SXSW Xcript: Will Wright Keynote

Will Wright gave the keynote talk at the Hilton Grand Ballroom on Tuesday, the last day of 2007’s South by Southwest Interactive.

Justin Hall introduces: Will Wright is a famous successful computer game designer. He created SimCity, which mapped birdseye urban planning into millions of minds. Working on a dollhouse for boys he created the Sims, which allowed all of us to manipulate suburban households. Billions of dollars of revenue, putting him in the highest echelons of entertainment. What does a guy like him do for a hobby? He runs the Stupid Fun Club in the East Bay. I visited the club, and the night I was there, there was a video shown of a robot laying on its side in theh street asking for help, and someone had taped the responses of passers-by to what was basically a homeless helpless robot. Then someone handed me a plastic visor and body suit, I put it on in the spirit of the evening, and suddenly this robot was rapid firing ping pong balls at me. I could see in the back of the room Will Wright behind the controls steering and watching and I think wondering how long I would stand there. This idea of experimentation and testing with things, he’s made it possible for all of us to experiment with the systems around us. Now Will Wright is building a simulation of the universe. Wow.

Will Wright (one arm in a sling): All those pictures you just saw [projected on screens before the talk] are from the Hubble. I broke my arm skiing, before you ask. I had way too much coffee today, so I’ll go fast. They asked me to speak here, I decided I’d come and talk about story. Then a week ago I read that I would be demoing Spore, so I’m mashing the two together.

I love the idea of naming a conference after its location, its coordinate system. To be more accurate they should specifiy the space-time coordinates [which he projects on screen] though of course that would be less memorable.

Why I hate stories that my computer tries to tell me. Story is kind of in the model we’ve inherited, movies are almost what we’ve been looking at as heritage. Before get to story: The way I look at world is as a simulation, all these things happening, a state of the world. We turn the clock and cetain things cause changes in other things. Story is following one causal chain and presenting it to the viewer.

Stories tend to be unchanging, very linear, whereas games tend to be open ended. Game is vehicle for player to explore difft paths and directions. Movies are primarily visual, games are primarily interactive, so whenever we take control away from player at all we are taking away the most important thing about games. Like going to a theater and showing a blank screen.

Games inherently are this branching tree. Linear sequence is the basis of story. There’s a topology difference, apart from that is the point of view. There’s a controlled POV when watching a story, so the storyteller can present an engineered dramatic arc. Games on the other hand stops each time you lose, the arc stops. So the arc doesn’t feel like a dramatic arc. One thing we think of is that linear drama more compelling for that reason.

Stories are really based on lot of properties. Language, imagination, but most important for me is empathy, the ability to put ourselves in the shoes of someone else on screen. Actors are emotional avatars. We can inhabit that person and feel what they’re feeling. Film deals with this rich emotional palate because they have actors. Games tend to appeal more to the reptilian brain, the basic instincts of fear and aggression.

But games have a different emotional palate, not that they don’t have an emotional palate. Pride and accomplishment, guilt, these things are felt in games, but are not felt in watching a movie. I once beat the hell out of my creatures in Black & White, I felt terribly guilty. I’ve never felt guilty watching film

The circuit in our brain that makes stories appealing to us is empathy. Whereas in games it’s more agency, the fact that I’m causing what’s going on on the screen. Movie: What’s going to happen next. Games: Can I accomplish this?

These models are cognitive technology, the original educational technology we evolved to understand the world around us, which involves abstracting the world. Both really respond to being stuck in time. We want to be able to move experiences outside either true space or time. A cveman telling another caveman a story about almost being eaten by tiger is transferring thath experience through time and space to another person. This is why we evolved these technologies.

Experiences we have thruough play or story, build abstractions, can use to predict future and change behavior. A lot of the role of a story is to build potential futures. A.C. Clarke: The best way to prevent the future of to predict it.

Let’s get into story structure more: When characters are introduced at the beginning, the structure of whats’ going to happen is very fuzzy. A wider set of possibilities could unfold as you think into the future. Once Star Wars starts, we see these sequences unfolding, and every time something happens it narrows the range. We fill in the adjacent possibilities as we watch this. When Indiana Jones is running out of temple avoiding traps, you imagine all these little branches of possibilities — what if that boulder crushed him — that don’t happen, but that’s what makes it interesting.

Other things about Star Wars, the grit and realism and back story, the dirty spaceships, this fills in that possibility space. Toward the end of stories you want to amplify, small events lead to dramatic outcomes. At the end of Star Wars there are two possibilites: the rebels are crushed or the death stars blow up, and it depends on two torpedoes finding their target. Minor events push big outcomes.

The biggest obstacle in interactive storytelling is knowing the ending. In linear storytelling the director knows the end, knows which minor events to present. We don’t know in interactive.

It’s like chaotic systems. If you drop a ball on top of an upside down bowl it might roll to one side. Move the ball just a little in the other direction, small change, produces a wide result [the ball rolling to the other side]. This is more extreme in storytelling. Because a glass of water fell, our hero ducked, the assassin’s bullet missed him, and he was able to go on and save world. In linear drama you can show causal chain.

A lot of linear storytellers are playing with the causal chain in interactive ways. One thing that’s pretty typical, to develop a lot of different subplots that over time merge together, like in Magnolia. This came from literature. Extreme version in Timecode, four sections of screen in parallel. One of my favorite types of stories, going along, sense of where its going, then takes total left turn on you. The Matrix is an extreme example. Here’s a hacker, then he takes a red pill and all of a sudden holy shit. Fight Club. Definite expectation not only in story chain but in entire package of movie.

Memento, at some point each future point in chain caused you reevaluate what you’d seen before, you had to go back and rebuild large regions of the causal chain in your head. Like a puzzle game.

Groundhog Day is one of my all time favorites. Interesting, felt most like a game. Linear sequence, but all of a sudden it’s 6:00am again, he does it again, then again, again. What’s interesting about Groundhog Day, it was a game, he had to restart. This is a really interesting example of where the audience knew the past, so every day the director could skip more and more of the sequence. In our imaginations we cover almost an eternity of experience in this succinct way.

As game designers we do know what the player did, he failed the same level three times, so why not let him skip it this time. We should be doing more of this in games.

Game stories, the initial things with games, some early adventure games, branching stories. Slide of The Cave of Time book. Very early games were branching trees. That gets very expensive when you have to develop a million branches. More common is the gating tree [which is composed of many branches that periodically all end up at the same place].

There’s different topologies for the gated story, little sub goals between the gates, little gates between large gates.

There’s also hybrids between gated and branching. Still, we have to develop a huge game world, and the player explores very few branches. That’s the downfall of the tightly topographical branching story.

There are more interesting approaches, like Facade, which incoporates the idea of story fragments that have trigger and result conditions. Certain triggers match certain results. Put them together like LEGOs and form story over time. You have a pool of these, you put them together by matching tags, chain together, makes some causal sense and has more potential than is being explored.

One of my favorite short stories was by Bruce Sterling, called Maneki Neko, about a computer you wear on your hip. The computer says, oh if you got few minutes, stop and buy milk at this convenience store. The character buys milk, then as he’s getting hom the computer tells him, oh go one block over here, deliver milk to lady. He opens the door, there’s a woman with screaming children, him bringing her milk totally saves days, it’s a minor effort on his part, for her iot’s a lifesaver. It’s like a karma computer that keeps track of the easiest way to help this person’s life and who can do that at the least expense. The more you obey, the more it gets people to help you.

So how do you let the player create the story? The generation we’re dealing with now is used to idea that media is malleable, YouTube, MySpace, media is something they directly mold. When I grew up and got my first computer, they were thought of as fancy calculators. Nowadays it’s more of a communications device. So I think looking at technology as player-centered rather than braodcaster-centered, that’s totally flipping it around, the masses are empowered to create and saer.

Watching players play games is very interesting, they invariably come up with stories to describe to others what they did. Unique to them what they did in game. Three categories: unintentional subversive expressive.
1. unintentional — come across bug and report it in story form. There was a great spontaneous human combustion bug early in the Sims.
2. subversive — this is more where players are trying to push boundaries of game, push envelope in different directions. Find cheats and exploits. In GTA there’s a cool exploit where you can call in a million tanks. In BF1942, players would put on coordinated exploits. Really fun for them to play with boundaries of system and present results in story form.
3. expressive — more like in Sims, players have an intentional message. In GTA I spent the game developing my character and I had homeboys, I found a bicycle I did bicycle tricks, I was this semi homeless guy, I didn’t like missions, but the character became very real.

In the Sims we saw, when players were playing while sitting next to each other, they were verbalizing story as played. This was their interpretation of complex simulation, which they’d usually reduce to one causal factor though in fact it was far more complex.

So we put up a Web page where people could put stories, we ended up with 100s of 1,000s. Players got very good at not just playing game but creating stuff in game and telling stories. Some were initially very short but eventually got huge. Small are the size of novels that people are writing, plus the whole machinima movement. Some stories very complex, about real life things like someone’s sister’s abusive relations, 9/11 stories made in Sims.

Thinking about story telling, I think it’s always been much more interesting to me over the last few years to think about not telling the story but listening. If we can teach computer to listen to player, stories be much more powerful.

One approach we could have computer come to some understanding of the story the player is trying to tell — are they doing romance, comedy, horror — where the computer is slowly learning the story in players’ heads. It’s like natural language parsing, like, demolishing a sentence into components.

In story parsing, we could have the computer understand this is boy meets girl or this is teen slasher. If we can do that, we can look at goal states. We can even present dramatic obstacles to achieving those goal states, and this amplifies the drama, the whole thing comes down to an epic struggle, the computer has figured it out.

If we can parse intended story, we can change the presentation, the lighting, music, events. If they detect we’re doing a horror story, maybe the lights get low, spooky music, then zombie pops out, rather than random events, Then you’ve created a movie you can then capture and replay to someone else.

This is probably more likely to happen using parallel learning where we observe millions of players and pull out patterns from that. The vision I’m describing is kind of like Truman Show. Computer is more like director in Truman show, has certain abilities to influence and present but can’t violate Truman’s free will. Truman show and Groundhog Day are two of the most relevant pieces of linear entertainment pertaining to games. Truman Show is dealing with someone in a game. I wish games were more like Truman Show.

Trends in future: There’s this concept from games called magic circle, when people play a game together they’re sitting down and respect rules of game. People outside are not expected to respect the rules: we don’t cheat, or talk with partner during bridge game. I think story has a lot of the same thing, shared experience, story circles are like campfires 10s of 1,000s of years ago, which then evolved more structure. At the same time, our idea of story has evolved more structure, and we now have large formal things, things like the three-act structure, then it’s been shrinking back again, television, shorter shows in living room all way down to video iPods.

There are lot of opportunities, it’s almost fractal, we now have 3 minute things pulled off YouTube, so in some sense the story circle has been diversifying in time and space, we don’t have to go to movie theater or living room. Games are doing same thing, diversifying in time and space, you can have epic 40 hour game experience, or 2 minutes on cell phone.

If you look at how much time the average person spends consuming linear entertainment, it’s fairly flat. Interactive entertainment is still riding a generational wave. Younger people spend more time with interactive entertainment than with linear entertainment. There’s this cultural overtake process, where there’s an uncomfortable mixture of people who have spent a lot more time with linear, but a whole generation coming up where interactive is more compelling. So what this is driving, rather than games being about sports, they’re more about hobby. They’re a tool of self-expression much like a hobbyist builds elaborate train sets.

A lot of people have design aspirations, and we’re now starting to fulfill those aspirations with computers. Players love making content in games, we’ve actually been really riding that wave over last few years, they also love sharing and collecting. Power of that collective effort is turning out to be amazing. One thing we’ve dealt with for a long time, is quality versus quantity issue. Get lot of content but frankly most of it’s kind of crappy.

As we give them better and better tools [flashes a screen from Spore], we’re going to get in process of increasing quality of what they’re doing.

So we have players building models as the play games, now the computer can start building model of players. How they move across the gameplay landscape, what they buy, what they’re good at, social networks they form, social interaction frequencies, what they do with other people, from that we can build fairly elaborate computer models displaying and predicting their behavior. This was something we did for the Sims.

Spore takes this collection of ideas, gives the player a simple set of levers, the computer amplifies the results, and represents them with higher output relative to what they put in.

That asset they made now has value not only to them but to other players as well. So we collect them on a server, categorize them, take the player model, what we’ve seen about how the player played the game, predict what they’d like to see, and how do we close this loop? We want to take the player out of world of being Luke Skywalker and put them more into being George Lucas, make them the creator. Can we extract the entire world from their imagination?

A lot of the work on Spore was about player tools. In the Sims, most tools were external. There was a lot of friction for players to create stuff. What we wanted to do was take all the tools and fold them back into the gameplay. Then collect and redistribute automatically instead of making you find and download.

The result is we can build this infinite world to play around in.

[Here Wright demo’s Spore, which I’m not going to write about here since it’s been seen before — though this demo was more elaborate than the one he did at E3 last year. At one point, Wright’s young creature is eaten by a much larger creature: “I wasn’t supposed to die”]

I want this game to be something that brings up a lot of issue for players, from the history of life to where my life is going. If you look at the history of the non -biological universe, it’s interesting, but not as interesting as the history of life. Life effects can have staggering effects. I went to Montessori school till 5th grade, and Maria Montessori, her idea was if you build toys and let kids play with them, they explore and discover the principles of the world around them by themselves.

The games I make I think of in those terms, very elaborate Montessori toys. How can we lead players to come across interesting ideas of philosophy and science. This can be understood as a philosophy tool, get you to think about meaning of life.

[Demo’s the Space Age portion of Spore: “Eventually, I want to see interstellar was between the carebears and the Klingons.” He adds water to a planet, then adds clouds, which make greenhouse gasses accumulate, then heats the planet until the water burns off: “Okay, the whole planet’s melting now.”]

One of the interesting thing about toys like this is you can give someone a sense of long term dynamics in short time. One of the biggest issues as humans is, we’re so bad at long term thinking. If we can take these long term dynamics like climate change and bring them into a five-minute experience like a toy, we can use this as a tool for long-term thinking.

As we play through evolution, we present it as a toy version, but it captures most of the dynamics of evolution. As we move into the future, we want the game to be more and more focused on some of the fictional landmarks we have. A lot of gameplay at these levels is based upon my favorite sci fi movies.

[Wright flies his spaceship to a planet and drops a 2001-like monolith using the Monolith tool. The creatures on the planet start worshipping the monolish, which upgrades their brain level. Wright’s creature leaves his spaceship and comes down to the planet, trying to get the creatures to take him as their god, rather than attack him. “I don’t think he wants to worship me right now.” So Wright does some miracles.]

I can also scan these creatures and they get entered into my encyclopedia. We came up with the idea of a Sporepedia of everything you’ve come across in the game up to this point. There’s a database on every creature, vehicle, building, city, planet, star I’ve visited. We can also make a trading card game with this that you can play outside the computer game.

You can take any technology and understand it as an extension of the body — cars as an extension of legs, TV of eyes, houses of skin. With computers, the most important thing they can do is extend the imagination. They’re an amplifier for imagination. We’re already using them for education and other things. How will this impact world going forward? To me it’s a question I think about as I’m falling asleep. Every now and then the world goes through a major paradigm shift, we have to rebuild our basic model of the world. Sometimes this is driven by social changes, it happens maybe once or twice a lifetime. But with technology, it’s starting to happen more and more often, some of it comes from the grassroots, but the rates at which major paradigm shifts are happening is more and more frequent over time.

When you look at games specifically and entertainment in general heading into the future, games have a perception as being simple and meaningless, but really they can be much much more than that. They can allow us to develop systemic thinking, build much more elaborate, more accurate models of the world around us. Because of that, they potentially give us the ability to approach the future with just a little bit more interlligence than we had before. They can allow us to change the world just a little bit for the better over the rest of our lives. And that’s the end.

  • Trackback are closed
  • Comments (5)
    • Alvis Brigis
    • March 13th, 2007

    That is maybe the coolest transcript of a presentation I have ever read.

  1. Great transcript and of course great presentation.
    Will Wright seems to be an extremely intelligent person – no wonder he thought of such games in the past and it’s him now making Spore…

    Waiting to see it live.

    • Alvis Brigis
    • March 13th, 2007
  2. Fantastic work capturing this.

  3. One of the most interesting transcripts I’ve ever read. Thanks for
    posting
    this, its fascinating to read Will’s conjecture and commentary on gaming
    dynamics. “Imagination Simulator” says it all, which is probably why SL
    residents protest so vehemently when the SL grid is down for upgrades –
    their imagination simulator is offline! I certainly know what the pull of
    creating in SL is like :)

    Great work, guys.

Comments are closed.