Exclusive Interview With Robin Walker
A few weeks ago we started a conversation with co-creator and software developer on VALVe’s Team Fortress 2, Robin Walker. Robin kindly agreed to answer a few of our questions. Our focus had been on the community elements of TF2 including mods, mapping and the future of the game itself. Robin has given fresh new insight into the history, current and future development of this great First Person team-based Shooter.
Kritzkast: TF2 went through many transitions in the conversion process from TFC to its current form, including a variety of near-realistic styles. Given the short-term meteoric success of titles such as Modern Warfare 2 and Bad Company 2, are you still happy with the period styling that you chose?
Robin Walker: This is a very interesting question, and one that we discussed internally a lot throughout the development of TF2 (obviously, MW2 & BC2 hadn’t shipped then, but Counter-Strike’s success alone was enough to generate the discussion). Whenever we started talking about it, though, we’d quickly run into the list of interesting gameplay elements that’d become more difficult to explain (rocket jumping, double jumps, the medi-gun, etc). Ultimately, TF2’s art direction is the right fit for its gameplay, and we think that’s still the right decision.
Kritzkast: There were to be a lot of things as spin offs for the TF2 franchise that either haven’t yet materialised or have but not in the form which we expected; cartoons became web comics as part of an update, Saxton Hale comics are just front cover concept art. Has this been a design or marketing issue or are you simply waiting for the right time to release “TF2 The Movie”?
Robin Walker: One of the strengths of the “entertainment as a service” approach that we’ve taken with TF2 is that it allows us to test things out before we heavily invest in them. So we’ve been able to try out a wide variety of things, and observe the community response. In some cases, like the Saxton Hale comics, we really weren’t planning on making the comics at all, and while the response to them was great, it didn’t make us think we should stop making games and get into the comic business. In other cases, the response made us keep working on them. Saxton Hale himself is a good example of this. He started with a small part to play, and the response to him was so positive that we’ve kept expanding him with every update since.
Kritzkast: As a team based game, TF2 is dependent on players wanting to come back frequently and play for the game to continue to attract new players. Does the recent foray into bots on koth maps mean that you’re seeing dwindling numbers? Is this another sandbox test for a different game using the Source Engine? Do you expect to be able to realise bots for other maps too?
Robin Walker: Like most things we do on TF2, there are several reasons why working on Bots was something that made sense. Several months ago we started digging into the data we had that showed a significant number of players who played TF2 only once. We wanted to know why they’d quit, and what we could have done to help them. The end result of that digging was the finding that the majority of them quit because they didn’t know what they were doing, and there was no “safe” place for them to learn the game at their own pace. The number one requested solution was some kind of offline training mode. That that got us thinking about Bots, because they’re part of that solution. We’d also been kicking around various other gameplay ideas we had for future titles, and Bot technology in TF2 could teach us some things that’d be useful. Finally, the technology behind the TF2 bots is the next evolution of the technology behind Counter-Strike’s Bots, and Left 4 Dead’s AI, and we’re always interested in driving that kind of AI further forward, because it enables new gameplay experiences.
Kritzkast: The official TF2 blog is still one of the most talked about areas of the community that surrounds TF2. I have visions of Saxton Hale standing astride the mouth of the giant Cave-o-Email each month. Whip by his side, shouting directions to his minions not to return without fresh material. How do you decide what makes it in and when to publish?
Robin Walker: The blog is written by the members of the TF2 team, so it tends to get updated rather haphazardly, as most of you have noticed. We try to update it regularly, but we’re always working on the next update, and gaps in posts generally coincide with some large chunk of work we’re all banging away on. In general, our philosophy is that while some words from us might be nice, fans are a lot happier when we ship something.
Kritzkast: Even within our own ranks the question of the gender of the Pyro is still argued over. Will we ever know? Chemical Ali, the artist for whom the styling of the scout’s mother was attributed, has created alternative female characters for each of the classes. Will valve be making these official and including them in the game as choices for players?
Robin Walker: We’re still working on Meet the Pyro, which will let you see the world through the Pyro’s eyes. We’ve seen Chemical Ali’s great models, but even with that work done there’s still a significant amount of work to do, and some design issues that we haven’t found good solutions for. Having multiple character models for a single class would have performance impacts due to increased memory requirements, production impacts in that new items have to fit onto both models, and content issues in that we’d need another set of voice acting. Finally, you have the issue of hit detection. If the two models don’t have identical hitboxes, there’ll be a competitive edge to picking one of them over the other. If they have identical hitboxes, you’re probably going to have issues in how well the visuals match the hitboxes in one of the models. In short, multiple models for player classes is something we’d love to do, but it’s not likely to happen in the near term.
Kritzkast: For some classes their roles have changed so much from their point of origin. The scout for example, has gone from a peck and run class to an over powered front runner. Will the character models be changed to reflect their new roles?
Robin Walker: Actually, we always considered the TF2 Scout as a flanking, high damage output character, such that we sometimes even refer to him as the assassin, a Spy who can’t disguise. We felt the Force-A-Nature pushed him even further into that role, having an even higher damage spike, and a lower ranged harassment capability. The Team Fortress Classic Scout was a different creature, being much more of a runner than a fighter. When we designed the TF2 Scout’s character, it was this evolution into a fighter that contributed to him having such an aggressive personality. In retrospect, we probably should have taken the opportunity to name him something other than “scout” at the same time.
Kritzkast: We have all had epic moments when we’ve tried for and pulled off the impossible move only to realise that we alone saw it and no-one will ever believe it really happened. Back in July 2009 you were recorded as talking of a new demo system. Are you any closer to releasing that into the wild?
Robin Walker: We’re closer, but unfortunately still not ready to release it. The primary function of it is to ensure that after that impossible move happened, you can say “I want to save that!”, and have the game not only save it for you, but also provide you with an easy method of getting it onto your favorite movie sharing site
Kritzkast:May 12th 2010, saw the introduction of the Steam Client on OSX. How soon will it be till you see TF2 being available on Macs? Up till this point you’ve been developing updates on just one platform. Will the introduction of a second (and possibly third) platform(s) make update releases and bug fixes far more sporadic?
Robin Walker: TF2 should be available for Mac in the next month. The Mac team was very focused on trying to reduce the cost of multi-platform support, and have done their best to automate as much of the process for us. As a result, we’re hoping that it won’t have a significant impact on the rate at which we release updates.
Kritzkast: Our conversation with Drunken_f00l revealed that valve has been thinking of integrating items management into an iPhone app, either under their own label or with the assistance of a third party. Should we expect TF2 innovations to be available on the iPhone/iPad some time in the future?
Robin Walker: We’re not working on anything right now, but we think both of those platforms are interesting.
Kritzkast: Do third party maps, mods and game modes need to be redeveloped for the Mac OSX? What assistance are you giving to the fan community to help them get to grips with working with both DirectX and OpenGL simultaneously?
Robin Walker: TF2 content is platform independent, so maps and models will just work on Mac, so they should just focus their time on making a fun map / mod / etc.
Kritzkast: We’ve seen PropHunt, Dodgeball and various other brilliant community created mods emmerge as add-on server modes for TF2. Is it your intention to merge these concepts into the final product, exposing them to all TF2 players, or are you simply happy to allow them to exist on their own terms?
Robin Walker: It’s something we think about on a case by case basis. As mods become more popular, we generally start by adding some functionality to solves specific problems the mod authors are having. If it continues to expand in popularity, and fits well enough into a TF2 customer’s expectation of what they might see in front of them when they join a server, we start thinking about building it directly into the game.
Kritzkast: The standard map base for TF2 has expanded rapidly largely as a result of community maps. Under what conditions do you attempt to take ownership of a map and has a map maker ever turned you down, are they even allowed to?
Robin Walker: We try to include a community map or two in every major TF2 update. We playtest a bunch of community maps internally, and we try to pay attention to what maps the community seems to be having fun with. Once we select a map we contact the author and see if they’re interested. So far no-one’s turned us down, but they’re absolutely within their rights to do so.
Kritzkast: Mentioned on the blog was one of the ideas for the Engineer update that was tested and removed. What other design ideas have you tried, tested and rejected in the course of your search for engineer update nirvana?
Robin Walker: One of the other things we tried was a secondary weapon that instantly teleported the Engineer to his teleport exit. So Engineers could leave their sentrygun for a bit to skirmish or collect metal, and be able to immediately teleport back to the sentrygun if something bad happened. It did work at achieving those goals, but we didn’t like the side effects. Teleporters stopped being much of a team focused tool, with Engineers placing them in places that made sense for their personal use, and not necessarily for the team. We also felt it was too easy for Engineers, almost eliminating the risk inside the decision of whether they should leave their nest to grab some resources. Both of these were solvable issues, but while testing this we found another idea that played much better, attacked the same problem of Engineers being rewarded for moving out from their nest occasionally, and had lots of other interesting applications.
Kritzkast: There have been several fake engineer updates by the community, ranging from guard dogs to chicken guns and ammo magnets. Many of these have been very detailed and well thought out. How much notice do you take of the fakes and what’s the chance of one of these ideas seeing its way into the final build?
Robin Walker: We love these. On the game team, we love seeing interesting ideas with more thought put into them than just text. Our web team loves seeing the way these pages have used the style of our TF2 update pages to do neat stuff, often challenging us to do better ourselves (see
http://www.engineer.fragfestservers.com/). Ultimately, good ideas make their way into the product by passing an analysis of what problems they solve, what benefits and disadvantages they have, the amount it costs to build and ship, and so on.
Kritzkast: Please, Mr Walker, may we have a guard doggie for our Birthdays? We promise to walk it and feed it and wash it when it smears its body in the dismembered corpses of our fallen foe. I shall call mine Kevin.
Robin Walker: See the analysis line above, especially the last “cost to build and ship” bit.
Kritzkast: With the fast approaching (valve time) last known update, the engineer’s, what assurances can you offer that Valve won’t simply drop the TF2 update program, close the blog and move on (to TF3)?
Robin Walker: We’ve never really planned too far ahead in TF2, because we want to be able to react rapidly to community feedback. So, we can’t give you any assurance that we won’t move on to another product at any point, but we wouldn’t have been able to give you that assurance two years ago, either. We’re already working on the next big update after the Engineer pack, though.
Kritzkast: I think I a lot of people are holding their collective breath waiting for the your trading system to go live. Are those plans still on the engineer’s table or did a wayward OMGWTFBBQ incinerate them all?
Robin Walker: We’re still working on trading. It’s been slower to implement than we’d like, mostly because it’s a feature that straddles across the development cycles of both Steam and TF2. So there are pieces of work that we need Steam’s dev team to implement (and they’ve been real busy on the new UI, among other things), and other pieces we need to do ourselves, and we’ve been real busy on the Engineer update and other things we believed were more important. To summarize: yep, still coming. Sorry for the delay.
Kritzkast: Will there be a place for 3rd parties to create there own trading facilities such as shops, trading posts and semi-automated trading?
Robin Walker: Like anything else we do, we’ll ship what we think is a solid initial feature set, and then see what feedback we get. If that’s what everyone is screaming for, then that’s what we’ll work on.
Kritzkast: On a personal note, do you and your dev team still play TF2 outside of the test environment? Are you any good even without invoking your God-mode-esq rocket launcher? Have you tried scrimming or are you strictly a pub/LAN player?
Robin Walker: We used to be good in the few months following TF2’s release, prior to which we’d been in months of heavy playtesting. Now, we’re all old and tired. Most of us still regularly play TF2 in the wild, but it’s almost all pub play. On rare occasions I’ll join a scrim, with some of the competitive folks on my friends list, to remind myself how terrible I am. Luckily, this isn’t very relevant to TF2’s design, because we try hard to avoid using our own play experience as data when we’re working on TF2.
