If word of this plan had gotten out, our inboxes would have been flooded with bridge purchase offers.)

The first problem came up immediately. Casual statements like “put up a company website” turned into Herculean tasks. We had to worry about registering the site, making sure it showed up in the right search engines, finding web developers, establishing a look and feel, creating content, fabricating a staff, inventing a product, creating email addresses and auto-responses for all staff members, registering phone and fax lines for the company, creating voice messages for the phone numbers, hiring voice talent, etc. etc. etc. OK, one web site down, five more to go… this week.

At the end of the two months, we all looked like zombies but we actually had a huge amount finished. We were nowhere near our goal of six months worth of content, but we were pretty sure that we had created enough to last for at least two months. The puzzles were cut up into four categories ranging from easy to whatever it’s called where your brains turns to much and slowly starts to ooze out your ears. We estimated that the former would take five minutes to solve while the latter would take the audience busy for several weeks.

Chalk up mistake #2.

The game took a few weeks to get off the group, but there’s a really important lesson that we learned here: We’re really stupid, and they’re really smart. That whole brain melting business we talked about, you know the three week endurance course of tedious mental anguish? Right, they solved that the first day they tried.

We wept like small children.

The content that we’d intended to last for weeks was being burned through in hours. This was the beginning of a very new kind of game creation. The kind where we don’t ever sleep, don’t ever eat, and don’t ever see our families. For the rest of the project we were barely able to stay one day ahead of the audience. The work load was grueling. The team worked for six months, seven days a week, 18 hours a day. But there was something rather special keeping us going.

A Team of Strangers

A group of people had come together to work as a team, and follow our story. We built challenges based on every discipline we could find. We wrote puzzles that required expertise in Photoshop, Greek mythology, 3D sculpting, molecular biology, computer coding, and lute tablature. We created strings of puzzles that no single person could solve on their own, and we found to our delight that it was working. The audience was forming teams, sharing ideas, writing applications, posting theories, arranging group meetings, programming distributed-client password crackers, creating art, and evangelizing the experience far beyond anything we would have been capable of.

By the end of the project, more than a million people had taken part. We challenged them on every front we could think of and they beat us every time. It was wonderful.

Evolution

We were not the first people to try something like this. The goal of transforming a person’s life into an entertainment venue has been around for a while. With the genre still in its infancy, I’m proud to think that we added something special. There’s a bunch of brilliant minds working hard on the next evolution of the concept, and we’re all looking for answers. How can we blend the fascination of a game with the emotional power of a work of art? Is this an effective marketing tool? Can it be profitable? How many people can we reach? What is a violation of personal privacy? The part that gets me most excited is that the only way to answer these questions is to build it and find out.