Non-Player Characters make up the entire population of the worlds we visit as gamers. These non-sentient beings (for now at least) are the shop-keepers, zombies, companions, covering fire, and nemeses of the video gaming world. Their behavior is moderated by their programming; they are only ever as smart, helpful, or sinister as the developers choose to make them. In some cases their programming is exceptionally rich: the shopkeepers in Bethesda’s 2006 release Oblivion would return to their homes at the end of the business day and even trade small talk with each other. In most cases they are very simple; the door guards in BioWare’s Mass Effect stand in the same place and do nothing at all until the player character walks up and says hello. It is not the complexity or simplicity of a given NPC’s programming that makes them a valuable addition to a game; it is their suitability to their environment and role. I expect a wolf to act like a wolf, a sniper to be able to hit something, and a military squad to work together to bring me down. Whether they are working with me or against me, I expect an artificial intelligence to at least give me an excuse to believe in it.
Clive Barker has been one of a select few well-known writers who has come out in support of the video game medium, and as a writer and gamer I am glad the medium has his support. As a younger man, Hellraiser scared the pants off me and left me with a lifelong phobia of Rubik’s cubes. Looking back at those old films I still have to tip my hat to the originality of the series. Imagine my excitement when I discovered that Barker teamed up with the developers at MercurySteam to create the first person shooter/ horror release Jericho.
This being a Clive Barker joint, gamers find themselves tossed into a hellacious suburb of purgatory populated by fleshy abominations who want to eat their faces. In Jericho, gamers take control of an elite squad of special forces, each one with phenomenal powers and a fairly well rounded back-story. This is an especially impressive touch given the size of the Jericho Squad: seven characters in total. Gamers make their way through the twisted level designs by possessing each member of the squad as their particular skills become needed.
The back story is interesting, the characters are over the top, but very cool, and the level designs would make H.P. Lovecraft queasy. All the ingredients are there for a blockbuster game, but Jericho was poorly received, both critically and financially. I believe that the responsibility rests mostly on the shoulders of bad AI. When not under the direct control of the player, the other six members of the Jericho Squad just can’t seem to get themselves killed fast enough. They walk into explosions, shoot into the sky, and even casually saunter up to spear-wielding centurions to be harpooned. I don’t pretend to be an expert in military tactics, but I think they would have gone over those things in wizard/commando boot camp. The squad’s AI is so bad that gamers spent most of the experience simply trying to keep them alive. This detracted significantly from the game, so much so that it made an otherwise playable title into a footnote in the medium’s history.
Sloppy AI is a problem for allies and enemies alike. Oblivion, as I mentioned before, had excellent programming for most friendly NPC’s, but every enemy in the game seemed to use the exact same pattern. A stealthy assassin would charge directly at you and attack. A powerful wizard would charge directly at you and attack. A hulking ogre would – you get the point. Since Oblivion was a combat heavy game this resulted in a great deal of employing the time honored tactic of running in a circle and shooting. Now, I recognize that a game as large and complicated as Oblivion, and later Fallout 3, will have some implicit programming limitations. That being said, it’s 2009 Bethesda! I shouldn’t be able to defeat my enemies with the same strategy I used to kill baddies in the 1993 release Doom. The days of my digital enemies repeating patterns and performing predictable responses to my actions should be over. There is no glory in winning a fight against an idiot, no matter how well rendered they are.
While NPC programming has had a bumpy road there have been some notable victories in the last few years. Alyx Vance of the much-lauded Half-Life 2 broke the mold of the traditional video game heroine and provided excellent covering fire for the protagonist rather than just waiting in a castle to be rescued. Enemies have gotten better too. Despite my many problems with the Gears of War series it did an excellent job of simulating a firefight. Enemies spread out over the terrain, took cover, and even performed flanking moves to take you out. That’s the kind of AI I want to see; I want to match wits against the next incarnation of Deep Blue, not the EDSAC vacuum-tube computer.
NPC programming should challenge us, it should adapt to us, but most of all it should make us believe in it. When this facet of a game is overlooked it takes away from the entire experience, it reminds us that we are playing with a computer, and it breaks the immersive aspect of gaming that we value so highly. Thankfully, the future of this aspect of video gaming looks very bright. We aren’t far from the day when developers will start hiring psychology professors to help design NPC’s; just look what Microsoft has in the pipes with characters like Milo. It seems that soon enough it may be hard to determine the difference between video game characters and real people. For the time being, however, I will be satisfied if NPCs at least display the survival instincts of carpenter ants and don’t actively detract from the gaming experience.