Recently, I have come across a series of YouTube videos by one Keith Burgun, presenting his ideas on (video) game design. While I do not necessarily agree with all of his conclusions, I do find his insight into many aspects of gaming keen and valuable, which is why I’d like to edit his main points, as I understand them, together in this post. For more in-depth explanations and examples, please check out the original videos.
As the foundation of his future analysis, the author proposes a taxonomy of “interactive systems”, which he applies to electronic entertainment, among many other things. This taxonomy divides interactive systems into four hierarchical categories, based on how the consumer interacts with them and what values they receive from it:
- A Toy is a “bare interactive system” with no goal. A good toy must offer a lot of possible interactions, as its inherent value lies in what the author calls “mapping”, wherein the consumer explores and maps out these interactions. A video game example of this form would be Minecraft.
- A Puzzle is a bare interactive system with an explicit goal. Its inherent value is “solving”, so a good puzzle must be challenging but ultimately solvable. A video game example would be Portal.
- A Contest is an interactive system with an explicit goal and a session limitation. Its value is “measurement” of how well the participants have mastered a certain discipline, so while a good contest does not have an optimal solution, each individual participant’s result must be accurately measurable.
- A Game is a contest that obfuscates the game state. Its value is “understanding”, and a good game only lets the player directly observe parts of its full state, so they can develop understanding of it and make meaningful decisions based on incomplete information and their own heuristic experience.
The author posits that games (in the sense of his taxonomy) are unique among entertainment forms in providing their consumers (players) with the “game fun“, which he defines as “heuristic learning”. Since human brain is wired to be very good at just such a learning, as well as at creative problem solving rooted in it, it naturally rewards us for engaging in games.
This heuristic learning (“understanding”) is facilitated by the game feedback loop:
- The player makes a decision and acts upon the game state accordingly
- The game determines the outcome of the player’s action, obfuscates the new game state, and presents it to the player
- The player observes the new (obfuscated) game state, learns about the efficacy of their decision, and applies this knowledge to their future decisions
“Obfuscation” here does not necessarily mean “hidden information”: it can also be strategic in nature, such as obfuscation of long-term consequences of a particular move in chess. Obfuscation is important, because “decision” here refers to choices based on incomplete information, situated on the continuum between random guesses (where no data informs the player’s choice) and calculations (where the player has all information they need to make an optimal choice).
Editor’s note: The author does not mention another type of “decision”, namely, a choice between incomparables, as mentioned by Extra Credits.
A good balance between these two extremes is what lets the player bring their heuristic knowledge to bear, and all of the player’s decisions throughout the game build up to the end state, i.e. winning or losing. A game must be technically but not practically solvable, complex enough to avoid solution, but simple enough to learn (the classic “easy to learn, hard to master” formula).
Elegance, Depth, and Width
The goal of a game designer is to create an interactive system with the smallest appropriate amount of rules. The author defines a game’s depth is as the measure of how much heuristic understanding it requires, as well as of how strongly it resists a provably optimal solution. The game’s “elegance” is the ratio of its depth to the number of its rules.
Editor’s note: At this point, the author echoes the findings of the MDA model: what he calls “rules” (inherent complexity), Hunicke et al. called “mechanics”, and his “depth” (emergent complexity) is the number of their “dynamics”. Thus, “elegance” can also be defined as a ratio of dynamics to mechanics.
“Width”, on the other hand, refers to the amount of “content”, which, in turn, is defined as additional rules that are individually nonessential to the game. Adding content is a common way for developers to extend the life cycle of games that lack a truly complex and thus sustainably engaging core mechanics. The drawback here is that while content makes it linearly harder to “solve” the game, it also makes it exponentially harder to balance it, which ultimately produces a handful of Pareto optimal solutions that dominate the metagame. Thus, additional game content often actually adds very little to the game.
Editor’s rant: A major problem with many popular pen-and-paper RPGs is that their business models rely on selling new source books every few months, adding new content that, instead of contributing meaningfully to the gaming experience, essentially turns it into a shopping mall trip.
Clockwork Game Design
In this section, the author presents his method of approaching game design, which he also describes in detail in his book. In this method, the game design begins with a “core mechanic”, which consists of a “core action” (what you do in-game) and a “core purpose” (what you hope to accomplish with that action). The core mechanic is intrinsically tied to the main goal of the game, which should be the ultimate expression thereof. With the core mechanic in place, supporting mechanics can be built around it, adding nuance and complexity to it without directly competing with it for the player’s attention.
The author illustrates it with an example form Super Smash Bros.. The core mechanic in that game is “attack opponent…” (core action) “…to knock them back” (core purpose), and the main goal is “to knock all opponents so far out, they cannot come back”.
Unlike deterministic variables, random ones are inherently unpredictable and, more importantly, unlearnable (their range is learnable, but a particular instantiation is not). Random variables in interactive systems fall within a continuum, based on when the randomness occurs in relation to the player decision:
- Input randomness instantiates the variable before the player makes their decision and informs the latter. One example would be randomly generated levels.
- Output randomness instantiates after the player’s decision and determines its outcome. An example of this would be rolling dice to see if an attack hits.
The author argues strongly against using output randomness, as it takes agency away from the players and delivers potentially incorrect feedback that hinders them in their heuristic learning (e.g. when bad luck ruins a good strategy).
Editor’s rant: This is basically the crux of why, for all of their pretense of being tactical games, D&D-style RPGs are very bad at being tactical: you only have a vague idea of your chances to hit an enemy, and are required to commit resources (your turn and possibly character ability points) to an attack whose resolution contains three to four layers of unpredictability (hit or miss, crit or normal damage, min to max on the damage die; you may also have to “confirm” a crit like in 3.*e). With this much randomness, any hope of tactical planning goes right out the window.
Games are, by design, technically solvable, and once solved, all decisions in them become calculations. As mentioned above, one way to prevent having your game solved is to keep adding content to it, but doing so only transforms it into a contest of who can calculate the new content faster. Another is to employ output randomness to intentionally hinder the player’s learning, but it has its own obvious problems.
The lack of a cap on look-ahead in a deterministic perfect information game can lead to “analysis paralysis“, and the solution is to hide the complete game state behind a carefully placed information horizon. This limits the calculation space and facilitates the “incomplete information” decisions (see above), however, it also shifts the game towards output randomness, as the player now has less understanding of how their decision resulted in the new game state. Therefore, a balanced information horizon is an important ingredient of a heuristically learnable game, along with a high elegance ratio.
Games where the player’s success is determined by their execution of particular inputs substitutes explicit output randomness with one “seeded” by the player’s manual dexterity and hand-eye coordination. While it is better than regular randomness, since it can (theoretically) be improved with practice, execution-based challenges are much better off in contests than in games (see the taxonomy). Informed decisions need time to form, therefore limiting time that can be taken for a decision (i.e. relying on reflexes) makes it an execution problem instead.
A theme is a metaphor attached to a gameplay system. It is used to communicate the ruleset to the player via symbols, and must be therefore, first and foremost, clear and understandable.