Following a half-hour play session of Left 4 Dead that ended entirely too soon for my tastes, I sat down with Valve Software's Erik Johnson for a friendly chat. There, in a secluded portion of Valve's E3 2008 booth, Erik let us in on the process behind the upcoming survival-horror-FPS, and how Valve was able to build a better zombie based on their experiences with Team Fortress 2. Erik also let us in on a few upcoming tidbits regarding the future of TF2, including hints as to upcoming animated shorts and planned content. TF2 fans, you won't want to miss this one.
GameCyte: In relation to Team Fortress 2, when did the development process begin on Left 4 Dead? You can see a lot of influence [of the former on the latter].
Erik Johnson: TF2, we started, I think, about 27 years ago. *laughs* In its current form, though, it's been just a couple of years. It started with Mike Booth, down at Turtle Rock, building a concept of a product that was strictly a zombie horde, and fighting your way through... We played that during Episode 1's development, so, some time in 2005. Just from that early prototype, you could tell -- especially from the way Mike walked us through how the director was doing things to craft the experience for the players in the game -- it was a game that we wanted to flesh out, and build.
We were building TF2 during the same time, so the crossover between the products is significant enough that, where it made sense -- things like the voice cues when players take context from the world and distribute that data to the rest of the team, which in a co-op game is even more important than in a chaotic game like TF2... I would say that was probably one of the single biggest crossover points. There are some technical/graphics pieces that get shared just because the Source engine is moving forward all the time... some lighting improvements, and things like that. A lot of the work that we did to get TF2 working on the Xbox carried over to get L4D working on the Xbox -- things like that.
GC: You have this one L4D setpiece that you're walking us through here today, where it starts on the rooftop, and goes past the gas station and the warehouses; how many maps can we expect to see L4D ship with? How many "missions?"
EJ: There's four total campaigns in the game, and each of those is about five pretty large levels. You played through them; they're a lot larger than your standard TF2, or one of our single-player, levels.
GC: Can we expect to see the same kind of new content, like you've been rolling out with TF2? That same rapid pace?
EJ: Yeah. That's been our goal from the first game we ever shipped -- definitely the first multiplayer game, Team Fortress Classic. We want to keep pushing out content to our customers and help build the user base.
GC: I noticed, between levels, that each player was getting a performance rating. Is that going to directly affect how difficult the next section is, à la the "director" portion?
EJ: No, the director's a little more granular than that. It's looking at a bunch of factors in every individual player: When was the last time they were vomited on? What kind of damage have they taken? How many bullets have they used? It's not just a "skill difference" thing, either; in the same way that we try to craft the experience in our single-player games, the director is the thing that we're using to craft the experience in the multiplayer space. That's the bet we're making. The screens you're seeing in between, those are mostly used as a way to communicate with the player about the state of the game, or how they did -- and also, how their teammates did. People can build strategies in the game around the data that they know about the people they're playing with.
GC: You can see some of the TF2 influence there. Like how every time you die in TF2; "Hey! This round, you killed this many."
EJ: Yes; especially for new players, it's a tricky problem to let them know what they're doing well at. For new players, especially in the multiplayer space, it's tough. You're playing against people that have potentially been playing for years longer than you. You want to use something to teach them the system and how to move forward in the game. In a co-op game, that's obviously really important.
GC: A purely mechanical question about L4D: Let's say four of us are playing, and one of the four goes down. Can you always revive someone, or if they get torn to pieces, are they just completely dead?
EJ: There's a couple of different things that can happen. A player that's on the ground, incapacitated: you can revive them, but eventually they will kind of bleed out. They'll come back at the next checkpoint, or they may come back inside the level if they'd potentially have to wait too long. So, you can be rescued inside of a rescue door.
GC: So there is room to drop back in.
EJ: Oh, yeah. We don't want people to just sit for a long, long time.
GC: I have some more TF2-specific questions, for the fans, if that's all right. While you've done a great job making a fun and even game, are there any glaring imbalances in TF2's classes that still jump out at you?
EJ: You know, when we talk about class balance, we still feel like the Demoman might be a little out of whack. It's probably the one that comes most often into play. And it's tricky, because there's a skill component to the sticky grenades that's interesting. Just making things less effective -- tweaking numbers, like, "this now does less damage" -- isn't really the approach we generally like to take with things like that. That's one we look at probably the most.
There's also imbalances the other way. Take sentry guns, for example, for the Engineer... take the whole build layout, where he has a dispenser, teleporter, and sentry gun. Sentry guns, for newer players, at level 3, are really difficult for them to take care of. For experienced players, it's really trivial. There's no in-between that's an interesting play for an Engineer. Dispensers and teleporters feel like they're a little bit less useful than they could be; that's something that we keep an eye on.
We felt like we took a stab at the Soldier's imbalance in one of the most recent updates, where he takes a little more damage from his own rockets -- that seems to be effective. We're always looking at the system and trying to figure it out.
GC: How do you feel the unlockable packs have worked, and been received, so far?
EJ: It's been positive. With the Medic, the first pack, our feedback was definitely that some of the achievements that we chose -- the difficulty in obtaining them was too high. So, we rolled that feedback into the Pyro pack. We actually went back to the Medic and changed the amount of achievements you need to have to get the unlocked weapons. As opposed to having to do unlocks for all the classes at once, that's one of the advantages we have in doing one step, is we can learn. So, with the Pyro, I think we did a much better job in making those interesting for players than we did with the Medic.
The Pyro class, also, was one that was much less useful before the update, and not only more useful, but a lot more fun, since the update. We're pretty happy with that release. We feel like we're getting smarter about them, and that was our goal.
GC: Has the "unlockable pack" idea been there since the (current) game was conceived, or was that something you realized after release?
EJ: It was after release. We knew a couple of things: We knew that we wanted to add more achievements, and we knew that when we shipped the game, we just didn't have time to implement all the achievements we wanted to, so we knew we were going to build some more of those.
Secondly, we knew that we wanted the game to change over time, because we look at a lot of statistics and we play a lot of TF2, and we learned from Counter-Strike over the years that it's fun to have the game evolve over time, especially based on engagement with your customers. We look at how people are playing, and we try to have our customers get more value, both in new content and in changes to the game rules. The idea came about with those two things in mind. We had a tool, and achievements, and a goal.
GC: Do you continue to plan to do the class unlockable packs individually, or do you think, in the future, you might do multiple classes at once? Would you try and avoid the period, after the release, where everyone plays the same class?
EJ: We might do multiple classes at once. We definitely talk about that a lot -- a world where half the players are playing Sniper is not a world we really want to have for any amount of time.
GC: *laughs* All right. When you unlock a new weapon, in an unlockable pack, it's given a "level" when it lists on the loadout screen. Any hint as to what that might actually mean?
EJ: No.
GC: No? Not a word?
EJ: No. No hints.
GC: Not a single one?
EJ: Well, to walk back from the system, there's a set of criteria that's applied to every one of the weapons based on what its abilities are. So, in the Medic's case, the Ubersaw doesn't crit, and it fires slower, but it gets you charge -- all of that is combined to form a "level" for a particular item. It's something that gives us flexibility down the road.
GC: How large is the TF2 team right now?
EJ: It's tricky for us... we have our technology base, and Source, and there's a number of people that work on that. They're working on L4D also, and they're working on all of our products. It's hard to count them. I would say the group of people that's spending most of their time working on TF2 is about 15 folks.
GC: But there's a lot of company-wide crossover?
EJ: Oh yeah.
GC: Does the Valve internal team do most of the pre-release testing, or do you guys farm out some of that?
EJ: We do most of the testing ourselves. We do bring in fans of the game and people "off the street" so we can watch them play, so we can learn and make sure we're communicating things -- not so much bug fixing. With Gold Rush, we spent a lot of time bringing people in and watching them play, making sure they understood how the HUD works, how to push the train, and things like that.
GC: Looking down the road, what is the date you have in mind for releasing the rest of the unlockable class packs -- the first round for all nine classes -- and the rest of the "Meet The Team" videos?
EJ: To try and do a class unlock... I think it was about five weeks between the Medic and the Pyro, and we'd like to keep that schedule. If you extrapolate out from there, with seven classes left, that's 7-8 months of class work for sure. The class unlocks are a little faster to come out than the movies. For the next release, the Heavy, we're going to have a short piece of linear media that isn't a "Meet The" piece of media, but we're going to have something out.
GC: So you have plans to continue this TF2 "cartoon" media?
EJ: Yeah. We think the movies have a useful role in our release cycle. One of the things we did in the last one, for the Sniper, we showed the Pyro's flare gun. We knew people would figure it out, and see it, but we were kind of surprised -- and we shouldn't have been -- we were surprised at the reception we got. In doing things like that, we're allowing our user base to be more invested in the product, and share this funny little thing that we did. It's something we want to do more, so the Heavy release isn't going to have a full "Meet The" video; we're going to do something that's much shorter but kind of a nod to that community.
GC: That leads me into the major question I have about TF2. Fans everywhere appreciate all the free updates done to TF2, between the class packs, and the shorts, new maps, new game modes, but all of this game content is being given away for free. How does the game sustain itself?
EJ: Well, we did this with Counter-Strike. We're not doing it strictly to be altruistic; this isn't charity work. We sell more copies of the game, the happier we make our customers, is our approach to it. We view customers as incredibly valuable. We want to make sure that a customer of one of our products buys our next product. We look at our relationship with our customers as a very long-term thing, not a product-by-product thing. Continuing to give them content, and continuing to give them a reason to tell their friends to buy the product, is something we try to drive. It's very profitable, it's something we're happy to keep doing. It's very fun work to do, as opposed to a three-year death march on a big huge product.
GC: And every time you've done a free weekend or a large update, you've seen a significant sales boost?
EJ: Yeah, absolutely. The community's growing.
GC: Looking beyond the horizon -- what happens when the game reaches critical mass, audience-wise? Obviously these great updates will keep the current audience playing, but you're not selling a significant amount of new product. What happens then?
EJ: We still sell a lot of Counter-Strike. Both CS 1.6 and CS Source. And we never really saw diminishing returns in investing in those products. We run into bandwidth issues, where we just don't have enough people to continually build content for those, but we never really saw a point at which -- to use my own term -- our customers were so happy that they were done being our sales force for our products. It's not something we really think about; it's a good investment.
I know it sounds simple, but making your customers happy is actually a really good business to be in.
GC: Is the plan to follow this same pattern of months of tweaks, patches, and free content with L4D? With the same kind of release schedule?
EJ: We've tried to take TF2 and boil it down to a formula, where we're doing a movie and a class -- and some sort of promotion, like a free weekend -- almost every month. They may not take that exact same formula, but we'll definitely be taking the same approach and try to deliver value to our customers every time.
GC: TF2 is already out, and L4D is coming out, for the 360. What do you think were the major positives and negatives of trying to develop for the console platform, as opposed to the self-published PC version?
EJ: Well, EA handled most of the PS3, so I don't have a lot of insight into that -- they did most of the work. On the 360 side, certification is definitely difficult, not having Steam is difficult, our lead times are much longer. Not being able to deliver the same amount of content to all our customers is something that's tricky for us. We ship almost an update a week for TF2 -- not major class updates, but there's some kind of software that flows out of Valve to our customers every week for TF2, and we just can't do that on the Xbox, and that's pretty frustrating to us.
GC: Any positives to come out of the experience?
EJ: Oh, sure. We did Half-Life 2 for the [original] Xbox... I don't think our timing was that great for when we shipped it, so in a lot of ways, this was our first big console release internally. We learned a bunch of stuff. We built the groundwork, in Source, for being able to make L4D work on the 360 a lot more easily than all of the Orange Box products. There's a lot of technology that was valuable there.
EJ: Absolutely. It's an audience that we weren't reaching... I think there's a lot of crossover, but there's definitely customers that had never played one of our products before because they didn't play games on the PC, and it was nice to reach those people. It was a positive experience overall, absolutely.
GameCyte would like to thank Erik Johnson and Valve for taking the time to speak with us, and for being so forthcoming and accomodating with regards to our TF2 pestering. Left 4 Dead will be available on PC and Xbox 360 on November 4th. Team Fortress 2 is currently available on PC, 360, and PS3, now featuring wild "weapon level" speculations.
My friend Bob Pantolamo released his indie game "Maximum Leptodactylus" last year, and he's following a similar release strategy. He's releasing new frogs from the Leptodactylidae family every week, and he says it's really not that hard once you've got the first version out there. He said he gets about a 70% increase in downloads every release, although when he shipped two new Eleutherodactylus in one week, he only got about 30%. Guess people prefer their frogs without Batesian mimicry.
We were building TF2 during the same time, so the crossover between the products is significant enough that, where it made sense -- things like the voice cues when players take context from the world and distribute that data to the rest of the team, which in a co-op game is even more important than in a chaotic game like TF2... I would say that was probably one of the single biggest crossover points. There are some technical/graphics pieces that get shared just because the Source engine is moving forward all the time... some lighting improvements, and things like that. A lot of the work that we did to get TF2 working on the Xbox carried over to get L4D working on the Xbox -- things like that.
GC: A purely mechanical question about L4D: Let's say four of us are playing, and one of the four goes down. Can you always revive someone, or if they get torn to pieces, are they just completely dead?
There's also imbalances the other way. Take sentry guns, for example, for the Engineer... take the whole build layout, where he has a dispenser, teleporter, and sentry gun. Sentry guns, for newer players, at level 3, are really difficult for them to take care of. For experienced players, it's really trivial. There's no in-between that's an interesting play for an Engineer. Dispensers and teleporters feel like they're a little bit less useful than they could be; that's something that we keep an eye on.
EJ: No.
EJ: To try and do a class unlock... I think it was about five weeks between the Medic and the Pyro, and we'd like to keep that schedule. If you extrapolate out from there, with seven classes left, that's 7-8 months of class work for sure. The class unlocks are a little faster to come out than the movies. For the next release, the Heavy, we're going to have a short piece of linear media that isn't a "Meet The" piece of media, but we're going to have something out.
GC: And every time you've done a free weekend or a large update, you've seen a significant sales boost?
July 16th, 2008 at 4:01 pm
My friend Bob Pantolamo released his indie game "Maximum Leptodactylus" last year, and he's following a similar release strategy. He's releasing new frogs from the Leptodactylidae family every week, and he says it's really not that hard once you've got the first version out there. He said he gets about a 70% increase in downloads every release, although when he shipped two new Eleutherodactylus in one week, he only got about 30%. Guess people prefer their frogs without Batesian mimicry.