Our goal as players is to understand a game so well that it ceases to be fun or surprising. The goal of the game is to resist our thoughts in such a way that this does not happen, while also not being completely incomprehensible.
Surprise comes from that which we did not predict, the perfect player has no such lapses in predictive power. The player seeks ever more predictive power, he builds ever more nuanced models of the game in his mind and refines them through his experience.
We should want the players’ failures to be failures of prediction rather than failures of knowledge. A failure of prediction is a new or novel situation, or outcome of his actions, that the player has not yet encountered, but which is governed by the rules he is already familiar with. A failure of knowledge is the result of rules that the player is not yet aware of, a failure for which the player cannot possibly be held responsible. Games should endeavour to teach the basic rules, and even much of the content, in as painless and quick a fashion as is possible. Learning the base rules of a game is not interesting. What is interesting is discovering a new wrinkle in a higher order understanding of the emergent properties of a game.
The need to understand may explain some of the allure of skinner boxes. In a low level, subconscious way, we feel the need to find patterns everywhere, this is how we gain information about systems. Skinner boxes however, have no deterministic properties and so our subconscious is stuck forever looking for patterns in the noise. This may also imply that, on a neurological level, our brains are rewarding behaviour that should lead to knowledge of a system, rather than rewarding the actual acquisition of knowledge. Since engaging with a system and looking for patterns in it typically leads to knowledge, that behaviour is rewarded, even though in the case of a skinner box there is no such knowledge to be acquired.
“reset to neutral” – this is fighting game terminology that captures the statelessness of a lot of games. There may be a lot of states in these games which match the opening state quite closely, aside from some numerical resources having changed value. There is no hard distinction between stateful and stateless. Rarely do these resources have no connection to the players’ options or choices, and these effects can range from very marginal to quite extreme. Chess is a strongly stateful game, every change in numerical resources is actually the removal of a piece. As each piece affects the board in large ways, and offers a lot of possibilities to the player, there cannot really be any “reset to neutral” in chess. Each captured piece changes the game-state massively, and in a way that cannot be reversed.
When a game is about learning, about gaining understanding of the game systems, this understanding comprised of mental models, we have to ask what these models represent? And what is it that we are trying to learn? Are we trying to find the underlying rules of the game, obscured as they may be in a piece of electronic entertainment? Or are the underlying rules quickly learned, and the deeper learning based on the ramifications of those rules? Put simply, is our learning trying to look deeper into the emergence stack, or are we building the stack higher?
This leads to a terminology that is reversed from the standard game design terminology. As the emergence stack intuitively builds upwards, “deep” understanding or knowledge of a game system is rather “high”. The most basic understanding of a game, provided it uses this model of simple rules leading to emergence rather than the competing model of an intentionally obscured ruleset, is actually the “deepest”. This understanding contains the lowest level of emergence, ie; no emergence. This is bedrock, the fundamental physics of the game world.
We can instead use the terms “low-level” and “high-level”, to better reflect typical parlance. A low level understading implies both that it is at the bottom of a stack of emergence and that it is somewhat basic and mechanistic. High level implies, as with programming, a less concrete, more abstract, but in many ways more powerful and useful understanding.
We could use the term “mapping” for the kind of interrogation of a system that is used to find the underlying rules, this is uncommon in board games where the player is typically told all of the rules up front. In a videogame these underlying rules can be hidden beneath layers of emergence, and are rarely explicitly communicated to the player, so the player must discover them through experimentation. We could use the term “shaping” for the process of building and refining high level mental models in order to reduce the reliance on simulation. I use “mapping” for low level exploration because lower levels of emergence are objective and so there is the sense of tangible terrain being discovered. “shaping” seems appropriate for higher level exploration because higher levels of emergence are biased towards achieving a goal, and are more divergent between players due to differences in personality and experience. This subjectivity gives the sense of something being created rather than discovered.
Content based games rely on particular arrangements of elements to provide novelty. Particular arrangements provide opportunities for systemic learning, but do not demand it. Mastery of a particular arrangement does not imply mastery of the underlying systems. Brute-forcing a solution to a particular situation can be a valid approach. System based games, on the other hand, should make it clear to the player very quickly that they are not seeking a single solution to a problem, but rather a set of tools for solving multiple problems.
The opposite problem may arise in system based games – degenerate strategies – that is, one, or very few high level strategies which are universally applicable. We want the player to have many different tools for solving problems at different levels, not just one tool for solving the highest level problem. Degenerate strategies appear when some level of understanding, one particular model at the top of the emergence stack is far too important, and a general strategy that deals with this level of emergence can safely ignore all lower levels. This means that the fine details of the arrangement of elements is not important enough, and so the player does not have to understand lower levels of the emergence stack. We might think of this as the smaller details being cumulative rather than interfering in a chaotic fashion.
Brute-forcing and degenerate strategies are undesirable because they do not require much systemic learning. I have suggested before that systemic learning is the primary value of games. In both cases the player does not need a very broad understanding of the game in order to be successful. In both cases the optimal strategy can be learned very easily by watching other players. Watching other players will always be a valuable tool for learning strategy, but it should not entirely obviate the need for experience, as it does in these cases.
Many years ago a friend was playing the real time strategy game Homeworld against another player. At a certain point in the game the other player had lost most of his units in battle but was able to teleport the rest of his fleet across the map. This was a very large map with a lot of resources, so he was able to keep gathering resources and teleporting away whenever my friend found him and began an assault. In one sense this was not a winning strategy, my friend had the upper hand and given long enough the resources this player needed for teleporting would have run out. In another sense it was a winning strategy, as the patience of the players became the focus of the game. My friend technically lost the game because he quit out of boredom after over an hour of this.
Similar scenarios occur in many games, often in games where defensive plays are too powerful and so the player who goes on the offensive first is at a disadvantage. I call these situations “patience games”. They are small games nested within the actual game, where the player who wants to play the actual game the most loses. I get an almost perverse amount of enjoyment out of patience games, but they are the result of poor game design. To resolve this, both players need to be drawn towards a central goal over which they must compete. Doing nothing should not be a viable strategy, and if it is the first player to not do nothing must not be immediately at a disadvantage. If they are, the game rewards reaction too much and action too little.