Supreme Commander 2

Supreme Commander is a fearsome beast of an RTS: utterly bewildering to anyone who isn't a long-term veteran of the PC's most mainstay genre, but profoundly satisfying to those who have always sworn by base-building. With Supreme Commander 2, Gas Powered Games wants to bring everyone in from the cold instead, as you can read in last week's Supreme Commander 2 preview.

But how the bleeding crikey do you make an RTS that's about shuffling hundreds of tiny, futuristic tanks across maps the size of Wales accessible? Eurogamer sat down with Gas Powered Games' mile-a-minute CEO Chris Taylor, the hyper-brain behind the likes of Total Annihilation and Dungeon, and made him say words. Lots of words.

Eurogamer: What kind of a sequel is this?

Chris Taylor: See when people sequel games, they tend to give it a paint job, add a little of this, a little of that. But you get the same core experience. We, on the other hand, completely rewrote the rendering engine. The old one used a dynamically tessallating model so that as you zoomed in and out there were different levels of triangle density and then the CPU had to push that over the bus to the GPU. It actually gobbled up a lot of the system performance, made the game require a pretty beefy system. This game we've improved that so that we push the geometry over, and we leave it there. It's a better architecture.

It's also got a brand new lighting model - we use a global illumination model, we use a point-cloud system, all-new shadowing, all-new water, new pathfinding system. You provide one graph, one destination on a map that will take every unit from that location to that point. So it's a wonderful system for moving hundreds of units. The other thing we've added is a little bit of neural net AI as well as the handcrafted AI we already use. With neural net AI, you run a hundred thousand simulations, and you build a database of tendencies - it can bring a more natural and a smarter opponent to the AI.

Eurogamer: So is it just the same SupCom experience but much more efficient, or something else?

Chris Taylor: Really the top of the list was to tell a story in the single-player campaign that has a focus on the characters. Think of the difference between watching the History Channel where it's all about Rommell and Patton, or instead watching Saving Private Ryan, where it's all about taking the girl up the barn, and how when they were kids they played together and oh, the roses... Y'know, all the stuff that takes us into the lives of the characters. That's an extreme example, but when you go away from the high-level History Channel version you get a more interesting story. In our story, it's about where do your loyalties lie, at home with your family or with your country? It's a good story, because it tests the character.

We go into the lives of our characters: we have a father-son story that we tell, we have a brother and a sister story. Going into battle, these characters are dialoguing during the fights, and you have these moments where it's really cool to feel you're in a sibling situation, or later where you're thinking, "I don't wanna do what dad tells me to do." We're trying to be place these characters in the cockpits. It's neat to be able to walk through that kind of experience. That's number one for us in this game. For most of the mission briefings, we're actually doing it in-game, rather than having all these bleeping screens you have to sit through first, saying "you must attack from the south..." We really put you right in the game and feed you the information you need.

Eurogamer: And you've really pushed the gigantic Experimental units to the fore - you were showing us a giant dinosaur, a structure that can catapult armies to the other side of the map...

Chris Taylor: We have 27 Experimentals - minor and major. Minors show up earlier in the game, 10 or 15 minutes in, rather than having to wait 40 or 50 minutes before you can build something awesome. We also have the notion of half-baked, where you can be building an Experimental unit but need it in a hurry - so you can push it out of the gantry after it's 50 per cent built. There's a percentage chance that it will fail, based on the completion: so if it's 90 per cent done there's a 10 per cent chance it'll fail. It cycles every 10 seconds or so, and decides if it's going to work, over and over again. If it fails, it's only going to be another 10 seconds until it tries again. It's like the guys in the engine room have oil spraying everywhere, are hitting things with a spanner - we simulate that. It's a neat bit of drama.

Eurogamer: Given all these different types of killing machine, how are you going about making good on your promise of this being a more accessible game than SupCom 1?

Chris Taylor: Some things had to be cleared out. I think the building adjacency bonus has gone, we dropped co-ordinated attack mode, that we don't believe anybody used, we dropped a lot of our travel time estimates system which we don't believe anyone paid attention to either. We made some tough cuts - think of it as we had to lighten the game up to add new stuff to it, because otherwise the overall result would be a game that's unbearably huge. You don't want a game where your friend is doing all these things to you and you don't know why you're losing. There are a lot of things that happen more on the surface now.

You don't have a limit on mass and energy like you did in the first game. So it's never a bad idea to have an engineer picking up resources. You just don't get punished that way any more - there's no downside to having too much.

The next big thing that we did is reduce the unit count from over 300 in the first game to around 120 or so. What this does is allow the player to learn the units then delve into the depths of the tech three. The tech tree is now this system by which you accumulate points, and can assign them yourself - range here, I'll throw some health, here's a demolisher mobile artillery unit...

You get these points by doing damage to your opponent, or you can build research stations. You've got land, air, naval structures to upgrade and the ACU, and it's tough - because when you're in a game and you're fighting with land, fighting with air, do you put your points into land or air? Or onto the ACU [the commander]? Adding a tactical missile launcher - that sounds like a really great idea, right? But you want it all. You have to a tactical decision as to what you're going to do.

Eurogamer: Do those unlocks carry between levels? Are they a persistent thing as in Dawn of War II?

Chris Taylor: In single-player it sort of does and it sort of doesn't. There's probably 170 total, about 60 per side. In the first mission you might get to build five, in the second you get 10, but we might fill in the first five for you. The third there might be 20 but we fill in 12. It's the lower-end ones that you had to research a couple of times, we just give you those for the future because we want you to concentrate on the new ones and learn all the toys in the sandbox. So by the time you've finished playing all three chapters, you'll be comfortable and familiar with every faction's units and upgrades. You will have done it all. Which I think is better than the old way. I think it was Blizzard who pioneered doing this; the RTS community feeds on each other's ideas quite a bit.

Eurogamer: Do you feel like conventional RTS has given up and is moving on - so many people just can't do the multiplayer now?

Chris Taylor: We're definitely focusing heavily on our story and our single-player game, we want to make a really rich experience for somebody who doesn't want to go online and fight other people. Should we be moving away from base-building like in some other games? I actually think base-building is really fun, and an important part of the game. Innovation doesn't mean we move away from that. Innovation means we do things like add strategic zoom, we have really cool Experimental units and half-baked systems and systems that really make the game more lively, but not at the expense of the core RTS experience that we've learned.

Eurogamer: Do you feel like you guys are holding fast against a trend of abandoning this core RTS experience?

Chris Taylor: Yeah. I think we're doing a good job of revitalising RTS but keeping it moving forward in a way that doesn't forget itself. If you love Dune 2 and Command & Conquer and Total Annihilation and Starcraft... I mean, certainly Blizzard is staying on that track, they're not throwing away the old formula.

Eurogamer: Well, they're not likely to give up on it given how much money they'll make in Korea from Starcraft sequels...

Chris Taylor: It's no secret that they have a double-edged sword there, because you need innovation, but if the Koreans up and stop playing the game... What they should do is they should take Warcraft IV and innovate on that, because they don't have the same risk and they do need a free hand.

But I definitely think there's been some desperate moves in the industry to find a new place for RTS. There was a game that had you fighting on the surface and below the surface - Armies of Exigo - and there was one where you could be fighting on a ground map and also in space. All these kinds of variations, but it's like saying let's add a fifth wheel to a car, or let's take a wheel off. But maybe we can actually make the car more comfortable, maybe we can make the drive less noisy or more fuel-efficient.

There's other places to go than just pure breaking something off or sticking something on to innovate. I actually wrote an internal essay about this - when games run out of places to go and apply a gimmick instead, it's a turn-off, and not having had the resources or the dare to do it, they would have sold more copies.

For years we've been beating the drum "innovation, innovation, innovation", but innovation for innovation's sake actually takes you backwards. You've got to have the right idea - like what we're doing right now with Experimental units. We know Experimentals are great, but we've got a nice balance between offensive and defensive, land, air and sea, so we can spread them around that way, and it's not nine different versions of the Universal Colossus. You only need one!