Ben Gonshaw: Digital Media Theorist & Game Design Consultant
Home Latest Updates Design Rants Musings Links Comments

This page will look wrong in IE. Use FireFox for better viewing

EMERGENCE: Gaming's Saviour, or False Hope? Emergence and Games

Game developers have got carried away with emergence as an exciting principle and tool. It is a very enticing concept, order for free, little work with huge results and an adaptive, reactive, dynamic system that maintains its cohesion and begets further systems and complexity.

To the pseudo-scientists and technically minded, it is intoxicating to imagine the similarities in form and function. From atoms up to the universe as a whole, everything in mirrored harmony with each level of complexity dependent on those below it. Each one superficially similar in stasis, dazzlingly diverse in motion, with fabled golden-ratio Fibonacci sequences spiralling off to infinity.

Has our obsession with exciting mathematical constructs made developers lose their judgement and forget that the purpose of such systems is to make fun entertainment?

The Promise of Emergence

Emergence promises both to help developers in creating better game experiences and to help the player in allowing them to play however they want to. In the past, when methods and processing power could not support complex systems, the player was always in conflict with the (mechanical) boundaries of the game, often pushing against them in directions that the developer had not thought of. The results were that either they 'broke' the rules of the game and it behaved oddly, or more likely, a strange artificial restriction kicked in and prevented the player from doing what they wanted to do. Complex systems promise to give the player the freedom to experiment without the game telling them that they cannot. This is especially useful for those who do not have much experience in playing games, because they can pick something up without knowing the artificial boundary problem and just play. The game will stilll respond to whatever they do instead of forcing unnatural barriers on them or just getting them killed repeatedly.

Visible Consistency

Games have a thin veneer of reality that allows a player to believe in it. The more 'realistic' you make the game, the easier it is for the player to do something that breaks their suspension of disbelief. As a completely abstract representation the player will understand if some blobs move towards you when you approach from the left, but not if you drop down from above. In a more realistic setting it seems odd that you passed the trigger point by taking a different route, but the men stand there with their backs to you and let you pick them off one by one.   More believable team tactics can be offered through emergence, more exciting accidents and traffic flows, better herds of wildebeest (useful for those countless savanna simulators), believable crowds that run or gawp realistically when there is a shooting or a streaker, instead of just standing around talking.

Improved AI

Programming an AI adversary is sometimes a difficult task, especially when there are many permutations. For example, in an RTS, the AI must be able to cope with many different strategies developed by many different players. Rather than make the opponent very strong numerous and dumb, a more satisfying experience is had against an inventive opponent. However, creating an AI that attempts to do the same job that the player does is incredibly difficult. Instead, an opponent could be made by giving each unit some basic rules, and letting them roam the map. With enough units they can spontaneously develop cohesive tactics.   This also paves the way for interesting new gameplay that could turn the tables on the traditional enemy horde. By making the player strong and numerous, but making the AI few and tenacious, you have an enemy that creeps up on you where you least expect it and behaves cohesively, rather than lumbering behemoths that outnumber you five to one.

Hidden Adaptivity

Under the hood, games can employ diverse complexity-based algorithms to aid enemy spawning, the placement of health packs and bonuses to throttle the difficulty of a game in real time.   Behind the scenes off-line, many simple agents, paired with genetic algorithms, could be used to generate levels, like the termites create their nest, or in the same way that the Game of Life flexes and homes in on a static state, so buildings and sceneary can crawl into place within the level.

New Gameplay Types

Physics promises much and right now, and although very little of interest has yet occurred, with the help of dedicated physics processors it may come to deliver. Combine this with clever logic creation and there is no reason why players cannot have combinatorial weapons and game types. Allow the player to create new types of weapon, or robot companions and the gameworld will be able to respond and adapt appropriately. Allow the player to create swarms of intelligent co-conspirators and loose them on the world and the world will have to respond appropriately. Adaptation is a key quality of complex emergent systems and games where the player pits their wits against an immune system or reversed, against a virus-like adaptating intruder, becomes a real possibility.

The Current Reality

 

Hinderance Not a Help?

Having convincing traffic behaviours in driving games certainly makes the world look more believable. However, does the other traffic add to the tension, or does it destroy the game? Sometimes traffic is in your way and sometimes it is not, meaning that your success is not dependent on your skills, but it rests arbitrarily on the current state of the system. This can be extrapolated out into other genres, when unexpected events which are products of the system put paid to your well planned actions.   All this emergence may help you to believe more in the gameworld, but if it makes you wipe out on the last corner of a time trial, it may cause you to throw the controller through the TV as a result.

 

Tweaky Tweaky

Programming emergent systems is not too complex a task in itself, but honing the system so that it does what you want it to do can be extremely difficult. Inherent in the nature of emergence is that the system produces things that you do not necessarily expect. Some may be harmful to the game, some may be undesirable, whilst a scant few might be useful. However, retaining the desirable outcomes and screening out the unwanted ones is especially hard, as even the desired behaviours will change when you try to alter the fundamentals of the system. This balancing act is made more difficult because the underlying rules may be correct, but the positioning of the agents may be wrong, preventing the correct behaviours from being seen. Working out which is which can also be a difficult trial. Try messing with the Game of Life to see that even though the fundamental rules remain the same, if you place the agents in bad slots almost nothing happens beyond 10 generations.

From a designer's perspective we don't want a precarious unpredictable world, because we don't know if it will still be there to play in after 40 minutes. If the world has a fairly stable complex system without player intervention, then that is fine, albeit nightmarish to attain. However, when the player enters and starts shaking things up, it's incredibly hard to build something robust enough to cope with what they can throw at it without collapsing. Think of it as an overfished ocean eco-system that just ends up full of plankton and no higher life forms. To counter that, either the player's goal must be to maintain the system or you have to cheat by constantly respawning or repositioning agents to try to keep the system balanced. If you have to cheat to make it work, you might as well cheat properly, script the whole lot and save yourself from the Sisyphean task of trying to tune the system in the first place.

 

Spotting Emergence In Today's Games

So what examples are there of emergence in current games? Very few, as there are almost no games that supply the necessary complexity to support emergence. Many that claim to be emergent are actually reliant on much smoke and mirrors to fool you that there is something more going on, when actually it's highly scripted. Play through it a few times and you work out how to get by those pesky scripters. Grand Theft Auto is heralded as having emergent gameplay when it actually has heavily scripted agents. There are few emergent circumstances to speak of. There is no living world and in its place is a supremely amnesic sphere of reality.

 

Even so, for argument's sake imagine that the game you're playing does have a sufficiently complex system within it to support emergence. Games will only support truly emergent self-organising generative gameplay when they have higher order systems within them. Right now we are stuck at the level of one flock of boids. You can spend time forcing the flock to break apart, reform or change direction, but that is about all. When gamers have the ability to make a new flocks and flock behaviours, or they can follow the interactions of 60 different flocks, then the higher order interactions will feel more like they are emergent.

 

What do you mean, feel more emergent?

 

Well, we are used to seeing emergence as the grand play of squares marching around a grid, or in the excitement of a shoal first forming. However, if there is already a flock and you put a lamppost in the way, even though the flock moves around the obstacle like a liquid this is an expected result of the system, it is not as visually amazing or as surprising as seeing a system created. This means that we might well miss true emergence because it doesn't have all the whizz-bang hallmarks of that first shoal's birth. It'll quietly sneak up on us and head right past us and we won't be any the wiser after it has gone. In a sense that is when emergence will have made it, when it's so good, so responsive and the results so natural, that we simply take it for granted.

 

 

Back to Rants