By Jim Rossignol on April 27th, 2008 at 10:53 am.
It’s a beautiful, summery Sunday morning. Sunshine is pouring down into in Bath’s hazy valley, and I’m sat by my office window, looking out across the trees, rooftops and rusting gas towers – one of my favourite views anywhere in the world. It takes me a while to notice the chat box that is blinking patiently on my desktop. It’s Eskil Steenberg, the chap who is currently working on his solo-MMO project, Love. At that moment he’s sitting somewhere in a Wifi-field in Tokyo. We talk.
Steenberg: Hi, online?
Steenberg: What time is it for you?
Rossignol: 10:28 AM
Steenberg: What time zone are you in?
Steenberg: Ah, thats okay. It’s 6 28 PM here in Tokyo
Rossignol: How are you enjoying Tokyo?
Steenberg: Very much, you might have seen on my blog that I’ve done some Akihabara shopping.
[We discuss an article I am writing about Steenberg and others. The subject is procedural content generation in games. We namecheck Introversion, Spore, Dwarf Fortress, and then move to discuss Steenberg’s work.]
Steenberg: My focus is very much on procedural story creation. That is my goal, and much of the content generation is aimed at reaching that goal.
Rossignol: Are these actual text-driven stories, or sequences of situations?
Steenberg: Rather stories and drama created by the actions of the players, and the reaction of the world. I want to replicate what I’m doing in my head when I write scripts.
Rossignol: So people dealing with emergent situations etc?
Steenberg: Yes, that sort of thing.
Rossignol: It does sound powerful. I think what MMOs are missing are these kinds of dynamic situations. You get it in Eve a bit, but only via player-player interaction.
Steenberg: Just like physics in games have become a natural cause and effect set up, so should your motivations, goals and experiences on a story level be. Eve is very much like that, but it is user created, the game doesn’t create any of the things that make Eve cool, the players do. But I think that Eve is a very interesting game in that direction.
Rossignol: The odd thing for me is it’s almost the only game in that direction, don’t you find that weird?
Steenberg: Yes, I do. Well there are a lot of games that get some of that by accident. My favourite game to accidentally create story is Populous the Third Coming, but even games like Halo3 and CS create those small pockets of events that play out as a story that involves: bravery, revenge, failure and so on, but they are very rudimentary.
Rossignol: So would you say that games, as story-telling devices, should simply facilitate player actions, so that we can tell our own stories? About our own actions?
Steenberg: Well that’s the state we are in now because the games are not proactive, they just say “here is this sand box, do something and maybe you will find it compelling.” The game is not actively doing something itself to create that. So games like Second Life (if that is a game), Spore, Eve, The Sims, and even GTA, create these interesting things because players make them interesting and use them in interesting ways, but that requires a lot out of the player and doesn’t immerse the player into the game, but rather invites the player to find the limits of the game.
Rossignol: So Love will be proactive? What sort of stuff do you want it to do to make players act?
Steenberg: Well that is the plan. Something that I’ve noticed is that building things gives players a lot of pride, and they are emotionally invested in what they build. Now I have given AIs weapons that can destroy buildings, that gives a very big emotional gravitas to what happens in the world. I think by creating AI that does both positive and negative things and that can react to player behaviour, you can get some very interesting gameplay.
Rossignol: So you think that story-telling in MMOs is going to depend on AI making interesting decisions as well as players making them?
Steenberg: Well in part. You need AI that can react and create characters, but I also thing there should be an over arching “Director” who can monitor the overall experience and make macro decisions. Like if the world is too calm, we need some more action, lets introduce a Klingon-style warrior race to the world.
Rossignol: So you want players to be forming “relationships” with AI as well as other people to make Love interesting? Your enemies won’t just be baddies to who attack, but beneficent forces too?
Steenberg: Well the more I’m working on figuring out Love’s AI, the more I think you are going to be able to befriend AI characters. They won’t have to be enemies. If you think: “I’m going to make a computer that can write Shakespeare”, you are going to loose your mind. You just cant get that done. But then you start breaking it down: Let’s take something like Star Wars. You’ve got this droid, that has the plans for the Deathstar. That is in essence a simple door key, problem that you have in Doom and million other games. You need object A to destroy Object B. Simple. Then you say why do we need to destroy B? Well, we’ve got to make B really dangerous, so we make it destroy something the player cares about, say his aunt and uncle. Each of these links are very simple cause and effect links, that strung together can create a story like Star Wars. So you create x, number of keys, and x number of bad things that can destroy x number of good things and you start getting a lot of variation. Obviously if the player fails to find the key making it impossible to destroy the Deathstar we let the game throw in a few different keys, while the players have to deal with the fact that the Empire destroys a lot of things wile they figure out how to stop them. Now imagine you create very many of these links, and the string them together in very long chains, and then dress them up very differently visually, you get a very complex story that just keeps going and going.
Rossignol: And this is made all the more complex by having multiple players, presumably?
Steenberg: My game is going to have multiple players, but ir really doesn’t need multiple players for the system to work. Take a look at all the events in Star Wars, (just to keep using it as an example) they are very simple in gaming terms. You owe someone money, you need a part for you spaceship, a probe finds your hideout, you need to find a master to learn new skills, you need to help your friends… and so on. None of these things are very hard to generate. But together they become this compelling story.
Rossignol: Why do you think that previous MMO designs have tried to do this statically, rather than dynamically, procedurally?
Steenberg: Well, again, the hard part isn’t actually generating the story, but expressing it. You need to be able to destroy the rebel base, build a new one, create cool characters and all that. That is very hard.
Rossignol: I suppose the challenge for you, then, is to create a system in which all this can take place, without seeming too chaotic?
Steenberg: Yes, you need to keep track of what the player does, notices, and can be a part of. I mean what happens if the player sleeps through the briefing telling him where to hit the Deathstar? The game needs to be flexible enough to be able to get out of that situation. Sorry, but I need to run.
Rossignol: Ok, cheers.