Enemy Territory: Quake Wars

Eurogamer: John Carmack responded to a point about Enemy Territory tick-rates in the question-and-answer session following his keynote. He said that animation was running at 30Hz and frame-rate at 60Hz. Is that something people should be concerned about?

Paul Wedgewood: There's a ton going in Quake Wars. You have to realise that there are 24 players, all of them can have deployables that can be firing, and be in vehicles, and then of course we're tracking experience points, objectives and their status, and have physics attached to almost everything. Even just overcoming the network challenges took some significant advances.

In terms of the server and the client, the problem comes when you try to process all of that data 60 or 90 times a second or have it uncapped, because the gameplay experience when a server a or client shifts constantly from 30 to 90 to 60 ends up much worse than having a locked-out 30fps. We started by reducing everything to 30Hz to see how that would work, but players felt the gameplay wasn't smooth enough.

So Timothy over at id Software worked on a solution that unhooked rendering from game sampling, and we have a much smoother experience now. We have unlocked frame-rates and people who were previously around 30fps are now getting 60fps because we've made a ton of performance improvements. That said, the animation problem can't be solved. It can't be. As the game stands right now, it's just one of the insurmountable hurdles.

The game's many vehicles give it a certain feel, but their impact changes throughout a level's various phases. Not much use underground, for instance...

Eurogamer: Carmack described it as needing major architectural changes.

Paul Wedgewood: Yeah, but ultimately the most important thing is that what you shoot is what you hit, and that's more important than having really smooth animation. We're faced with the choice of giving you a smooth gameplay experience with really good hit-registration and really good player-prediction and really good networking, or something that is heavily interpolated and gives you the impression that everything's running really smoothly, except that vehicle isn't really where you think it is and that animation isn't really playing and half of the game is client-side prediction.

Eurogamer: Illusion or the truth.

Paul Wedgewood: Yeah exactly, and gamers are more obsessed with knowing that what they're shooting is actually where you say it is, and that's what we've achieved with things like the anti-lag code. Say you're on 150ms ping and I'm on a 33ms ping - if someone's head is under your sniper rifle when you shoot them you actually get that hit, that hit is registered appropriately, and that levels the playing field without imposing any penalty on anybody else. The low-ping guy doesn't suffer for your lag.

Beta 2 features the Valley map. Around 38,000 people played Beta 1, according to Wedgwood.

Eurogamer: Because if he gets hit it's because he was exposed at the wrong time. Right. I think people were just looking at the comments Carmack made and were uncertain about how significant they were.

Paul Wedgewood: I really don't think it affects anybody at all. I think we've definitely made the right choice for the gameplay. Of course everything in game development is a series of compromises, like whether to have destructible geometry. You could have destructible geometry, but not have great vehicle physics - you have to decide whether these things are purely immersive or whether they significantly benefit gameplay, and if they're purely immersive then you end up thinking there's not much point simulating reality in this situation and emulate reality instead.

If it's immersion plus gameplay or a pure gameplay mechanism like constructing a bridge with a pair of pliers, we realise that that damages immersion for some people, but it's the best mechanic for making the game competitive and fun, and so we will sacrifice immersion in the pursuit of better gameplay, because gameplay has better longevity. The game isn't going to look good three years from now, but the game should be as fun or more fun.

Eurogamer: Could you give an overview of the new maps you're presenting?

Paul Wedgewood: Sure. In Quake 2 the retaliation against Stroggos is led because the humans have learnt how to use slipgate technology and the first map, Area 22, is really the telling of that battle. It's an arid-themed map. It has the Strogg attacking, initially to take the power out of a big EMP that the GDF have erected. When they take it down, the Strogg are able to deploy a mining laser from orbit. This blows the doors off a bunker and gets them into the final objective room which, just like every other map, is entirely unique. You've got the big slipgate with all of the lab equipment around it. You'd never confuse it with any other. Each map really works as a mnemonic.

Eurogamer: So that's one.

Paul Wedgewood: Moving onto Ark, on the coast Norway - a kind of fjord coastal inlet, in arctic conditions - the interesting thing is that the objectives are the same as on Area 22. You have an EMP to attack, a mining laser in orbit, doors getting blown off and an underground objective at the end. But they play entirely differently.

What's going on is...in the Quake universe, the GDF are a paramilitary organisation rather than a kind of organised military unit, and that's the reason they're still using projectile weapons and their vehicles are in poor condition. And it's good, because it gives us a nice asymmetry. If they were more futuristic - laser guns and stuff - it wouldn't be as fun to play against the Strogg. Anyway, because they are a paramilitary organisation, they've developed most of their presence as a rapid reaction force, because the Earth has been through a series of natural disasters and there isn't a ton of commercial research going on any more that's generally concerned with advancing Earth - it's mostly just about holding back the flood. The Ark is one of the few locations where a bunch of billionaire philanthropists have constructed a series of biodomes and they're doing something that's benevolent - storing strains of DNA for species of plants and animals that are becoming extinct.

For the multiplayer gamer, it's perfectly reasonable to argue that on his third time he's playing the map he doesn't give a crap about any of this stuff, but as designers it means that we can come up with a cool location for the map, a cool series of objectives, and when you make that transition from the outdoor part of Ark, going into the biodomes, having loads of vegetation, then going into the facility, the game changes. There's a really significant transition between the way the game feels while you're trying to blow up the EMP to when you're then fighting across the bridge and you have this big choke-point and all of the aerial combat comes into play, to when you then move into the biodome and you have the underground experience which is a lot more like Wolfenstein: Enemy Territory - kind of run-and-gun, assault rifles, leaning round corners and all that kind of stuff.

Eurogamer: And the third map?

Paul Wedgewood: The third, Island, is one that's being developed over at Nerve Software, and takes place on Reversion Island in the South Pacific. It's the location of the last surviving facility that holds data on the first appearance of a Strogg slipgate. It gives us this James Bond-esque mission for the GDF. The Strogg have taken possession of the island and this data is behind a vault door in an underground complex. An engineer has to construct a generator first to bring power up on the island, to power up the satellite which is in the transmission towers up on the top of the mountain.

You then have to blow down a vault door and get the data disk, which is in a briefcase - there's even an escape route down through a staircase to a speedboat in an underground tunnel, which you can then use to escape round the island, or you can try and go straight up the hill with the Strogg forces bearing down on you. When you transmit it to the waiting GDF scientists, that's what led to them finalising their research into slipgate technology. And we have slipgates as a theme in a few of our maps, but the ways that we use them is quite original from map to map, and we have a couple of quite cool surprises for players when they buy the game...

Do you like these QuakeCon screenshots? You know, they wouldn't give us a press CD when we were there. Meanies.

Eurogamer: So all 12 are very distinct.

Paul Wedgewood: If it was 12 maps that were in desert or a jungle and they just had flags that you captured on them, then it becomes a kind of indistinct and repetitive experience. Touching back briefly on the mnemonic idea - you know, um...I first read about mnemonics...I think it was Hannibal Lecter had a highly developed mnemonic memory palace, but the basic premise is that it's much easier to remember things if you store them in physical locations in your head. One of the things that I use is I have these 52 locations in my house, and I can stick things in those locations and I can tour through the house and I can remember all 52 things in there.

But for a player, the idea is that if a map is completely unique and distinct, when the map loads you are instantly reminded that this is that map where you place the landmines on that little strip of road and you watch hogs come haring past and they get blown over the top of your head. The mnemonic reminds you of the thing that you had fun doing, but the sequence of events ends up being a story you can tell other people, whereas it's less interesting to say 'I spawned, I ran forward, I stood there, the flag was captured'.

Eurogamer: Could you talk a bit about the work that John Dean has done on bots?

Paul Wedgewood: Sure. John developed Fritz Bots for Wolfenstein: Enemy Territory, and id Software were really impressed by his work. Well, we were really impressed by his work, but they were impressed enough to actually go and hire him! The design goal was to have bots that played like real human players.

Because we're in beta at the moment, there are still some bugs. Occasionally you'll see a tank spinning round in circles or a Strogg and GDF guy running along side by side completely unaware of each other because their view frustums aren't intersecting with one another, but, setting aside those couple of small things, I think these are probably the most advanced implementation of AI yet in a multiplayer combat game.

They're not just target practice. They understand how to use almost every weapon, item, tool, deployable in the game. They understand how to use them tactically as well. A medic will make a decision about whether to revive a team-mate versus whether to heal someone to cover him while he's reviving the other player. A Covert Ops will jump into an Anansi, fly to the enemy objective, bail from the helicopter, select a camping spot, wait for the objective-completing character class from the enemy to turn up, stab him in the back, steal his uniform, and then stand around knowing that the enemy team is less likely to shoot him if he's the guy who's supposed to complete the objective, and then carry on going round and getting back-stabs. And there's no preset pathfinding, so we actually discovered routes that bots find between objectives that I didn't even know existed.

With the real game, if the bots did everything they were absolutely supposed to but humans didn't, it would suck when you played against humans because they would all just run around selfishly motivated by their own personal objectives. So what we do is we reward heroic behaviour in the game with experience points that lead to unlockables, attribute-modifiers and that kind of thing - rewarding you for making the other gamers' game more fun.

Eurogamer: What else does the mission system do for you?

Paul Wedgewood: It will try to give them the best thing they could be doing at that given time. At one level it works a bit like a tutorial system - a new player will come to the game as an engineer and it will recommend you deploy a turret. Then it will give you missions for things like repairing deployables. Some of those are pre-scripted by our designers, but they're not particularly innovative. But as a second method of generating missions, some are based on intelligence your team gathers, and this works equally well for the bots as it does for the players. If you put down radar, and it determines the location of artillery, that generates a whole bunch of destruction or hack missions for the soldiers or Covert Ops on your team.

The third tier of this is that they can be generated by players on your team clicking on things. The context-sensitive menu allows you to point-and-click at something and it generates the most important order right away, so if your vehicle's damaged you can point-and-click and it generates a repair mission for an engineer. If he points at a plasma mortar and you're playing Covert Ops it will generate a mission for you to go and hack into the plasma mortar and disable it.

Another cool aspect of the bot implementation is that, ordinarily, even if you do play with bots offline, you might only do it for the first week or two. Here you can separate their tactical skill, their aiming skill and whether or not they complete objectives. So if you want to practice your sniping, you can sit up on top of a hill and just snipe, but if you want to go through all of the maps learning where all of the objectives are, you can set the bots not to complete objectives and you can be the guy who does all of the objectives playing through the game continuously.

The Strogg all shout things in Stroggy voices. It's all a bit WarCraft II. Hurrah for that.

Eurogamer: Wicked. So - running out of time - are you going to do a third beta, or is that it now until release?

Paul Wedgewood: I don't know. There's no definitive plans on what will happen or what won't. Ultimately betas depend on whether we can make the fixes that we need to in patches as opposed to full releases. With Beta 2, so much had changed that we couldn't really deal with it in a single patch, so that's what generated Beta 2.

Eurogamer: Once you're done with Enemy Territory, are you going to continue to collaborate with id?

Paul Wedgewood: We hope so! At the moment our focus is exclusively on getting the PC done, and as soon as that's done we'll be focused on whatever additional content and things we can make, to make people continue to have a really good experience. id Software are really committed to the ongoing support of the game, and so whatever work we do I would hope that Splash Damage continues its role as leading on the development on the PC sku.

But as for the next game and stuff, I have tons of ideas flying around my head, and I've been discussing a bunch of ideas with Kevin Cloud over at id Software, but to do anything more than just continue to have ideas flying around would be a distraction from the PC Enemy Territory, and everyone would hate me!