CR: Do you think that this kind of thing is... guys like Ernest Adams have been talking about this for years. This modular kind of procedurally-generated narrative, and games should be doing this. They really haven't been. And a lot of people have been talking about it. And it seems like the natural way games should be going, but it doesn't seem like people try it very often. Do you think this is actually where things are going to go? Do you feel like you're on the cusp of actually driving something?
PR: Yeah. I guess even I, when we went to GDC, and Clint was there, and John was talking about the repetitious stuff on his side. I did my talk about narrative design. We were kind of shocked, honestly, that... personally, I would have expected that by now, at this stage that we're at, in this generation, that there would be a whole schwack of other people all solving the same problem, kind of at the same time.
And I fully - given the number of talks on narrative that were happening - I fully expected to have somebody show up and go, "Hey, look what we did." And totally just render moot everything we've been working on. And it never really happened.
GR: There was Façade...
PR: Well, maybe Façade is even a good example, because it's like... I know Michael Mateas showed up at my talk, and followed it, and was there. I only found out he was there afterward. Otherwise I would have been probably a little nervous. Because I had attended the Façade talk before, and I was just like, "Ow, my head hurts."
But yeah, I walk out of there, and Michael's sitting down with Clint, and they're having this super-in-depth conversation. And Michael's, like, super excited, and blah blah blah. And I'm, "Huh, I wonder what that's about." And Clint tells me afterward, "Well, Michael saw your talk, and he was going on and on about some of the potential of this approach." And I was like, "Yeah, but didn't he solve all these problems? Him and Stern solved this years ago." We're just kind of emulating it.
But I think that in fact, a lot of people have just kind of been gun shy about trying to tackle some of this stuff. And who knows, maybe when our game comes out, it'll be proven why they should be gun shy. Maybe it'll be like, "Yeah, you guys totally went down in flames on this." But I think that what's potentially - to get back to your original point about "Are things moving in this direction, or should they be moving these directions?" I think that they absolutely should be.
Chris Hecker did his talk about the sort of decoupling of structure and content. And his point about AI, like the fact that we currently lack a kind of Photoshop of AI, a way of intuitively authoring the behaviors that will ultimately make for a more robust and life-like in-game agent. And I think that that is at the heart of this issue. Right? We really need to be able to handle our story systemically. And that means we need characters that can behave systemically when we give them things to talk about, when we give them situations to react to.
The micro-narrative solution, which I think is a very good solution, at the end of the day still needs to be... it still boils down to the smallest indivisible piece of content. And right now, that's still - as I said in my talk - a piece of animation, a piece of recorded dialog, all of these kinds of things. Or even a state change in a state machine somewhere.
And the thing is, we need to get to a stage where we can make that smallest, indivisible piece of content as small as possible. So that ultimately we can build AI-driven engines that generate narrative.
But I mean, we're way off, I think, frankly, from that interactive fiction/fantasy, and being able to build... with all due respect to Chris Crawford, I think we're a ways away from being able to have that in a form that is readily applicable to games. And maybe he doesn't care, because he doesn't think games are valid anymore. But I think that we still are trying to develop this in the context of making games.
And I think it's going to be an easier sell to people if we start with something that they're somewhat familiar with, in terms of the form, and they're able to pick the controller up and run around and shoot, and drive, and all that and then start introducing these other more systemic narrative pieces, so that it starts to become an expectation. We want to raise the bar. We want it to be... we want to set the standard by which other games have to try to follow us.
And even if we screw it up, I think we want people to say, "Yeah, these guys were pointed in the right direction. That's the direction we should be going in." And who knows? It remains to be seen. Like I said, we're totally willing to live with the possibility that we'll only get a piece of that delivered. But I believe very strongly in this approach.
CR: How much more recorded dialog will this mean, really, versus a traditional script?
PR: That's a very good question. As a shooter, we have quite a bit. We're running in at about 100,000 words of dialog at this stage, which is kind of more what you would associate with an action/adventure title, or a small RPG. It's certainly a lot for a shooter. I'd have to sit down and look at the specific allocation of that.
The thing that makes us really different isn't so much the quantity of dialog. It's the fact that so much of it - like fully 80% of it - is handled through AI dialog systems, as opposed to through scripted events. So when I'm encountering guys in the world who just happen to be there, and they're able to serve up relevant content to me, they're drawing on a large pool of content.
The actual content that's kind of like scripted, in the sense of, you walk into a room, and the door shuts, and certain things happen - that represents maybe 120 pages' worth of script dialog, in the kind of classical screenwriting sense. So I don't know that it's necessarily a lot more content. I just think it's content that's delivered differently. It's chopped into smaller and smaller pieces, and it's put together more like Lego blocks, than it is like a typical scene.
CR: Do you think that normal players will notice that there is something unique going on narratively if they don't play it twice? Or will they just think all the consequences that occur are intended?
PR: It's a valid question. I think they're going to start to notice it when they realize that events are happening to them involving characters that seem like they have a lot more direct connection with the events of the game, than maybe a bunch of other characters that they've also met that don't seem to show up in those situations. It's a tough call. Honestly, I'm not sure. I'm not sure.
It may turn out to be one of those things that's a lot more clear to people on hindsight, or by conferring with friends who are playing the game. Or by playing again. And I'm okay with that. I think at this stage, it's less about being really flashy than it is about trying to create an experience that is at least fun, and understandable, and meaningful.
And if it's not 100% clear... I mean, honestly, I think if they're not able to differentiate between our highly systemic building block approach, and something that's a lot more scripted, then that's actually the least of our problems. That's actually a good problem to have.
Honestly, I don't think anybody's going to mistake these highly systemic moments for, for example, a scene from Call of Duty 4. You know what I mean? I think we're not at that stage yet. People are going to look at it and go, "Okay, that feels a little... a little mechanical. That's probably something that's been put there because of something that I did." And I think that's still probably going to be a factor.
CR: I remember a few years ago, I was talking to Michael Land, who was one of the LucasArts composers during their heyday in the nineties, and he did - his soundtrack to Monkey Island 2 was amazing. It was really procedural. It was very interactive and very cool. I was a music major and I was so impressed by that.
LucasArts' Monkey Island 2: LeChuck's Revenge
And I complimented him on it, and he said it was a very rewarding thing to do, but the problem was, it was done well enough that players didn't notice. And then it became hard to justify that expense in a game, if people were like, "Oh, it just had a great soundtrack." Is that something that you worry about?
PR: Probably I'm just about the worst person to try to judge this because I've been so close to it for so many years, but I kind of, at this stage, I tend to view this sort of dynamically-assembled narrative as kind of... it's a little bit like if you're looking at editing in film. It's almost better if they don't notice it.
It's not a hundred percent true. I still want them to go, "Wow! The story's really good in this game." Obviously. I think, at the very least, especially if you differentiate between story and writing, if you think of writing as the data that's generated in order to support the story, I think Susan did a great job. The other writers that we brought in - Armand Constantine did a great job. Kevin Short did a great job.
But the thing is, people will react, I think, potentially, to the writing, to some of the characters, in terms of the dialogue and the voice acting. They're going to either like it or not like it. Some of them they won't like, some of them they'll like. And I think that some people will tend to conflate that part of it with the story.
People, when you say, "Yeah, but that's not what our story is. Our story is all of these systems that allow you to do this and this and this, and certain things to happen," they'll go, "Well, yeah, but that's just your game." Right? And that's okay. I don't have a problem with them not being able to disentangle those. But I think that is a... that's a tough call.