Sponsored By

Analysis: Can't Someone Else Do It? AI Advisors In Strategy Games

Examining design issues and gamer experience within the strategy genre, writer Troy Goodfellow looks at how two different games -- Distant Worlds and Hearts of Iron -- approach the problem of AI advisors.

Troy Goodfellow, Blogger

May 3, 2010

6 Min Read

[Examining design issues and gamer experience within the strategy genre, writer Troy Goodfellow looks at how two different games approach the problem of AI advisors.] Though not the dominant genre it once was, strategy designers still beat at the same problems that faced Bunten, Meier and Tiller in the genre's heyday. At their best, these are games about planning and understanding systems, not simply reacting to the newest thing on your screen. Strategy gamers love to pretend that they are gods, and not distant Deistic gods simply setting things in motion. Strategy games give you worlds to build and worlds to destroy. Many of them make you believe you can do whatever you want, but the control freak tendencies are sometimes constrained by information overload. Most of us can only process so many connections, possibilities and events. For example, classic empire building games (often dubbed "4x" for their emphasis on exploration, expansion, exploitation and extermination) often run into the same problems that real empires do. Once you have built an empire over a certain size, the management of that empire becomes more cumbersome. You might have a large army scattered across more fronts than you want to manage. You might have too many colonies that you don't really care about nagging you for new orders. If the game is really complex, you might not even be comfortable with how everything fits together so some decisions you make just feel like crap shoots. Enter the virtual viceroys – the AI controlled advisors, governors or generals that you can call on to either advise on or manage the things you don't want to care about right now. Lots of games have used these 'virtual viceroys' at a basic level. You could set planets or cities to produce structures or units on their own (games in the Civilization and Master of Orion series, for example.) The global political sim Supreme Ruler has cabinet ministers with their own personalities to whom you can assign priorities; you can always set them to follow your concerns, though, so the personality thing never really comes into play. But designing to give the player confidence in the virtual viceroys is harder than it looks. You have to overcome the strategy gamer's instinct to control everything and, most importantly, navigate Meier's Maxim; if a game is a series of interesting decisions, won't a player want to make them all? Distant Worlds Code Force's recent science fiction 4x real time strategy game Distant Worlds is all about making the player comfortable in ceding control to the AI and does it better than many larger budget games. You can cede control of just about anything to AI control and on its default starting setup almost everything is automated – colony taxation, fleet management, and various other imperial duties. By starting you off with a lot of automation, Distant Worlds is the inverse of most 4x design. It doesn't ask you whether you want to cede control; it asks you whether you want to take it. And one major component of your imperial economy – commercial trade – can not be controlled at all. These are private enterprise beyond the reach of your throne. The capitalists decide which resources to exploit, and all you can do is make sure their fleets don't get raided. This little bit of reverse psychology proves to be integral to making the illusion of the effective AI governor work. It's kind of a sleight of hand. You see things working well enough in most cases that it might take a while before you realize that the fleet construction advisor is forgetting to suggest fuel supply ships or that fleet management isn't working as smoothly as it should. But until that moment, you see the ships moving around and doing their thing and you can convince yourself that you can wait a while before you take over that job. At which point the game even asks you whether you really want to take over. Distant Worlds' core assumption is that empire management is a difficult thing and that you can't really be bothered to take your mind off the big picture stuff. You can slowly work your way up to mastering every task but there's no need to. Hearts of Iron 3 Compare this to the design philosophy of last year's Hearts of Iron 3. This giant World War 2 game from Paradox is often only manageable because of how many ways it lets you give up control. You can decide at the outset to give up some decision making power, but the default understanding is that you won't until you feel you need to. You can set general objectives on each front and your AI commanders will move the troops to get there. You can ask the AI for recommendations on what to build and it will give you a long list of armaments you are lacking. Almost any system can be transferred to computer control, from diplomacy to research - both of which require a deep faith that the AI's plans are kind of like yours. Once you've let go, it's hard to break yourself of the mindset that you could be doing it better. It's one thing to give up the boring duty of finishing off Denmark's army, but do you really want to blindly follow the AI prescriptions on what units to build? How does it know what I plan on doing? Hearts of Iron 3 is a much more complicated and complex game than Distant Worlds and even when you understand everything you never feel like you have precise enough control. It is easy to simply pass the busy work onto the computer and hope for the best, but there is always that desire to reassert control. All told, it's not really clear if Distant Worlds' virtual viceroys are any better than those in Hearts of Iron 3. The construction advisors in both games have major blind spots insofar as recommending what your war effort needs right now. Both games had recent patches focused on the AI's military management, and problems with AI opponents are often reflected in the effectiveness of AI advisors. Designing Backwards The differences in game structure, however, lead to very different initial responses from me. If you play Distant Worlds, you have to volunteer to take control of a new job. If you play Hearts of Iron 3, you have to choose which job you can trust to an iffy AI. The former asks you to make a choice to control, the latter asks you to make a choice to trust. So the general community response for each has been radically different – Distant Worlds users seem more likely to praise the option to only do what they want to do as they learn the system and Hearts of Iron 3 users are more likely to worry about what they are losing. The solution to the problem of accepting virtual viceroys, it seems then, is to change the terms of the conversation. Instead of starting with the gamer as Grand Poo-bah, in charge of every office the empire has to give, the mechanic is more palatable if you start with more discreet responsibilities. Strategy games are about learning systems, for the most part, so by letting the player choose which systems to learn in which order, you can give the player greater control over his/her play experience. [Troy Goodfellow is a freelance writer based in Maryland who blogs about strategy and war games at Flash of Steel. He also hosts Three Moves Ahead, a weekly podcast about strategy gaming.]

Daily news, dev blogs, and stories from Game Developer straight to your inbox

You May Also Like