Multum in parvo. It’s a Latin phrase that I feel encapsulates the concept of emergence. Complexity borne of the combination of simple things. Linearity in games can offer the benefits of structure and focus, but it comes with a loss in novelty after being experienced once. Linear games, while respectable in their own right, also tend to lack offerings for replayability. Many games nowadays include multiplayer to extend longevity, but that is only one solution to this problem. Creating scenarios and game mechanics where emergence can flourish is a seldom used alternative solution, and if it’s applied to single-player content it can still function independently of the accessibility of servers or the solvency of the game’s developers. But before I highlight exemplars of emergent gameplay, let’s quickly go over some early history of the concept.
John Conway’s cellular automaton the Game of Life could be considered a distant forebearer of emergent gameplay. Admittedly, there wasn’t much playing of games to speak of (as the word automaton would suggest) since it was entirely deterministic, but the Game of Life was one of the first programs that could consistently generate the key component of emergent gameplay: complex behavior that emerged from interactions between a few simple rules. By merely establishing initial conditions of whether particular elements of a grid (called cells) were on or off and then following the rules prescribed by the game, we could generate “creatures” like Gliders and Oscillators as well as more complex systems in the form of Guns, Spaceships, Eaters, Puffers and Breeders.
Fast forward a bit, and we see video game developers begin to implement artificial intelligence into their games, usually by way of establishing behaviors for enemy characters. One of the simplest forms of this could be accomplished by scripting, where specific entities would follow rigid and often unchanging behaviors or patterns. Scripting could work for the very first time a player encounters a situation, but players could easily break or exploit scripts with a little experimentation. (We still see this happening in modern times, such as in the ability of the player to bypass certain enemies in Mirror’s Edge while they are performing their entrance script.)
So, developers endeavored to make behaviors more intelligent and less artificial. One of the ways they accomplished this was by making their scripts more general. Rather than forcing a specific entity to go to a specific location when a specific trigger in the code was activated, programmers could just write and combine small procedures. When they avoided scripts that were defined in absolute terms, they could write code that would, for instance, instruct entities to take cover or run away or flank an enemy or toss a grenade and so on based on the changing circumstances of a battle rather than due to some deterministic view of how it should happen.
Developers could use more general scripting to benefit allies as well. I think Mercenaries: Playground of Destruction and Overlord serve as particularly good examples of this type of AI design. In Mercenaries, you could use members of a faction friendly to your cause as back-up or as simple cannon fodder (not unlike the UNSC Marines that appeared in Halo: Combat Evolved, which was released before Mercenaries), but things really got interesting when the dynamics between each of the in-game factions came into play. The Chinese hated the South Koreans and vice versa, the Allied Nations focused on systematically weakening the North Koreans while protecting innocent bystander Civilians, the Russian Mafia was comprised of opportunists who interfered with everybody and the North Koreans hated every other faction and were likewise hated by all. There were a number of scripted scenarios where opposing forces would duke it out in epic battles, but some of most interesting interactions were unscripted and happened when belligerents caught wind of each other while simply driving or flying around. Factor in the ability of the player to disguise themselves as part of any of the factions if they are inconspicuous in an appropriate vehicle, and you can see how the possibilities for manipulation, mischief and mayhem begin to mount.
Overlord, which also encouraged mischievous behavior but lacked a speculative history narrative, was essentially a Pikmin game with a Tolkienesque high fantasy setting, gremlins replacing the titular Pikmin and a mute Sauron-esque protagonist replacing Olimar. Oh yeah, and there’s a jester that you could kick as well as a gnarled old advisor gremlin called… Gnarl. In Overlord, a game mechanic involved the gremlin minions automatically picking up armor and weaponry as they stumbled upon it after defeating enemies or destroying the environment. Defeating bosses and minibosses sometimes generated special and unique weaponry or armor, which minions could pick up and use just like any other piece of equipment. It was here that two simple concepts, equipping minions and the provision of special equipment drops, combined into something more than the sum of its parts.
Generally, the gremlins are indistinguishable from one another due to them all wearing roughly the same equipment, especially since numerically superior items automatically supplant “inferior” equipment (even if the result is not particularly aesthetically pleasing). This changes when a minion picks up a special piece of equipment, as they become unique by virtue of being visually distinguished from the rest of your gremlins. Additionally, their special appearance reminds you of your previous conquests and provides a great continuity to the narrative that you’ve been carving out for yourself as you know exactly how long this minion has been with you and exactly what trials the two of you have faced together. Players can become emotionally invested in these special minions (especially since their unique equipment tends give them greater strength and make them the de facto leaders of your gremlin horde), specifically going out of their way to cherish and protect them even though such a thing is never demanded by the game. Consequently, the bitter taste of defeat can be all the more worse when you feel a twinge of sadness and regret over losing a favorite warrior, a sadness that is never felt for the legions of non-special clones that will inevitably rise and fall over the course of the game. Fans and developers apparently loved the uniqueness aspect of minion management so much that the ability to name and resurrect favorite minions was implemented in Overlord 2. I can certainly understand why, as I often felt more invested in my distinguished minions than in countless VIPs, tacked-on family members and love interests from various other games.
While the implementation of general procedures for AI is one of the most common forms of facilitating emergent gameplay, it’s still sometimes rather rudimentary and far from the touted marketing blurbs of “You’ll never play the same game twice!” Even so, modern developers have made great progress, with open-world environments (a la the sandbox games of the Grand Theft Auto and The Elder Scrolls franchises) facilitating interactions that are scripted enough to provide a specific narrative, but which include enough procedural generation to make experiences unique by reacting to the differing play styles of each person.
This is all closely related to open-ended decision making and problem solving, and thus any system that includes many manipulable variables and entities facilitates emergent gameplay. Though sandbox games are a popular form of such a system, even games from other genres, like SimCity or Dwarf Fortress, can result in complex interactions between their various components.
Often, unscripted emergence can take the form of exploiting glitches or bugs, with sequence breaking being one common result, or can even simply be borne of scenarios that weren’t accounted for by the developers. One of the most infamous examples of the latter was seen in World of Warcraft’s Corrupted Blood plague. A simple infectious debuff spell cast by an isolated boss enemy turned into an epidemic when players’ in-game avatars began to act as vectors for transmitting the disease to populated areas. Normally, the damaging effects of the spell would go away when the player character died or when the disease wore off over time, but two major details exacerbated the problem. Players of the Hunter class could use pets that could be summoned and unsummoned at will (including in the area where the boss dwelled). If a pet was unsummoned after contracting the Corrupted Blood debuff, the disease wouldn’t disappear by clocking out (as originally intended) and would instead be put into a kind of temporal stasis. If that same Hunter later resummoned that same infected pet in a populated area, the debuff would snap out of temporal stasis and the pet could be used to to inadvertently (or maliciously) spread the disease to members of the populace.
That alone probably wouldn’t have resulted in a large-scale epidemic, as informed players could at least try to steer clear of mobile vectors or infectious hotspots. The real problem began when it was discovered that some NPC characters could not only catch and spread the disease, but that they do so without suffering its negative effects (aka being asymptomatic carriers). Players could, to at least some extent, avoid interacting with other players, but avoiding NPCs was far more difficult as they were responsible for a number of services like quest-giving, armor repair, skill training, buying & selling goods, item storage and renting methods of travel, all of which were important functions used by many people a day, and often multiple times per day. In areas where the plague had taken root, skeletons of players that had perished littered the streets and surviving players abandoned the place in favor of less fatal locales like the Western Plaguelands. Blizzard and various players tried to impose voluntary quarantines, but a mix of confusion, skepticism and griefing caused it to be ineffectual. Eventually, quick patching and server resets were used to end the plague.
Although many players suffered inconvenience or loss due to the plague, I believe World of Warcraft was made that much more interesting by this large-scale emergence of a complex result from simple and disparate mechanics. I’m not the only one who thinks so, as the incident (particularly the parallels between real-life and in-game responses to epidemics) attracted the attention of real-life epidemiologists. Developers at Blizzard even went on to deliberately engineer a plague with similar behavior as part of the preparations for the release of their Wrath of the Lich King expansion, but it involved more tight scripting and thus I don’t think it’s quite as notable as its source of inspiration.
You could chalk up the Corrupted Blood plague happening due to an oversight on the developers’ part since the plague and its fallout wasn’t intended. But can the seeds for emergence be intentionally planted by developers? Of course they can, but the scope of their impacts has been largely limited due to the complexity involved in not only creating mechanics that can result in emergence, but doing so in a way that the interactions don’t completely break the game. Fortunately, EVE Online can serve as an illustration of deliberate effort to promote emergence.
Though EVE Online includes some NPC-generated missions and a heaping of lore to establish a narrative, many of the interactions with living and artificial entities is player-driven. This is definitely intentional, as the developers specifically chose to make the game unsharded and run on a single server (to maximize the number of players that can interact with each other) as well as to empower players with the ability to establish corporations and alliances, manipulate the economy and monopolize territory. Players, if clever enough to succeed, are freely permitted to establish mercenary groups as well as engage in heists, extortion, embezzlement, fraud, racketeering and piracy. It’s a cutthroat universe, but the paucity of prohibitions and the ability to form effective organizations with an abundance of finances and personnel allows for some truly spectacular events to occur.
One famous example involved the group known as GoonSwarm managing to rob and later dissolve arch enemy Band of Brothers with the help of a defector, but my favorite story involves the infiltration and eventual dismantling of a corporation known as Ubiqua Seraph by members of the mercenary group known as the Guiding Hand Social Club. The GHSC was tasked with killing the CEO and delivering her corpse to the client that wrote the contract. The group spent months planting operatives and gaining the trust of officers and even the targeted CEO in order to gain access and prepare for the killing blow. The momentous day came, with one particularly wily operative managing to convince the CEO to pilot a very valuable ship, and the GHSC accomplished their goal while getting more than they bargained for. Simultaneously, members of the GHSC engaged in battle with the CEO’s ship while others pillaged and embezzled goods from Ubiqua Seraph bases. In the end, the CEO’s ship was destroyed, the corpse of the CEO herself was successfully recovered and billions of ISK (the in-game currency) in property damage (combining assets stolen and destroyed) was racked up. Some condemned the act as morally reprehensible in and out of the game while others were impressed by the expertise employed and the scope of damage inflicted. Many games are restrictive of PvP interactions and uphold player rights much more strongly, so I contend that EVE Online is one of the few extant games where something of this magnitude could happen, and I also contend that it is the result of multifaceted structure emerging out of simple individual components of weaponry, ships, equipment and guilds.
All of the aforementioned games share a common thread in that they facilitate or even encourage emergent gameplay while at the same time including narrative elements that are rigidly defined. Do any games exist without this limitation? We need only to look at Spelunky, a hybrid of roguelike and platformer genre gameplay. As is typical of roguelikes, it features permanent character death as well as the ability to interact with the environment in a variety of ways. There are a select few fairly rigid constants in Spelunky, like the presence of four thematic areas always occurring in the same sequence and containing four levels each, as well as the final boss and subsequent ending, but overall it is one of the least rigid and most replayable games I have ever had the fortunate to encounter. The aim of the game is to collect treasure and dive deeper and deeper into the cave system in an effort to acquire a particularly large piece of treasure, but large parts of the environment (including the locations of treasure and equipment) are algorithmically generated. If that wasn’t enough, true to form of the roguelike genre, the player, the items, the enemies, the traps and the environment itself can interact in a rich variety of ways. Additionally, even the game’s introductory text is randomized to a degree. It’s one of the rare few games that approaches the “every game is different” asymptote, even if the narrative of each session is likely to end in the death of the player character.
This is but a brief foray into the subject of emergent behavior in games, not even touching upon Deus Ex (a game that popularized the concept of emergent gameplay) or countless other MMOs, roguelikes and games of other genres that offered and continue to offer so much in the way of possibilities for emergence. Nonetheless, I hope I have enlightened you somewhat, and I hope that you, like I, can’t wait to see what the future holds for emergence gameplay.
Note: Every image except the Dwarf Fortress one was was taken in-game by the author. For the in-game images, the author claims Fair Use under United States law for educational purposes.