Celia Alexis Wagar's CritPoints

Game Loop has become an industry-standard piece of terminology for video games. It’s taken as a default, a forgone conclusion, or necessary for a game to function. It has a role in game development similar to 3-act structure or the Hero’s Journey in storytelling. These structures are presented as inevitable, ever-present throughout history and culture, and essential to good storytelling or game-making, but many stories and games don’t follow these structures and are still successful and well-regarded.

Some people argue you cannot make a game without loops, or tell a story without 3-acts, or say that the Hero’s Journey is the monomyth from which all other stories derive, but there is nothing definitionally inherent to games or stories that necessitates these things.

Theories & Falsifiability

If we look at these things similarly to how one might regard a scientific theory, we have to ask: Is this explanation of storytelling or game design falsifiable? Is there something that could prove it wrong. If these patterns automatically apply to every single game and every single story, then why are we encouraged to intentionally create it or structure it to fit these patterns? If these things automatically apply, why should we put conscious effort into acknowledge them? If you can’t present a counter-example of something that doesn’t follow 3 act structure, or doesn’t exist in a game loop, then how useful and predictive are these theories? Rather than view these structures as inevitabilities, I think it’s more useful to view them as lenses that can be applied sometimes, but not all the time. There are already people questioning how useful 3-act structure is for video game stories.

One of the big challenges to 3-act structure is that people can’t agree on which parts of a narrative actually fit the 3 acts. The argument goes that for any given story, you can cut it into a beginning, middle, and end, but you could also cut it into a beginning and end, or into 4 parts, 8 parts or any number of parts. Why 3 specifically? Why is 3 good? Why is 3 inevitable? If 3-act story structure were falsifiable as a theory of storytelling, then we’d be able to say either, “Here are stories that are not 3 acts. They are not good.” or, “Here are sequences of events that are not 3 acts, they are not stories.”

If game loops were falsifiable as a theory of game design, we’d be able to say, “Here are games without loops, they are not good,” or, “Here are interactive systems without loops, they are not games.” If loops simply occur whether we want them to or not, how does awareness of them allow us to change anything? What makes a loop good or bad? What is this framing actually describing about a game, or allowing us to examine?

Do all writers necessarily write with 3-act structure in mind, and do writers who don’t follow 3 act structure produce good stories? If writers are able to produce good stories without following 3 act structure, and we can say their stories fall into 3 acts after the fact, then what’s the actual importance of learning 3 act structure? What does it allow us to predict?

If you look at the game loops that people present for games, it’s not always clear that they loop as cleanly as people might claim. For Minecraft, you might say that the loop is Explore > Harvest > Craft, but what if someone bounces back and forth between harvesting and crafting, and never explores? What if someone harvests as they explore but never crafts? What if someone harvests and never stops? Is this actually a loop, or is it just a list of things you can do?

If I play a Rhythm game, I could very well hit a non-repeating sequence of notes. Is this a loop, because I see note, then hit note? Is bouncing between song selection and playing songs a loop? What if I just let it autoplay songs for me? Does this mean that all songs are loops? Does this mean everything is a loop? Are stories actually loops? If we have an explanation that cannot be disproven by a counter-example, then we aren’t actually explaining anything, we have a tautology. String theory went through a similar problem, it was a mathematical reframing of physics that attempted to harmonize everything we know into a single theory, but it didn’t produce many testable hypotheses, so it couldn’t be disproven, but it also didn’t grant us any additional explanatory power over the world. Good scientific theories allow us to predict how the world works and give us a working knowledge of the universe.

Predicting & Understanding

Three act story structure and game loops, being framed as inevitable and inescapable, don’t allow us to predict what will make good stories or good games, only enumerate what’s already in them. If everything can be framed as a loop, then nothing is actually a loop, and looking at a game as a series of loops doesn’t actually help your understanding.

For contrast, I like to frame games as node graphs. I value complexity in games, which I frame as differentiated state space. My thesis is that these node graphs produce games with more differentiated state space when there are more edge relationships between nodes in the graph, and when the graph is more asymmetric overall. This allows me to predict that games with small groups of nodes that are isolated from one another, (ie. games with different modes of play with a small number of unique mechanics that do not affect one another) are complex as games that have large bunches of interconnected nodes (ie. games with a single mode of play with many mechanics that affect one another).

This theory of games has falsifiability. If there were a complex game with many isolated groups of small numbers of nodes, then it would contradict my theory. Since most games designed this way are not very complex and most games designed with more interconnected nodes are complex, I would argue that my theory largely holds up. You’re welcome to disagree or provide counter-examples of course.

I can also imagine games that contradict my predictions, but these games have many limitations. So this graph theory oriented approach is more of one method of creating a good game, rather than a definitive explanation for everything. Not to mention that it’s good as defined by my personal lens for quality, effective complexity, which isn’t exhaustive.

If we essentialize these structures as inevitable and essential, then they lose their predictive capability. If we view them as the one true way to success, then we prescribe quality to narrow adherence to a certain structure, when we know that this isn’t inherently true. I believe that these structures should be treated as lenses or templates, rather than facts of the universe. They’re a path to success, rather than the secret to success.

A Brief Disclaimer

Feedback loops are very real, and an important tool for any game designer. I have written about feedback loops before, positive and negative. It is entirely possible to build a game that structured around a feedback loop. Not every game is structured like a feedback loop; for example, Celeste, or any Masocore Game. You cannot obtain any permanent advantage or disadvantage, and everything is proportioned exactly to overcome the challenge right in front of you. Cookie clicker is designed as one massive positive feedback loop.

It’s a common component of competitive games to give out positive feedback to players that are winning to enable them to do better, as well as negative feedback to players that are winning too hard to stop them from running away with a victory. My card game, Charmed Chains, has a negative feedback loop built into the core mechanics: When one of your creatures is destroyed by your opponent, you get to draw a card.

My article is not criticizing all loops in all contexts, just this specific idea of “the game loop” as a lens for viewing all games.

The Takeaways

The real question here is, what are loops intended to describe? Loops are intended to describe timescales of play. How often am I going to make a decision that interacts with this layer of the game? Loops with only immediate action verbs are short-term timescales, also called Tactics. Loops that feature game systems with persistent or permanent payoffs are longer-term timescales, also called Strategies. Starcraft and other RTS players similarly talk about Micro and Macro to refer to moving army units and ordering workers, production, and building. Rather than trying to fit our games into some notion of a loop, I think it’s more helpful to consider how different actions in a game are tied to different timescales. Is the game satisfying on a moment to moment basis? Is it satisfying on a level to level basis? Is it satisfying considered across an entire playthrough? Are there choices and payoffs that tie into each of these timescales? When considered this way, we can tie these ideas more effectively back into my conception of depth: How do these different timescales come together to create unique experiences?

And as a matter of opinion, I think the moment-to-moment timescale is the most important one, and the most difficult one. It’s very easy for a game designer to draft up something like The Nemesis System from Shadows of Mordor. Games have had NG+ modes and other complications in them for a very long time. Long-term systemic consequences for game actions in the form of XP, stats, level variants, and so on can be balanced on a spreadsheet. However, building a strong melee combat system is tremendously difficult for even a skilled designer, and is exactly where Shadows of Mordor tripped up, which undermines the fun of everything else it does.

We also saw this type of thing happen in Metal Gear Solid V: The Phantom Pain with its “Revenge System”. Enemies would “react” to tactics you used in a level, and deploy defenses against those tactics, which you needed to send soldiers on missions to remove. This culminated in almost every enemy outpost being covered in body armor, with night vision goggles, flashlights, and helmets. While this is a neat system, the game definitely could have invested more in its core stealth systems, creating more enemy investigation patterns, and having them do more to interact with you than simply spotting you faster and being harder to take down.

These types of systems are easy to invent, easy to describe, and therefore easy to market, even if they do require a ton of human effort from programmers and artists to implement. This makes them tempting. I think that in order to move forward on game design as a discipline, we need to look harder at the root mechanics that happen moment to moment. What types of skills are we testing? What types of interactions are we stressing?

By looking at these systems as different layers of timescale, we can contextualize games like a layered cake, and we can plan around how immediate decisions affect the short term (like attacks and dodging consuming stamina), how short-term things affect the mid-term (Using up your ammo means you need to buy more ammo later on), and how mid-term affects the long-term (Spending money on ammo lead to a hike in the price of bullets eventually, and enemies wearing more helmets).

I think this model of looking at timescales is probably going to be more valuable to designers than something as abstract and disoriented as a “game loop”. If game loops are meant to give games structure like storytelling acts, then I think we should be considering whether they’re the right tool to imagine that structure.