CA On Rome II's AI, Scope & Incendiary Pigs

Veni, vidi, voice recording

Feature by Alec Meer Co-founder

Published on July 3, 2012

Yesterday, you first endured me waffling on about what I'd seen of Total War: Rome 2, and then harkened to a cleverererererer man than I shedding further light on his studio's intentions for their next epic historical strategy game. In this second and final part of my long chat with Total War series lead designer James Russell, we cover the geographic scope and scale of the game, why it's not Supreme Commander BC, the importance of multiplayer and - here we go - whether and how the AI has improved.

Editor's note - as there's currently a shortage of promotional screenshots for Rome 2, I've had to get a little inventive with the illustration. Special thanks to Kieron Gillen for providing the last image here.

RPS: Is there any part of you that thinks ‘maybe we don’t want to call it “Rome” because that’s just a part of it, actually it is all these dozens of different nations’?

JR: Yeah. I think Rome is the most identifiable and obviously Rome was the winning power in a way, in that game you’ll start off small, you’re not going to start off controlling lots of different regions. You’re going to make your own empire, so anything could happen. When people think of the idea of the Roman world, I think in the popular imagination it includes the enemies of Rome and the different cultures that they faced, I think it conjures up a very rich world, partly because the empire was so global. It’s different from the idea of Greece.

Any empire that became so powerful obviously does horrendous things, I think you only get military power through that, doesn’t make it ok, but I don’t think they’re any different from any other power in history really apart from perhaps, well actually no, I was going to say a slightly declining brutality with time, but actually no, that’s not valid at all. It was a militaristic society but it became much less so.

In Total War obviously we start in the Republican era and we obviously want to get to the height of the Empire, but the height of the Empire was under Trajan technically, but the Empire was pretty much at its peak at the time of Augustus, and then it lasted for hundreds and hundreds of years after that, before it split, before Constantine. It was actually quite stable and peaceful for a long time.

I remember seeing some interesting estimated statistics of probability of death in a year for different situations in the armed forces, and a Roman legionary in 100AD was very, very safe. They really weren’t doing a lot of fighting. They were garrisoned across a huge empire. There’d be the occasional little war. This was after it was established, we’re looking at this at the time when you’re making a Roman Empire, which obviously was a much more brutal age.

RPS: What is the geographical scope of it? Is it just Europe or are you going to have Egypt and North Africa as well?

JR: It’s definitely bigger than the Rome 1 map, so it will obviously include all the areas that the Rome 1 map included, like Egypt, but we’re going to go further east in particular. Different directions.

RPS: Multi-player seemed to be a bigger thing in the Shogun 2 games than in earlier titles. Is that a philosophy that’s continuing generally?

JR: We’re not really ready to talk about multi-player, all we’re saying at the moment is that we’ve got big plans for multi-player.

RPS: In your experience does it seem most of your audience are in it for single or multi-player?

JR: I think the core Total War experience generally is I think about playing your campaign. Lots of people play multi-player campaigns, that core single player deep strategy campaign experience is obviously the focus of the series. We’re looking at doing something different for multi-player. I guess that’s all we should say really.

We’re really committed to constantly improving the AI, because that’s a very important part of the experience, particularly for single player obviously. We’re doing loads of new stuff with the AI, part of that human face thing that we were talking about is to make the AI in diplomacy, for instance, feel a lot more human and a lot more convincing. So we’re doing a whole bunch of things related to that, to add a lot more personality and let the player understand the motives of the AI more. We’ve got these two separate systems; there’s the diplomatic relations system and the core AI intentions system. What we’re doing is combining those into one. The AI intentions and the way it behaves diplomatically won’t be separate systems that are communicating, the will be the same system. So we’re putting love and hate at the very heart of the AI’s intentions.

What it means is it’ll allow us to tell the player, perhaps if they spy on the AI or whatever, why the AI’s doing certain things, or warn the player that it might do certain things. So for instance, you might think ‘well why is the AI refusing to trade with me?’ and you might be able to find out that it’s because it really covets this particular region that you’ve got, and it wants to attack you. ‘Why won’t it ally with you?’ Well, it actually doesn’t trust you because you broke an alliance before with someone else. All of these kinds of things will actually be able to unify because they’ll be all part of the same system, so will be able to tell a player more about the AI intentions. It’s all part of making it feel more human, giving it more personality, and just making that world come alive.

RPS: And you’ll be able to affect those decisions?

JR: To an extent, yes.

RPS: And AI in general I guess has become a bit of an albatross round the CA necks, at least in the perception of some. Do you feel they’ve had valid concerns or has it become an unfair accusation that you can’t shift no matter how hard you try?

JR: There’s a number of things I’d say there. Obviously as a player, you want a game that offers you a convincing game, and an enjoyable game, and a challenge. No amount of talking about how difficult it is to do a good AI is going to change that, you’ve got to create a great experience. I feel that we’ve made significant strides in the AI over recent years, and the Shogun 2 campaign AI for example, is I think a lot stronger than any campaign AI we’ve had before. So we’re constantly improving AI. Certainly it’s the case, working on Rome 2, we’ve got more AI effort than we’ve had before that’s for sure. We’ve got design and programming working on AI, we’ve got a bigger AI team. It’s something we are determined to make work well. I don’t feel that it’s a massive weakness as a series in a way.

CA's PR: What is slightly galling is adopted perception as well - you can see conversations on forums, and someone can quite easily say ‘oh, the AI’s been shit since Empire’ or something like that, and it simply isn’t true. Especially with Shogun 2, so much work went into improvement, and even Napolean was a leap over Empire. It’s just people have a perceived perception. It’s like: here’s a thing I know about Total War, so I’m going to drop that into the conversation, without having actually played through the different iterations.

RPS: Have you been able to ascertain what amount of the playerbase have these concerns?

JR: We certainly know it from a metric perspective that people certainly want a challenge at different difficulty levels, and the vast majority of games are played on Easy and Normal. So there’s a relatively small proportion of games played on Legendary or Very Hard. As an example of how things are different in Shogun 2, on a normal difficulty level the game cheats in favour of the player. So the AI’s not cheating - we’re cheating in favour of the player there, so that’s different. In Rome 2, we’re pushing it even further. We’re really wanting to make the AI a great, convincing human opponent, both in battle and in the campaign.

RPS: Can you talk in layman’s terms about what it is you’ve done to really tighten up and expand these routines and stuff?

JR: I mentioned in terms of the diplomacy unifying those systems. I think it’s also about having more people and also design attention given to exactly how we want the AI to work, how we want it to play and win the game rules that we are designing. I think it’s important to think about AI right from the get-go. So we think ‘Ok, well we’ve got this system designed, how do you win that system?’ And then we talk to the AI programmers about how to do that, genuinely within the context of those rules. Similarly on the battlefield, when we design the city lay-outs, like you saw in Carthage…

RPS: If all of that’s navigable, that must be a huge task on so many fronts…

JR: The point is that we design those lay-outs thinking about the AI, that’s the point. We don’t make the world and then get the AI to try and deal with it. We design it thinking about how the AI’s going to navigate it, and play with it. It’s just that change of approach that is what makes the AI stronger, and just making sure that we’ve got a really really solid amount of programming effort - and a really solid amount of design focus on directing that effort and working out the optimal things for the AI to do.

RPS: And in terms of those cities what’s the scope of the sieging you can do, you can go all the way round it, can you be trashing anything, creating your own path by knocking down this row of buildings? How dramatic can that get?

JR: Partly those cities will be pre-destructed because we want to create that sense of armageddon, when you’re besieging a city, so we don’t want to have a pristine city for you that you then completely destroy. In a Total War battle they’ve all been under siege for some time, so we want to create that sense of dark, foreboding. Actually it should be apocalyptic when the city’s been under siege for a long time. But really I think it’s about making the gameplay strong, and what makes the gameplay strong is having multiple capture points, because it means you’ve got a lot of cat and mouse play within the city.

In previous Total Wars often when you just had one plaza, effectively once the walls are breached, the defender was incentivised to come back and put everything in the plaza to defend it. Having multiple capture points means that you’ve got cat and mouse gameplay, you can heavily defend one capture point but then they can come round and flank you or create another breach on the other side of the city. Also, having the cities being bigger means there’s more scope for proper developing gameplay inside the city, rather than it being all about breaking the walls and trying to get through that one breach.

RPS: Or just ending up with a queue of dudes waiting to get in, which is what I’ve had a few times.

JR: Yeah. But also the combined battle, the fact that you’re coming onto the beach, you can have potentially defensible points near the beach, which means that they’re potentially an incentive for the defenders to come out of the city to attack you as you’re landing. There’s all sorts of possibilities and I think that’s the focus of the siege battle design, is really making the gameplay as interesting and varied as possible. So it’s not necessarily that every single building will be completely destructible.

PR: Not all battles are going to be that big, that’s essentially a boss battle. That’s the capital city of an empire.

JR: But we do want to have variations. We’ve got the barbarian settlements which all look different... The guys doing the building work are actually trained architects, the artists in charge of all of that. They’re really obsessed with exactly how a Greek city looked, and how a Carthaginian city looked, all that kind of stuff.

RPS: Building wattle and daub huts to live in, I hope. And you’ll get to defend sieges as well presumably?

JR: Oh yeah.

RPS: In terms of the naval aspect from the defensive perspective, are you able to knobble the boats as they’re coming in towards your cities?

JR: If you’ve got the artillery for it, then yes. Also you might be able to have your own naval reinforcements, so you might have a defending navy. You saw Carthage there, the Carthaginian navy got destroyed, but Carthage there had a big harbour, you’ve got your own ships in there. If you were defending Carthage, you could have your own ships in there and come out and have a naval battle before you even land.

RPS: It must be possible to have some incredibly long battles with cities that size and that many different elements going on as well.

JR: I think the key thing is that you’ll have some battles that will last longer than others, the battles tend to be a good length. The very longest battles tend to be 45 minutes or something like that, maybe an hour I think. We’ve got an hour-long setting, a time limitation of an hour. Those are the very longest battles, so they’re reasonably contained. It’s very rare to get that long. They vary from 20 minutes to 40 minutes for a big battle.

RPS: It says a lot about what that visual scale suggests, because I was watching it thinking ‘ah, this is Supreme Commander BC, this is going to have to take three hours’.

JR: No no. There’s no save point in battles.

PR: But as you saw, the action telescoped there fairly quickly, from being in the sea, onto the beach, breaching the wall and getting the city gates open by capturing that point, getting into the city. Although that demo of the Carthage battle only gets halfway into the city. It's what, about 9 minutes long? I think that’s a very quick battle but yeah, I think that could be a lot longer.