Do you ever have time to do level design these days?

CB: No, I haven't done that in a good long while. I'm more of a facilitator at this point. It gets to the point where the game is up and running, and I just get to play it and make notes -- it's what we call "shoulder surfing."

Lee Perry, our senior gameplay designer, does a ton of it as well. [He'll] just go into a person's office and ask, "What are you working on?" -- looking at it, and saying, "Can you move that cover over by five feet?" or, "That guy really annoys me when he spawns. Can you spawn him from a different angle?" or, "People are getting kind of lost here, can you put a little ammo at the end of the hallway to kind of lead them through?"

All those little things make the difference between good to great, right? And, you know, it's also just saying, "I have to see a gun with a chainsaw on it," and knowing when to dig in your heels.

There's a lot of mutual respect amongst the leads in regards to when people pick their battles. I'm not going to sit there and dig in my heels and say, "Oh, I have to see a character with a bozo hat on or clown shoes."

I'm not going to sit there and be stubborn about something that I'm not a believer in. I know if Lee says, "I have to see a Boomer with a shield and an explosive flail," I'll say, "Dude, that sounds cool. Do that."

What do you think makes a good game designer? Is it more of an instinctual thing?

CB: It depends on the game, it depends on the designer. For the kind of games we do, I think I'm fairly effective at what I do because I know when to pick my battles. I know when to push for the things I want to see, but I also know when to back off and give people enough rope to be creative, not sitting there and saying, "Here's the exact blueprint for how your level's going to pan out, and don't deviate from it."

You have to trust people enough. Look at the design process for the tickers, where we knew we had this level where the truck had to make it through and we wanted to have obstacles for the player to clear out before the truck would advance.

The initial idea was that the Locusts just had straight bombs or land mines that the player had to clear out. I said, "That's not that interesting." So Dave Nash and I sat down and were like, "What can we do to make this cool, man?" Well, the locusts usually use creatures. Well, what kind of creature would it be? What if it's this kind of small thing that scurries around and has a bomb on its back?

So Matt Tonks went and did the AI for it, and he made it more cockroach-like, so it avoids you until the last minute when it charges. I thought it would be cool if you melee this thing and you could send it flying on its back, and turn it into basically a big portable exploding barrel. So then it got built, and we saw the big beaver teeth that were on it -- okay, those teeth need to chatter.

So all of sudden there's this iterative process, and we wind up with a very cool memorable creature that you're then fighting in a dark tunnel. Suddenly the fun and the magic happens, and you kind of let things happen organically.

You need to allow for time for iteration. Thankfully, our tools also facilitate that with things like Kismet and UnrealScript.

Even with that iteration, though, these games didn't take as long as some modern triple-A games to develop. By the time you actually get to the point where you're modeling everything, do you feel you've basically figured it out? Do you do a lot of redevelopment? I've talked to Valve about this, and they seem to be happy to toss essentially-completed stuff away.

CB: I think it completely depends on the shop and what their development methodology does. If Valve is happy to throw out a bunch of completed stuff, cool. Valve makes a lot of amazing stuff, and if that works for them, more power to them.

We do a lot with levels in shell form, or when we're prototyping. It costs money to have an artist sit there for months and build a character that's eight trillion polygons, right? So the more we can answer that question before we commit to something, the better. I think we've done a pretty good job of it.

One of the reasons we're able to develop this game in two years is because we have tools that allow you to rapidly prototype. A level designer can have a level shell up and going in a matter of days, if not hours. We rough in all the levels before one artist lifts a finger. We have designers who are able to use our Kismet tools to cobble together a prototype version of the Bloodmount or the Mauler Locust before anyone even builds the thing.

We have these maps called POC -- proof of concepts. Lee would sit there with Kismet. He would cobble together a Boomer with an epic flail, and create what the gameplay mechanic would be. When he shoots, he deploys, and when he gets close to a certain point, then he charges, and all that kind of stuff. Then you show that to a programmer, and who codes it up, and then the concept gets done.

There's very much a pipeline that way. That's the way we work. I think a lot of it is building a better Swiss Army knife, in many ways.