Valve's Writers And The Creative Process

[In a rare interview, Valve writing lynchpins Marc Laidlaw and Erik Wolpaw talk in-depth on the narrative structure and pithy quips behind critically feted games like Half-Life 2, Portal, and Left 4 Dead 2.]

Valve has developed a particularly strong reputation for story in its games, recently, and much of that credit can be laid at the feet of its in-house writing team. Whether it's the seamless character developmeent in Left 4 Dead, the engrossing dystopian sci-fi of Half-Life 2, the pithy quips of Team Fortress 2, or, of course, the wickedly demented GLADoS in Portal.

Marc Laidlaw has been with the company since before the release of the original Half-Life in 1998; he started his writing career with well-received cyberpunk and sci-fi novels but moved into games once he discovered the narrative possibilities.

On the other hand, Erik Wolpaw made his name in games journalism -- both for GameSpot and, perhaps even more famously, with the satirical review site Old Man Murray; he worked at Double Fine Productions on Psychonauts before finding a home at Valve.

Here, following a well-received GDC Austin lecture, the two discuss writing process, inspiration, and mindset; and some of the creative happenstances that lead to the memorable and beloved stories and characters in Half-Life 2 and Portal.

It was interesting to hear you talk about the creative process and how difficult it is. You joked around about not doing anything for six months, during the incubation period. Do you find that that's typical for you, in writing for games -- just sitting on it for a while?

Erik Wolpaw: Yeah. I'm kind of an inveterate procrastinator, but sometimes that procrastination panic can inspire good stuff. There's a certain process of grinding through something, but then there's also those flashes of inspiration. I honestly don't know what the process is. Some connection just gets made in your brain, and you hope it keeps happening and doesn't stop.

Then, another part of the process filtering through that is that you look at something or read something, and some connection gets made. You think, "Ooh, I can apply this to whatever I'm working on."

Part of that is setting up this filter in your brain that's like, "I'm working on this thing now." So, everything I read, everything I look at, and everything everybody says to me is put through this filter of the project I'm working on. "Is this useful?" Everything you're telling me right now is first going through a filter of, "Is there something in there that I can use on the project that I'm working on?" and then it passes through and I answer you.

Marc Laidlaw: I don't know if this is true for all [creators], but it's a writing thing. If I were working on a story, and I just have a bunch of pieces floating around in my head, everything that I'm experiencing, like walking down the street, or reading, is filtered through that. Usually, that's where the missing pieces come from.

So the difference in applying that -- putting in a writer and working on a game -- is that you've got a writer who is applying that filter to the game. I don't know if anyone else working on the game necessarily thinks in terms of that, although you do have people coming in and saying, "I saw this movie last night! We should do this scene in our game!"

EW: It's not so constant. When you're really in that mode, it takes a lot of energy to keep that filter up. It's there, and you don't really notice it, but you notice it when you're not thinking about it anymore. When I'm done thinking about this thing, it's a relief, and I can start thinking about the next thing.

And you guys are obviously part of a team as well, so that filter you're talking about will also be processing stuff that other people are talking about.

EW: Absolutely. When they're talking in a room, you'll just hear a snippet of conversation and be like, "Oh my God! Can we do this? This totally fits."

Can you think of any specific instances where a snippet snowballed?

EW: One instance where the connection was made was with Portal, where we had the boss battle designed. It involved you burning up those spheres. At some point, we were sitting around talking about it, and somebody said, "Well, we never trained anyone that you can burn things in these incinerators. We're just springing this on you right there."

Note: Pictured cube unsuitable for companionship. 

There was a connection that was made right there, because we had this part in the game where you had the level with the Companion Cube. At that point, at the end of that level, you just left the Companion Cube. You just abandoned it because it couldn't go any further.

There was this connection made, like, "Holy mackerel, this can be way stronger if they force you to kill the Companion Cube, and we're also going to train you that incinerators burn things in this world." That was this moment of insight that just came from design arguments that were happening, and all of a sudden we hit on this perfect thing.

I know it's difficult to describe a creative process, but how did the character GLaDOS spark?

EW: It was just sort of luck. When I was working on Psychonauts, we'd hook up this temp dialog. We'd just get people around the office to hook it up. One depressing thing I noticed is that a couple of times I'd run out of people to do, and I'd just use a voice-to-text thing.

And people were laughing at that way more than what the lines were worth. I realized, "No amount of writing is funnier than this text-to-speech thing reading it." I was always thinking about that, and was kind of bitter about it, and by then, I was like, "I'm going to leverage this and use it to my advantage."

So that's what started it. And it was easy, too. I'm not a fan of temp voice, but it's got a place in the planning process and you can't avoid it. So I was getting the best of both worlds. I was in control of temp recording. I didn't need anybody to read those lines, and the weird artifacts of text-to-speech was going to make that stuff seem even funnier. So that's where I started with that.

But we were still fishing around, thinking that this would be announcements coming over the facility's speakers, and there was still going to be something else, like the villain, or somebody to interact with. It quickly became apparent that people were reacting to this voice. It may be a trivial realization, but it was a big realization for us. "Whoa, this voice is just the facility that's angry at you and will confront you at the end."

And you just built it from there?

EW: Yeah.

It was interesting hearing the narrative analysis [at GDC Austin] and what Tom [Abernathy] was talking about up there. Was it interesting for you guys to hear people take a literary analysis of the story of the game? Does it surprise you with the things that people come up with? Are you like, "Well, that's not what we meant, but...?" How much of it is deliberate?

ML: People have fun connecting the dots, and the patterns that emerge might be different for different people. Some people have done really elaborate analysis of the meaning of these things, and it's dangerous as a creator to get caught up in that, because it ruins what people are going to get out of it later. We put these things together out of the things that are floating around in the environment. It's an intuitive process, and people receive them in an similar way.

There's intuitive stuff going on, but it's not like we had an agenda that we set out to make a dramatic statement. We never talk about theme. We just leave that to students to analyze. It's enjoyable to see what they come up with, but setting out to carve the meaning of your piece in stone and then building everything around that is not a very natural way to work.

So you're not thinking in metaphors most of the time.

ML: No. You want it to have resonance, and it's cool when you're doing something that you think people are going to recognize something -- because it's reminiscent of something going on in the world at the time. You look at a character like GLaDOS in terms of other science fiction AIs like HAL, and people are going to see this as a series of characters that are a convention in science fiction. But we're not thinking, "This is a statement that we're going to make."

EW: You try and write stuff that's truthful. When GLaDOS is talking to you, one of the rules I had is that she shouldn't talk to you like a computer. She shouldn't be all like, "Oh my nuts and bolts." She's got this computer voice and she is a computer, but she's talking to you like a regular person. In that sense, we've got a theme. It's not metaphor. We're just trying to make her sound like a person who is angry at you and is manipulative might sound.

Tell me a little more about writing in competitive multiplayer games like Team Fortress 2.

EW: There isn't really a lot of space to put anything, so you just try to be as brief as possible and then observe the game as much as possible.

ML: As far as somebody outside that but watching Chet [Faliszek] and Erik work on it, this is a case where the collaborative process worked really well. Chet would come in and say, "Look, we need to write 100 lines this afternoon." And then Chet and Erik would sit down with the spreadsheet, literally.

EW: That's the classic writer's room thing. You just sit around and try and crack each other up, and 80 percent of it is stuff that's so foul that it could never make it in the game. [laughs] But then you're left with that 20 percent that's decent. We all did that when we were sitting in the same room. We'd be like, "We've got to write all these lines."

It's way less fun to do stuff like that alone, then?

EW: Yeah. Stuff like that is the perfect collaborative effort. Left 4 Dead has some of that stuff in it as well. Just write short, because you don't have a lot of space to say those things.

Speaking of Left 4 Dead, that's another game where you get dropped into a situation and things just kind of unravel. With Left 4 Dead 2, are you taking a similar approach to how story is told? Do you think the games are better off the way Left 4 Dead was, or is it important to have more of a backstory?

EW: I'm sort of speaking for Chet here because he's the lead writer on it, but I've been helping with that lately. Splitting the difference on Left 4 Dead 2, it's not necessarily a more complex narrative. It's more of a coherent narrative from Campaign One to Campaign Five. They each connect in the characters and situations. There's an art to the environmental story that's being told about the progression of the infection, from point A to point B.

The thing with Left 4 Dead is that it's replayable. As much as I'm against stopping the game to give you any exposition, I really think it would be a bad idea in Left 4 Dead. It's designed for people to play it 20, 30, and 40 times.

So we try to put enough lines in there so that you'll have an opportunity to hear new things, mainly because lines only play in certain situations so that as you're playing, you won't get totally sick of hearing that information.

Can you speak as to why you decided to have an overarching story?

EW: Only because we thought of a way that was lightweight and didn't seem like it would impact the game that much to do it, and it seemed like it was an interesting thing to pursue. In talking to people and listening to playtesters, it was something that, while not the most important thing...

ML: It really helps to have an organized theme or setting when coming up with levels which drive gameplay. You have to get really specific. You can't talk in vague generalities like, "It would be fun to have some combat to some sections of a place." You have to have resources. You have to have a list to say, "We need to build a city street. We need signs to hang here. We need to know what the signs say." It becomes very specific details. So if you have a good overall story or theme, that just drives all that.

Like, "Okay, we're doing this in the South." Suddenly we have this environment, this one, and this one, and you've got something to build from. If you don't ever come to some kind of focus, story is a good way to provide focus that's really low-cost. If you have a good idea for a story, nobody has to build anything, but they can build things out of that idea.

EW: And having those constraints early on, because the general story arc was determined right after Left 4 Dead, it helped to produce a full game in a year, because there wasn't a lot of time wasted dressing out sets in ways that we would later change.

It gave everyone a lot of focus on what these sets should look like, and what sort of information we should convey in the environments. Because writing is a lot of work, but building everything in those games has to be manufactured and placed by hand. It takes a lot of time if you go down the wrong road.

Can you talk a little bit more about how the whole train thing came about in the intro of Half-Life?

ML: It was a code thing. One of the level designers was constantly asking for more features they could use in the level editor, and trains are sort of a general-use thing. Every time a platform moves in Quake, that was a type of train. Usually they just go back and forth, and they didn't move very smoothly. So one of the programmers on Half-Life created a track train that would move over long distances and turn and pivot on certain points and go around curves and certain things like that.

I don't remember what the original request was for what the original level designer had in mind, but I remember when I heard about it, the first thing that came to my mind was an actual train. I just took it too literally, and I started thinking in terms of, "What could we do with a train in Half-Life?"

Then we came up with an idea like an old carnival ride, where you would literally be on a train. And to help set the pattern for the whole game -- the whole game is sort of this linear ride experience -- we start you off, literally, on a train.

And then the rest of the game is basically that haunted house feel, where around every corner, something jumps out at you. So it went from being a request for a technical feature that I sort of misunderstood what it meant, but it suggested something narrative, and we all started to see potential in there.

Without having the writers on board with the process the whole time, do you think something like that would've happened?

ML: Oh no, definitely not. I think in the case of a writer coming into Half-Life really late, they'd have a game that was very much built already, and would just be there to come up with lines of dialog for the security guards and scientists to say.

If I looked at what the game looked like before I arrived, it was sort of a series of unconnected lab rooms. They were building this huge, sprawling base, and you'd just fight your way through there, Doom-style. Every room would spawn a bunch of monsters, and you'd shoot them. And you'd go to the next room and fight some more monsters and go to the next room.

At some point, that's what we were building. That was the origin of us looking at it and saying, "No, we've been telling people we were doing something different, but this looks like everything else. Back to the drawing board."

At that point, we tried to deliberately integrate the story and level design in a really concrete way. And everybody wanted to do this. It wasn't my idea to hire me as a writer. The people at Valve wanted this kind of narrative experimentation to go on in this game. I was just there to try and help enable what everybody wanted.

A lot of gamers are infatuated with Alyx as a character. Can you tell me more about her development?

ML: With Half-Life 1, we didn't have a great animation system. Characters didn't have fingers. They had these potato-mitten things for hands. It was a huge breakthrough for us to move the mouths of the players when the speech was coming through. So we didn't try to get too ambitious. I thought we were pretty ambitious with what we were doing with characters already.

People responded really well to having Barney the security guard along, just because he had infinite ammo and they felt bad if they walked him out in front of a turret and he got killed. We realized, "Wow, people are getting attached to a cipher." Clearly, people wanted to glom onto a buddy and project on to that, and they love having this experience.

So we set out to create some characters that would take advantage of that. We realized we were going to have a better animation system, facial expressions, and the detail in the modeling was going to get better. And our animation was just going to be a lot richer. That was the point where we started to think in terms of the actual characters.

So we started to create an interesting cast of characters that was connected to the original game, in terms of all being either Black Mesa scientists or their relatives. The events of the first game still have meaning for the characters in Half-Life 2, even though it's many years later. Alyx wasn't originally associated with Black Mesa at all. Her father was a soldier out in the wasteland, and she was a courier kid who ran messages between different resistance groups. Then there was a separate character who was an inventor scientist guy from Black Mesa -- Eli -- who lived in a junkyard.

At some point, we started condensing the game. Whole middle sections went away, and I kind of jokingly said to people, "I think this Alyx character over here is the daughter of the inventor character over here."

I thought about this for a week or so, and then said, "Okay, we'll eliminate this other character who was originally her father, because we got rid of this whole section of the game."

And then it all fit. Suddenly, she had this direct connection to Black Mesa. She had this science background going on, and more tech, in her case.

We had done different background stories for her, and we kept revising those, and again, it was a matter of the artists coming up with a look for her and finding a model who had an interesting look, and then animators wanting to try some new expressive stuff with her, since we had facial expressions.

We had much better lip syncing. Then we started hunting around for an actor, and found the perfect actor to play her part. All of these things caused her character to gel.

Then there was a lot of playtesting to get to the point where she was a welcome character instead of somebody who did so much handholding. This is less of an issue in Half-Life 2, because she wasn't on all the time. She was also a really good foil for the character, because Gordon is not actually accomplishing his own goals.

He's accomplishing his co-op goals with her. He's helping Alyx do things that she is concerned about. So we were trying to do a thing where it wasn't about one guy who is saving the world heroically for his own reasons. You're doing it with your allies and friends. It's part of a larger effort. It's not just Gordon Freeman against the universe. It's Gordon Freeman's part of this group.

So Alyx was a great voice for the character in some way, and you're rescuing her father. You're not rescuing Alyx. You're doing things that are valuable for her, and she's a stand-in for this new world's struggle. She knows what is important for you to do, and she was useful.

She gives back to you emotion, which is the only way we can tell the internal story of Gordon Freeman -- by the way the other characters treat him. So having the characters like you and glad to see you, you think, "Oh, I'm an important person in this world." Alyx was a great way of affirming that, and the things that are perilous to Alyx are going to be things that you care about.

So it's not the typical quest. Saving this girl's father is a lot different from saving the princess. This is something she cares about and you want to do because it's important to her. Otherwise, a player is just playing a game and doesn't have an emotional goal outside of that. She's a good carrier for the actual emotion in the game.

In general, what do you guys think of the state of video games writing? Are we going in the right direction? Are we as far as you guys would expect?

EW: I don't know how far I'd expect us to be, but I think it's moving. I've definitely played a lot of games now where it's not just at the base level. It seems to be rising. I'm not embarrassed to be playing lots of stuff. There's just a ton of good and interesting stuff. And stuff that's to my taste, too.

I thought that both The House of the Dead: Overkill and MadWorld explored these weird areas in a legitimately interesting and funny way. Overkill was hilarious, but deliberately. It wasn't Resident Evil 1. It was deliberate, and it worked. And there's a ton of stuff. Maybe I'm just picking and playing less games than when I reviewed games, but most of the ones that I've played, the story or the writing or the acting has impressed me.

ML: The level of translation is also... I've been playing Monster Hunter lately. Even there, they obviously put effort into the little training tips. They're always entertaining to read. That right there is a huge difference.

EW: I played Plants vs. Zombies recently, and one of the biggest things that kept me playing was seeing the new descriptions of the zombies. Whoever wrote those did an incredible job. They're hilarious and short.

ML: I read every piece of litter on the pier in the first Penny Arcade game. The interesting thing is that games have attracted better writers from outside, but also the writing of the people who grew up in games, who would otherwise be writing in other media, now writing for games, and that's all that they do, and they're really good at it.

EW: They're good writers who have an intuitive sense of what works in games. I think that's a factor in having grown up comfortable in games. It's only getting better.

ML: That's definitely a really good trend. And the games that are really interesting are the ones where they're making things that are only just now possible. I think Half-Life happened at the earliest that it could have happened, in terms of the story we were trying to do then. Every few years, it seems like there's a new level of stuff that's possible, and we'll start seeing games take advantage of that.

It lags a little behind technology in some ways, but incremental improvements in animation and character...I think the things you see in Left 4 Dead, with people speaking on the fly... the way the characters are created in Left 4 Dead, it's impressive to me. We're creating these characters in a casual way. We're not making a big deal out of it, like, "Oh look! We've got a speaking companion character who is going to follow you around!" You just take it for granted now.

It's more like, "Who are these people? Who are these characters?" It's more about that than, "Here's a character talking to me in a game." They also leapfrogged the uncanny valley. Some things are still stuck there, where they're trying to be real and so it's creepy. But you can be comfortable with more stylized characters, and they feel more real than a real character. I'm really excited to see that stuff happen, where it's taken for granted, in a way.