Just as with neurology, where impairments due to injury can tell us a lot about the structural layout of the brain, dysfunction in games can tell us something about their properties. One such dysfunctional state occurs when a game is solved, or mastered, or “dead”. When we have mastered a game completely it ceases to hold our interest, and so for the master it ceases to serve any function. From this we can determine that there are two vital elements in games, at least one of which must be a source of pleasure and a motivator; the sensation of improving or gaining mastery, and the acquisition of new information – learning.
We can think of examples of activities in which there is very little to learn at the higher levels of ability, things like athletics or body building. These activities almost exclusively demand repetition and practice in order to improve, and yet remain engaging on that basis alone. On the other hand, we tend to see it as something of a failure if a game boils down to repetition and practice. When that happens it usually implies that there is little left to learn from the game in terms of strategy or high level knowledge, and we are simply drilling execution (Here I am speaking more narrowly about what might be called strategy games, this will be my focus throughout this article, I am ignoring explicit execution tests such as guitar hero). This means that the unsolved or “alive” status of a game depends on its ability to present new information to the player. There are however, different flavours of information, and I think understanding this will not only help us to tackle the dark art of game design, but also to more deeply understand the appeal and function of games.
I would like to draw a rough distinction between two types of information, systemic information and surface information. The absorption of these types of information I would similarly call systemic learning and surface learning. It is important to state at the outset that this is a spectrum rather than a dichotomy, and where information falls on the spectrum is wholly dependent on the person learning this information. I don’t think that there is anything inherent to information that could bias it towards either end of the spectrum absent an observer.
What is systemic information?
Systemic information is highly interconnected, which means that it is usually inherently more valuable and useful to an individual. This is obvious when we look at an arbitrary piece of trivia. The vocalisations of the blue whale are typically between 10 and 40 hertz. That fact on its own tells us almost nothing. The more we know about the behaviour and biology of whales, the more we know about how sound propagates in water, the more interesting and informative this fact becomes. Enough interconnected knowledge will also reveal other layers of information. We might start to speculate on the advantages of this range of frequencies and why blue whales might have evolved to make use of it, we might see how it relates to the frequencies produced by other species of whale and what that in turn implies. Because of this, as we learn more and more interconnected information, additional pieces of information can be much richer, as the relations between pieces of information are in themselves a form of information. Each additional data point reveals a little more about the underlying patterns.
In essence, the more a new piece of information can be anchored to currently held knowledge, the more systemic it is, conversely the fewer anchor points it has, the more surface it is.
Why is systemic information important?
The information that is most useful to us is systemic in nature, it helps us build mental models. Mental models allow us to make predictions, both to refine our mental models further when they prove inaccurate, and crucially, to improve our chances of survival. Our intuitive understanding of gravity is evidently very useful for our survival. It is easier to evade predators by understanding their behaviour and avoiding them, than to try to outrun them. More complex models, such as our models of the behaviour of other humans, help us in more nuanced ways. It is telling that the very structure and functioning of the human brain is so well suited to storing systemic information and complex interrelations.
Games are all about mental models. As with any system that we are trying to understand, we make and test predictions based on our models, these predictions lead us to success or failure, and we update our models accordingly. Games are like toy worlds in which we apply these very real skills, and from which we get much clearer feedback than we often receive in reality. This is the unique and central property of games, other art forms can tell us about systems, or ask us to think about systems, only games get us to live inside and internalise them.
Surface information and systemic information are not mutually exclusive, surface information can always be a data point that reveals deeper connections. However, there are different types of trivia that are more or less likely to be useful in this fashion. The latest gossip about a reality TV star says much less about the world than the results of a new study in a journal of psychology. The properties of an item in Dota say a lot less about the game as a whole than properties of the creep spawn system. I think also that specific details are much less important for systemic learning. The value comes from the intuitive knowledge and predictive power we gain from this knowledge, the purpose of gathering specific data points is to nudge our models in one direction or another, we don’t need to actually remember the finer details. This is actually a great form of data compression when you think about it. There is however a certain appeal to surface information, and this seems to come from the specificity. Think of the interests of collectors, or the stereotype of the star trek nerd. There is an appeal to being very detail oriented, to remembering a lot of very specific facts, regardless of how interconnected these facts are. This is more speculative, but perhaps there is also survival value in this kind of detailed information, in those situations where new information does contradict our mental models in surprising ways it would make sense to be fixated on details.
“Great minds discuss ideas; average minds discuss events; small minds discuss people.”
I quite dislike that quote, but it does roughly map the spectrum. Ideas are vague and interconnected, and of much larger scope. Events and people are more concrete and detail oriented. In games,the underlying systems are vaguely understood, but far reaching, finer details are more specific and quantitative. I think this is also part of why most people find it very hard to talk about game systems (no, not because they are small minded), these aspects are almost inherently vaguely understood. It’s much easier for Street Fighter players to talk about how many frames a certain hit-box lasts than to talk about tempo and spacing.
There is value in both types of information, although I think there tends to be a lot more value in systemic information, however there is also the factor of relevance to consider. Either type of information can have value in the abstract, and yet be completely irrelevant to us. This would be information that does not help us survive, nor does it fall into the purview of any of our areas of interest. This information tends to simply strike us as boring or uninteresting, but I think the nature of surface information sometimes slips past our relevance filter. I think there are two reasons for this. The first is pragmatic, if systemic information slots into larger structures, like mental models, we cannot dismiss new information until it is clear that it is not relevant to these models. Therefore at least some checking has to be done. The second reason is that our biology, as the messy product of evolution, cannot cleanly control our actions, it can only loosely couple certain actions with reward or punishment, either a dopamine hit or pain. So for us to want to gather information, it is likely somewhat pleasurable at a very low level in the brain. This makes information gathering a motivator, part of the drive that causes us to play games. At this low level there might be very little difference between relevant and irrelevant information, it would all produce some slight amount of pleasure. At a higher level some other processes, or as a last resort our own capacity for reasoning and judgement, might produce a predilection for relevance.
This largely indiscriminate, unconscious approach must have made perfect sense in earlier times, when there was only so much information to be gathered, and much of it would be about your immediate surroundings and therefore useful for survival. The factor of relevance may not seem too … relevant to the topic of game design, but it is something to keep in mind. Players will absorb any information the game offers up, whether or not it is relevant or useful for them.
Concepts such as depth are higher levels of abstraction above this concept of systemic information. A deep game will offer many opportunities for learning systemic information. Presumably, as players increase in skill, they are gaining a deeper understanding of the game, and the information they learn becomes more systemic as they have a greater number of anchor points to connect to that information. This means that, ideally, as they get better at the game the information they receive actually becomes more interesting and nuanced. In reality that value will likely taper off. When playing games we also care about being successful, as this is the proof of the accuracy of our mental model. There must come some tipping point where the amount of information needed to meaningfully improve our mental model outweighs its interconnectedness, and the value of new information, relative to some increase in success, starts to decrease again.
Another consideration, although I am now deviating from the topic somewhat, is that when increases in subtle systemic knowledge have a large impact on success, it is likely that the earlier, less systemic learning had a relatively small impact on success, and the game is therefore very frustrating to new players. On the other end of the spectrum, the game that quickly rewards the earlier, less systemic learning, is unlikely to greatly reward very subtle improvements in understanding later on, and therefore will lose players earlier on their path to mastery. It is interesting to consider ways to support both extremes.