Centralized vs. Decentralized Complexity


Magic: The Gathering is very bizarre in its massive influence upon game design as a field. As one of the most successful and long-running tabletop games since, I dunno, chess, Magic has been played by tens of thousands of people across the ages, harboring a unique appeal due to its complexity. Magic is a tactician’s wet dream, with an uncountable amount of strategies to adopt and invent. You can win a game of magic by just summoning comically strong creatures, by using relatively simple effects as the centerpieces of combos, by taking advantage of methods to cheat out strong cards without paying the costs, or by just playing an unreasonably large deck.

And yet, new players are still pretty regularly coming into Magic, drawn by the game’s popularity and the promise of exciting gameplay. This is in spite of the fact that the game’s official rules are 200 goddamn pages, and features chestnuts like this:

Magic players knew what this would be a picture of without even having to look

The secret is that Magic is not difficult to learn initially. The core gameplay loop of Magic is fairly simple, all things considered. Untap your stuff, draw a card, tap some lands, play some stuff, make guys fight, tap some more lands, play some more stuff, end turn. With some very minor hiccups (usually surrounding how blocking works), learning the fundamentals of Magic is, all things considered, pretty easy, if you already have a bit of experience learning board games.

The difficult parts of Magic‘s rules come from outside of the core rules. It’s the individual card keywords, limited to a few cards from a few select sets, that introduce the bizarre and perhaps unintuitive rules. Sure, maybe Ninjutsu requires a more nuanced understanding of how combat works, but it’s on 10 cards from 1 set. Maybe the decision on when to take the +1/+1 counters versus Servo tokens for a creature with Fabricate is a complex tactical decision, but if that’s overwhelming, just don’t play the 14 cards that use the keyword.

This leads to my point: Magic has decentralized complexity in its design. The core fundamentals of Magic, the bits that are present in every single game of it and are essential to playing the game, are pretty easy to pick up. You can understand the entire turn structure, the color wheel, combat rules, all of the card types, and a bit of basic strategy in, I would guess, an average of about 1 to 2 hours. The lion’s share of complexity lies in a bunch of components that, while part of the game, are not essential parts of the game, in particular cards and keywords. By the simple fact that Magic decks are built by the player before the game begins, a player gets to pick and choose what complexity they deal with.

A new player, or perhaps just one who doesn’t like dealing with complex decks, might go for an intuitive strategy: summon the big smashy creature and make them smash things. Throw a bunch of fireballs at your opponent’s face. Just gain so much life that whittling you down to zero becomes a chore. Experienced players, or those who want to interact with the game at a more technical level, can run decks which operate on that more technical level, complex decks like Eggs or, my personal favorite, Oops, All Spells.

Of course, a single player can only control half of the cards they see in a game of Magic. Their opponent, after all, might have built a complex deck while the player opted for a simple one, or vice versa, but I actually think this is the part of Magic‘s decentralized complexity most applicable to all games. Because a single game of Magic can only feature a limited number of cards (decided by players’ choices in deck construction and limitations of the format being played), and only a subset of those cards will be in play at any one time (well, unless you’re playing a very stupid game of Magic), even though the player could run into mechanics or strategies they are unfamiliar with, they are only encountering a small subset of the mechanical possibility space at a time. By playing game after game like this, discovering the game one deck at a time, what would be an extremely complex game to learn piece by piece is easily absorbed in bite-sized chunks.

Compare this to one of my favorite board games, Archipelago.


While I love this game to death, I positively dread having to teach it to a table of new players every time. Archipelago is a banner holder for centralized complexity. The most complex parts of the game, the parts from which the strategy and interesting mechanical interaction arise most of the time, are the fundamental actions taken over the course of the game. Buying and selling on the foreign and domestic markets, moving your characters, harvesting resources, bartering with your fellow players, and dealing with worker insurrection are all essential to the game; every player will perform all of those actions every game of Archipelago. For a player to understand the game at a base level, they have to understand all of these actions, that that leads to a pretty big wall that players have to surmount in order to start meaningfully playing.

So, the downsides of centralized complexity, and by inverse the upsides of decentralized complexity, are fairly apparent: decentralized complexity makes it easier for new players to start your game, and ultimately lets you build extremely complex games knowing that your players can digest that complexity in pieces, rather than as a whole. It can add a sense of discovery to a player’s process of learning the game, as they discover new concepts and rules, and it also lets you build an extremely modular experience, much like Magic. However, centralized complexity must have upsides, too, or else you’d think designers would stop designing games in such a mode?

Well, for one thing, centralized complexity gives the designer a bit more control over gameplay as a whole, and lets them focus the experience. The original Coup, a tight bluffing experience with only 5 types of cards, is an extremely refined experience, with interactions designed to create a tense, paranoid play experience, from the relationship between the Contessa and the Assassin to the discord sowed by the Ambassador or Inquisitor to the bullying brought about by the Captain. Coup: Rebellion, the sequel which allows players to swap identity cards in and out from a pool of 25, can vary wildly. Sometimes it recaptures that tense paranoia of the original, sometimes it devolves into brow-beating militarism if the card pool is too aggressive, sometimes it’s too easy to gang up on people and beat them down without fear of repercussion. Some combinations of cards in Rebellion surpass that of the original game in terms of interesting, but others dip far below it by being boring, or by simply not offering anything unique.


Archipelago was a game designed with an extremely tight experience in mind, the experience of putting players into the logistically crushing and emotionally dulling role of a colonial authority. By centralizing its complexity, the game ensures that this experience happens every game. Since the design guarantees that all of these complex components are going to be in the game, every game, each is tailor made to heighten this experience by playing off of one another. The worker rebellions cause stress on the local and foreign markets, which in turn can force players to work their workers even harder to recoup the lost materials, which can just make the next worker rebellion even worse. This feedback loop couldn’t exist if any one of these components was modular or optional.

Furthermore, centralized complexity ensures that once your player has learned the game, they’ve learned it, tip to tail, and their mental capacity in regards to the game can be dedicated solely to strategy. With a game like Magic, you are, at bare minimum, learning new rules interactions every time a new set comes out. You constantly have to readjust your understanding of the rules, which means you’ll have to constantly readjust your strategies. On top of that, if you don’t play for a while and come back, that need to learn is just compounded as you “catch up” with the game. Meanwhile, once you learn Archipelago or Coup, you learn those games, and every play experience after that can be solely dedicated to play and to strategizing, instead of learning. If I don’t play Archipelago for a while and I come back later, I don’t have to “catch up”. Once I shake the rust off, I can get right back into the game.

Obviously, neither centralized nor decentralized complexity is the “right” answer, and a designer should carefully consider which mindset to adapt in building a game. When well done, a game with decentralized complexity can last forever, constantly adding in new elements to make the game fresh without overwhelming new players, like Magic. Centralized complexity when well done, however, can live on forever by allowing players to strategize with the same game components ad infinitum, allowing masters to use a small, common pool of mechanics in novel ways forever, like ArchipelagoCoup, or even chess.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s