Emergence Room

As I already pointed out, I see a huge value and especially a unique trope of game stories in emergent narration (also often dubbed “player stories”). Many developers, both AAA and Indie, try to build on the idea of highly systemic games that have no set embedded narration, but live through their story emergence. On 2014’s GDC, Irrational Games’ Ken Levine talked about a systemic game design concept at the core of a fictitious emergent narration game. From the bottom of my heart I appreciate such approaches and look forward to the narrative possibilities, as I see the future of game stories in this kind of storytelling – the stories spawned are always meaningful to the player and are driven by his or her efforts to play the game (being a storytelling device no other medium can yield).

The goal of emergent narration, as I perceive it, is the following: We want the players to encounter the same amount of thrill and quality in their emergent stories as they experience through linear media – not through telling them but through having them experience these stories driven by their own actions. I order to do so, we have to craft believable ecosystems in which the player can act as he or she pleases, so that each and every combination of actions creates meaningful and exciting events. These of course need not be pre-written but rather system-emerging, yet they have to convey powerful emotions still.

And at the moment I see a couple of problems with that. Not that I doubt that these achievements are possible, but I see the problem rather in the approach than in the idea. The issues I perceive can be summarized as follows:

  • Usability: If the player has great freedom, he or she needs to know where the borders of this freedom are.
  • Loss of embedded content: If we embrace the “player story” so much, we endanger ourselves to lose the embedded story instead of combining them to get the best of both worlds.
  • Consistency: Especially in games with randomized events (or through player disregard), stories or combinations of events can easily appear hilarious or unfitting.

Luckily, these are not unsolvable problems:



The usability problem I see in systemic games with high degrees of freedom is that the player might easily get lost in his or her possibilities. Additionally, players might not know what exactly they are capable of (so what the elements of the game system are) or in what degree their actions affect events and NPC behavior.


This is the easiest problem to solve, as communication is usually the key to it. If the player gets informed through a tutorial or a less classical approach to teaching him or her the game mechanics, the confusion can very easily be overcome. The player needs to be familiarized with each and every component of the game (but not with their recombination, as finding these out is the core of most systemic games) in order to make meaningful decisions.

The second part of the problem can very well be addressed through feedback. If the player gets an indication for which actions get noticed by the system and which are mere tools to hold the game elements together, he or she can decide on relevant game actions much easier. If, for example, the PC steals a car and parks it right in front of his or her spouse in order to impress them, then the player needs to know if the system recognizes this or just disregards it as a parked car with no tie to the NPC. Without such feedback, the player may easily feel betrayed or has his or her immersion into the game world broken as the freedom’s instability shines through. Having the player always have a look onto the game system makes things much more accessible for him or her, even if that means integrating HUD or abstract menus into the game (this will break the immersion in a far smaller degree than the consistency break of the system not recognizing the player’s actions – this is part of knowing that games are just exemplary depictions).

Loss of Embedded Content


The strength of pre-written content is that a distinct person or team can care for its quality and progression. If we exclude this part of game development from a game in order to completely embrace the emergence of systemic game design, we endanger just inverting the problem. When every game element is randomly constructed and consists just of the player’s interaction with the system, the overall experience can very easily feel dull, repetitive or artificial.


This one-way result can be prevented if embedded narration is not demonized but instead though of as a tool at the hand of the designer and writer, capable of improving the liveliness of the game world. Let me give you two examples how this combination can be achieved.

Example 1

Let’s take the theoretical game Levine describes in the talk linked above – an open world Fantasy RPG (with fighting!) with different villages (dwarven, orcish a.s.o.), where each NPC’s sympathy towards the PC depends on the sum of about three individual dispositions, each representing a passion (e.g. “hates elves”, “wants a temple for the old gods”, etc.). In that game, to create player incentives, you could hide a number of legendary weapons in the game world. Each of these weapons have a distinct, pre-written legend to it, describing its origin, powers, previous owners and eventual loss to history. This enables the game to randomly decide the exact whereabouts of each weapon and their connection to the game world.

These connections could be NPC’s passions or backstories, which all are pre-written (but can also be randomly chosen from among a variety upon each playthrough). This way, each NPC is created with a number of pre-written plot hooks that can connect him or her to the treasure (e.g. family inheritance, secret hideout places, treasure maps etc.). Per plot hook, any part of the world’s lore can be connected to it, the legendary weapons being among them. Besides creating motivation for the player to learn about the NPC’s passions and backstories and to befriend them, this also enables the game writers to contribute to a believable game world without taking away any agency from the player.

To get this example a little bit more specific, let’s assume there is the legendary Sword of the Lion in the game world that causes earthquakes when rammed into the ground. The Sword’s origin and story can be traced inside of the game (e.g. through library books or historians), and these sources can also contain hints on its wherabouts. But one player will find this sword through befriending the Orc Smith who found a secret diving spot on her favorite fishing site, while another player might dig up the sword in the cellar of his just-married Dwarven husband, and yet a third might come across the sword atop a magical tower that she gained access to through eliminating a tribe of wood fairies that magically hid the tower from human eyes.

Example 2

Instead of adding pre-written content to the game, instead a pre-written structure can be applied. To pick the easiest, let’s assume you want to spice up your emergent narrative through topping it with the three-act-structure. Though remaining completely emergent and player-controlled, the game will thus lead the player through 3 different (st)ages, where the game elements of possible decisions (and, depending on the game, random recombinations) vary in degrees of epicness, tragedy or danger. This could also mean a different context for each of the stages (beautiful summer, harsh autumn and deadly winter, for example) that forces the player to rethink each of the available game actions and elements as they are set in different contexts.

Hence, a general structure and possibly also a distinct ending (e.g. through time pressure) can be put on a systemic sandboxy game, helping to forge a story with a type of arch and progression we can relate to as it is akin to the linear types of narration we are used to from entertainment media.



The core problem I experience with the immersion of emergent games is the feeling of some game elements or events being hilarious, inappropriate to a certain situation or presenting an anticlimax unaware of previous events. This can stem from ignorance inside of the system or from pure negligence from the player’s side (which isn’t much of a problem in a singleplayer game, but can destroy other player’s immersion in a (massively) multiplayer environment). Especially Maxis’ “The Sims” has managed to present some really puzzling occurrences at some point, like a llama-dressed college promoter at a wedding party.

I am aware of the fact that this is in fact only a problem of games that integrate pre-written content into their system. Purely systemic games (like Civilization) do not have this problem as the choice of game actions is abstract and self-restricting enough that inconsistent choices are nearly absent.


To address this problem, one has to differentiate between the two named sources of inconsistency: Player unpredictability and interactions of game events. I will illustrate each of these with short examples.

Player Unpredictability

Please note that there is no point in blaming the player for performing these actions, as the utter ignorance of the game for certain actions can yield a great deal of fun, which is a valuable part of emergent narration that should not be ignored. Instead of restricting player freedom for the sake of sacredness of the art, I advocate playing with the unpredictability of the audience. In order to keep up consistency, integrating the player’s possible ambiguity to the narration and having such events have consequence is a much better solution in my eyes. The Elder Scrolls Games, for example, do not punish the player for running around naked (other than the mechanical loss of armor protection) but instead have NPCs notice this fact and talking to them about it. This weaves another player action into the game by simply adding a new line of idle dialog, but with the advantage of increasing immersion and meaningfulness while still enabling the player to do as he or she pleases (and mocking his or her surroundings). The same can also be done for constantly jumping or dancing characters or other awkward actions.

As for more narrational moments, a player leaving his or her own wedding should not have the events wait for him or her to return (or just reset the level), but this player decision needs to become part of the game, too. To not having to write (and record!) an adequate line per NPC for each situation, each game event could simply have a termination scenario that triggers when the player actively ignores or destroys events. Imagine, for example, a burning house the player comes across. Should the player decide to ignore the house fire, the inhabitants die (instead of waiting for the character to come back for them). I know that such design decisions have been abandoned previously in some game’s designs on purpose so to not force the player to do things, but it contributes a huge part to the feel that the player’s actions have a real impact on the world – and can also keep up replayability.

Also, events could even build upon the player’s unpredictability. In the above wedding example, how about there is an assassin present at the wedding that can only be spotted and hindered when the player leaves the ceremony? This is an example that should not trigger always, of course – these events need to rely on previous player actions or random factors in turn, according to the game’s design principles. Correspondingly, what about a certain treasure can only be obtained from the burning house when the endangered inhabitants are ignored and not rescued?

I do not argue that a game should have an answer to every possible player action (as this would be an unachievable task to accomplish) – but it should provide backdoors or reactions for each huge set of player actions (like a general termination scenario).

Interaction of Game Events

As I already pointed out, games like “The Sims” sometimes suffer from the inaptitude of game event interactions. And though Maxis’ game manages to deal with this quite well through the game’s intrinsic humor and irony, other games may not perform so well when dealing with randomized content or reactions.

In order to prevent game events from taking place during an inappropriate moment, a game could take advantage of assigning different moods to the current situation. This would limit certain random or player-kindled events to only become active during one of certain different moods. Hence, the random man in a chicken costume is hindered to break the funeral of a deceased child, as his “neutral” and “fun” moods do not fit the currently active “tragedy” mood.

An alternate solution would be to work on the transitions of game events. Like dominos, randomized content or AI decisions could have certain IN- and OUT-labels. Following this, an event with a specific IN-label can only take place right after or during a certain OUT-label event in its direct vicinity. This could also help serialize events to build upon each other and support the solution of rising tension curves mentioned above.

The most important remark on these “events” I talked about is that they should never stand just on their own. They can be used to re-contextualize the game environment, provide feedback to player actions or present incentives, but they must always be connected to other systemic game elements in order to keep up the consistency.


To sum this long post up, I wanted to help identify common narrational problems for emergent games and propose solutions to them, as I see emergent narration to be the future of interactive storytelling. I hope this post might help developing a distinct approach for this highly-acclaimed trope.