Interview with AI Wars creator, Chris Park

If the average RTS were a comic book, Arcen Games’ AI War: Fleet Command would be a novel by Iain M. Banks or Alastair Reynolds. With a focus on co-op and a formidable, truly alien computer opponent, AI War eschews the usual micromanagement and clicks-per-second in favour of deep strategy, and conflicts that encompass entire galaxies.

Hellbored sat down for a chat with game creator, founder of Arcen Games and all-round smart guy Chris Park.

How would you explain AI War: Fleet Command to somebody who has never played the game before?

AI War is an RTS/4X hybrid, centering around a conflict between the last survivors of humanity - you and up to 7 friends – against a pair of far-more-powerful AI opponents. It’s a David and Goliath simulator, in many ways, and challenges the human players to engage in careful strategic planning and guerrilla warfare. The tactical AI is becoming well known as some of the best around, which is important for a game like this, and the overall game design is built around emphasizing long-term thinking and tough choices on the players.

Aside from our focus on co-op, the unusual genre mashup, and our great AI, our other claim to fame is the incredibly high unit counts. Thirty thousand or more units are in most games, and they commonly reach double or triple that number. In one case, on very nice hardware, I actually know of a player who has over two hundred thousand units in his campaign. Most of those units belong to the AI, but the human players themselves are routinely moving vast numbers of ships around - one thousand to seven thousand ships is pretty much the normal range per human player – and we have a pretty innovative interface built around making that scale meaningful and easy to control. Rather than this just being a gimmick, we use it for the emergent behaviour in the AI, and to encourage group tactics in place of unit micromanagement on the side of the humans.

In short, I guess, it’s a very different sort of game that only an indie developer would ever create. That makes it hard to sum up briefly, because it’s like trying to explain modern wars to knights from the middle ages. The general idea is the same, but there are so many important differences that it’s a fundamentally different activity than what people are used to. AI War’s depth is not going to appeal to everyone, but we’ve found a great following amongst those players who are into hardcore strategy games, who are interested in co-op, or who like classic RTSes but who don’t like the fast-clicky nature of many of the more recent entries in that genre.

How has the fan response to the recent expansion The Zenith Remnant been? What are some of the new features that have been introduced?

Fan response has been really great so far. The game was available for preorder for several months, with players who preordered gaining access to our beta versions. We had dozens and dozens of players giving us direct feedback, and that was very helpful in shaping the product as we went. By the time the official version was released, we already more or less had the blessing of the core community, which was a good “sanity check” to make sure the expansion was what people wanted. At the moment, after about two weeks of the expansion officially being out, we’re somewhere between 10% and 15% of the owners of the base game now having the expansion; it would be more, but the expansion has caused a surge in sales of the base game, too, so our overall total playerbase has nearly tripled since November.

The expansion is mainly focused on increasing the long-term strategic options and game variety. It adds more to discover out there in the galaxy, both bad stuff that the AIs use, and good stuff that you can capture for yourself. The base game has a ton of variety already, over 120 hours of content if you were playing full campaigns and wanted to see every last feature as of 3.0 (we add more content through monthly free DLC, so that goes up with time). But, the expansion adds 120 new ships to the existing 250, a dozen more AI personalities, another 40 minutes of music, more map styles, and a variety of new features such as “minor factions” (three of which were also added as part of 3.0’s free DLC). The massive golems and the minor factions are probably the most popular new features of the expansion - although, various of the 13 new smaller ship classes (4-5 ships in each) are also hugely popular.

The universe you’ve built is certainly very interesting. What are some of the works of fiction that have inspired the game world?

Many thanks. Some have criticized the base game for being a bit too much of a blank slate, story-wise, but others see this as a strength. Personally, I tend to imagine that I am Ender while I am playing. I was always into Star Trek TNG as a kid growing up, and Star Wars, and the Ender series, amongst others. Those all were an influence here. The Forever War was also a bit of an influence, as it sparked the idea for how to handle wormholes (of course, I later learned that Sins of a Solar Empire does the same thing -- go figure). A lot of players tell me that they picture themselves as part of Battlestar Galactica while playing, but (gasp) that’s not something I’ve yet watched. It’s on my list, I swear.

With AI War, I basically wanted to create a custom campaign engine, rather than any one specific story. With realtime campaigns that last 7-12 hours on average, the planets and events start to take on a lot more significance than you typically see in the RTS genre. Of course, this is something of a hallmark of the 4X genre: in Civilization IV, for instance, there’s not a specific overall story to a conquer-the-world game, but you wind up with a grand epic by the time you finish one, an epic that is very specific to you and what you did. AI War is built around the same sort of concept, and I think that’s why players are able to imagine it in the context of so many of their favourite sci-fi franchises.

What are the challenges involved in developing as an individual, as opposed to having a team at your disposal? Do platforms like Steam make it easier to get a title out there?

Well, for years I was a hobbyist game developer, and that was always working alone. In a lot of respects that was easy because I could do everything just the way I wanted, and in whatever order I felt like, with the downside being that I had to do everything -- which in the case of art, music, and sound effects meant I had to find those for free, which is limiting in a huge number of ways.

With AI War, it started out with just me doing everything (and using art from free sources, mostly Daniel Cook’s work), but by the time the game released I had already contracted a composer to do the score for the game. Later, months after our 1.0 release when the game was starting to make more money, then our team also grew. At present the company consists of me fulltime, a fulltime composer, a fulltime artist, a new part-time programmer, a part-time designer, and a part-time techwriter. The last two also help out with forum moderation and support. And we’re still overworked, collectively.

Now that AI War is out there with a growing playerbase, there’s no question of me being able to do everything myself. Our community is partly guided by a lot of longtime forum members who answer questions for new players and are on the front lines of suggestions and bug reports. It definitely takes a village, and in many real senses it always has - even when it was “just me,” I had a trio of alpha testers (my dad, my uncle, and my uncle’s colleague) who contributed immeasurably to the development of the game with their feedback. I don’t think there are any true one-man projects past a certain point of complexity, even when one person is doing all of the design and programming.

Regarding digital distribution platforms, they are definitely invaluable. AI War is coming to retail in a couple of territories (Germany and Russia at the moment), but we never would have gotten those deals if we hadn’t had existing success in the online space. The first digital platform to take a chance with us was Stardock’s Impulse, and our early successes there paved the way for everything that followed. In a very real sense, without some form of digital distribution I don’t think you’d see very many financially successful indie developers - or, at any rate, even fewer than you see now.

While graphics and the incorporation of scripted events have come a long way since the days of Dune II, sometimes it seems like there hasn’t really been any remarkable advances in the AI of computer opponents. What is it that makes building a capable, strategic AI so daunting?

Well, there are a lot of academic approaches to creating theoretical AIs, but not all that many of them are used in games. I think that the main limiting factor in many games is that AI isn’t really a priority, and a lot of AI programmers lean on approaches that are... outdated, I guess is the word. On the other hand, there are a huge number of AI innovations in various genres, even if the innovations are not always in-your-face apparent to players. FPS games in particular have some pretty amazing AI in a lot of examples: the Far Cry games, Killzone 2, and so on. Certainly there is always room for growth, but I think that some genres tend to do better than others. Turn-based strategy games often have some pretty wicked AI, for instance, and have for years.

In the RTS space, there are a ton of challenges, and I think it traditionally has been one of the weaker genres in terms of AI. Often this is because everybody focuses on pvp, and so the AIs are just a poor man’s stand-in for a human opponent. In pvp RTSes, often success is derived from making really intricate choices in a fairly narrow decision space – in other words, you don’t have a huge number of truly viable options, and success is predicated on executing your strategy in an optimal manner. Often those games are built around doing lots of micromanagement and positioning of units - it’s all very intricate and precise, and requires a lot of intuition and experience and reaction speed at the upper levels of play. Generally speaking, that’s about the worst environment to throw current-generation AIs into, because they tend not to react terribly fast with so much data to crunch, and they tend to do poorly in cases where there is a high “opportunity cost” to making a lot of sequential decisions. In most RTS games, there is both too much and not quite enough for the AIs to do: there is too much for the AI to think about to be effective, but not enough latitude in potential activities for the AI to do something really surprising.

So with AI War, part of why I was so successful with the AI was my novel approach to the game design itself. In other words, I set out to make an RTS game that would be an ideal breeding ground for good AI. That gave me a huge advantage right from the start. Secondly, I took a radically different approach to the creation of the AI itself, basing it around emergent behaviour and fuzzy logic rather than the common decision trees. This is something that AI researchers have been talking about for decades, and you even see the influence of this in some of the better FPS AIs, but these techniques have never been applied in quite this way to a commercial project with so many agents, so far as I know.

My hope is, now that I’ve demonstrated its viability, other developers will experiment with these methods; I’ve written rather extensively about the theory behind what I’ve done with AI War. Rather than being worried that others will copy my innovation, I rather hope they do - and that they improve on it, even.

You mentioned free DLC and other updates being made available. Do you have plans for more expansion packs like the Zenith Remnant?

Definitely! We do free DLC every month, and that will continue for the foreseeable future, hopefully another 2-5 years at least. Every 10-12 months during that time, we’ll also put out a paid expansion pack of about the same scope as TZR. AI War is still growing, and my goal is to form a long-term player community around it rather like what happened with Total Annihilation. So far, so good, in our niche.

Finally, what words of wisdom would you like to share with any other aspiring game designers out there?

My best suggestion is this: make the games you always wanted to play. That might sound obvious, but if you look at a lot of indie games that are unpopular, you see a lot of derivative work, a lot of trend following. People hope to make the next World of Goo or the next Flight Control, or whatever. But the next game of that popularity won’t be anything remotely like those past titles - those have been done. It’s important to focus on creating the games you always wanted but never had, and at least some success will follow if you do a really good job and then manage to market it.

Thanks for the interview!

My pleasure!