Asymmetry in Games



A few years ago, I had written an article called “Debunking Asymmetry“. I think that that piece makes some mistakes about how it framed some of the problems of asymmetric forces in games.

Quickly, a definition – “asymmetry”, in this context, refers to the player or players having different abilities from the start of a match. A Street Fighter II character, a StarCraft race, or a Magic: The Gathering deck all would qualify (for the purpose of this article, I will just use “character” to refer to any of these, as a shorthand).

For clarity, let’s say that emergent asymmetry – i.e., I’ve captured all of your pawns and you haven’t captured any of mine – does not qualify for the purposes of this article. There are also other kinds of inherent asymmetry, as in any game where one side is “on offense” and one side is “on defense”, such as Counter-Strike, or the board game 2 de Mayo. Neither of those are what I want to address here today, mostly because I don’t see those as a problem.

What needs to be discussed is what I’ll call “videogame-style asymmetry”, which has the property of allowing players to choose their asymmetric force before what we would normally think of as the true “match” begins.

In this article, I will explain the problems with videogame-style asymmetry, as well as offer some solutions as to how we can get its benefits without the pitfalls.


Current Thinking

First, let’s talk about what we currently do and why we do it. Where did this idea that you “pick a guy” before a game begins come from in the first place? Chess, Go, and other ancient abstracts don’t have that feature. A football or baseball player doesn’t have the ability to choose “which character” he will be before the game begins. Arguably, a coach or team manager is doing something similar, and that might be one of the first examples of this taking place. However, it’s still not quite the same sort of thing as “picking a character”; a sports team manager is negotiating for the best players he can get, not picking between a bunch of strategically-equal forces based on his preferences, experimentation, or other whims.

Regardless of where it came from and how it came to be, it’s now become a dominant design pattern for videogames of every stripe. Why is that? After studying this and speaking to many on the topic, here are the primary reasons I hear over and over again.

  1. Adds variety”. The number one word people will utter when you ask them why a game needs characters is “variety”. Basically the idea here is, without the asymmetrical forces, the game would get stale too quickly. Some even believe that “the more variety, the better”, and would advocate for perpetual addition of content to a game system.
  2. “Makes Learning The Game Easier”. This one’s pretty straightforward: if you section off parts of the game into “characters”, and allow players to just play as one character, they only have to learn that single character to get into the game.
  3. Supports Different Play-styles”. Players do have different “play-styles” – the kinds of decisions they make, whether they’re more or less aggressive, and so on. The claim here is, you can choose a character that supports your play style, and the argument is that this is a good thing.
  4. “Supports Different Player Preferences”. Let’s say you love StarCraft, but you hate the Terran units – you just find them really boring to play as for some reason. Well, now you can just play as Zerg or Protoss and never have to play as the Terrans.



I have issues with almost all of those points made above, or issues with some of the underlying assumptions of some of those points.

  1. “Adds Variety”. We already have non-asymmetric games that are interesting for years – look to designer boardgames like Puerto Rico or Through the Desert or even ancient abstracts like Chess and Go for examples. Many of these games are interesting and have plenty of variety purely as a result of having relatively strong rulesets.
    The underlying assumption with the “variety” claim seems to be that a system on its own – without a bunch of “content” – just can’t be interesting. People advocating this view seem to be unable to imagine just a gameplay system that itself is deep and has tons of emergent variety for years. This is understandable; we are still in the early days of “interactive system design”, and frankly, most of the systems we make aren’t very interesting – they do need something – asymmetry, tech gimmicks, an important social message, in order to maintain an adult’s attention for very long. However, it’s not hard for me to imagine that we will get far better at this discipline in the years to come.
  2. “Makes learning the game easier”. I’ll grant that this point has some merit. However, the way we do things now, videogame-style asymmetry seems to have a net negative effect on ease-of-learning, mostly due to the invitation it sends designers that reads, “please, add way too much content to this system!” It’s great that you kinda only have to learn to play as one character to get into a game, but to really play a game you do ultimately need to learn *all* of its rules – not just your character, and most of these games have a lot of characters. So in practice, this might mean an easier on-ramp, but a much longer road.
    I should also mention that you don’t need videogame-style asymmetry to get this effect. You can wall off some features in a “tutorial” mode (which could be skippable) and ramp up, even without asymmetry.
  3. “Supports Different Play-Styles”. A symmetric game can do this, too. Any sufficiently deep game with enough strategic possibility space is going to support different play-styles.
    Finally… are we really sure it’s even a good thing to “support the player’s play-style”? To me it seems as though much of what games are about is learning, often times learning highly counter-intuitive strategy in order to get better. So, let’s say you’re a naturally aggressive “rush-down” type of player. Well, sometimes in a strategy game, you have to go against that – you have to learn not to do that thing you have a propensity to do sometimes. I think trying to cater to someone’s playstyle is kind of similar to catering to players who have low skill. Players should be pursuing optimal play, regardless of their personality – and in that pursuit, their personality will still come through in the kinds of errors and novel moves they make.
  4. “Supports different player preferences”. This argument is also pretty easy to take apart. What if I “prefer” not to play against a Ryu – does that mean I shouldn’t have to? What if I “prefer” the enemy Ryu to only shoot fireballs at me? What if I prefer to only play against players who are vastly worse than me at the game? I think there is a role for “options” in a game, but they should be things like turning on or off music, or changing the window size. You should not be able to turn on and off various parts of the system itself. You have a favorite character, sure, and I can see how it could be framed as a negative thing that you can’t always play that character. But there are tons of kinds of things that “only sometimes happen” in every system, some of which you might love, but only happen infrequently. Perhaps you have a favorite monster, or a favorite combo – should you just be allowed to experience that one narrow “favorite thing” over and over? Beyond the fact that such a scenario makes very little sense in the context of a strategy game, isn’t it pretty likely that “supporting a player’s preference” and letting them have one chunk of the system over and over again every game is the fastest way for them to get bored with that thing?
    I think some of this is due to the “videogames are broken toys” influence; we’re used to being able to play with our strategy games. The idea of a finely-tuned and optimized strategy game which would fall apart if we start messing with the rulesets is somewhat alien to us in the world of videogames.
    Finally, this “players can just avoid playing with the parts of your game they don’t like” aspect also invites designers to be more lax about adding mediocre or even bad content to their games. After all, if players don’t like it, they can just not play with it! There are a few problems with this idea, but the simplest to explain here is the fact that your “crappy characters” become basically traps. Someone’s going to play that crappy character not knowing he’s the crappy character, and that’s a big waste of time for that player.

Those are my rebuttals to the claims of why we should use videogame-style asymmetry that I’ve heard. But I have other problems with the design pattern as well.


How do you choose?


One argument against asymmetry that I find increasingly convincing over the years is this: how do you choose which character to play? Let’s assume that you have some fighting game, with a few dozen characters. Let’s also assume that these characters are basically balanced (kind of a big assumption, but bear with me). Also assume that you have to make the “first” pick – like you can’t just “counter-pick” your opponent or anything like that.

This choice puts players in a position where they have to decide between several conflicting interests. Usually, it comes down to some mixture of these two: should I play as the character that I think would give me the best chance to win, or should I play as the character that I think would result in my having the best experience? Essentially, having to choose between trying to win as a player, or trying to craft the experience as a designer.

Here’s where people tend to make a big mistake, which is to think that this is an “interesting decision”. But interesting decisions in strategy games are decisions between two strategies, or tactics, not deciding between a strategy and some external thing, such as, “would it be fun to do that?”

In my view, it is absolutely a designer’s job to ensure that players are never put in that position. The “strong thing” should always be the “fun thing”. Most players already accept this concept when it comes to in-game interactions.

This is made worse by the fact that in the real world, characters aren’t all balanced with each other, and the player always is best with one to a small handful of characters. Both of these facts add up to the player having to make that kind of a “good choice or fun choice” decision.

Even if one were to say something like, forget what’s fun – play to win! Pick strategically! The “strategy” layer here is exceedingly thin and solvable. You know nothing about how the match is going to go – you have to pick this gigantic chunk of game-state information all up front, completely blind, so you just have to pick based on very simple criteria (who are the best characters / the characters I am best with).


If It’s Important, Why Is It Optional?

This is one of the most comedic aspects of this discussion. Defenders of videogame-style asymmetry will tell you that their game needs the 10, 20, 50 or however many characters it has. If that’s true, then why is it that me and my friend can do nothing but Ryu mirror matches?

Is the argument there “well, yeah, you can do that, but then you’re choosing to do something more boring”? If so, why are you even letting me do it? There is no guarantee that players will have any idea that that’s “more boring”; they may just think that that’s a perfectly fine way to play the game (and shouldn’t it be?), in which case they’re just getting a worse experience out of your game, for no fault of their own.

Maybe that seems like an extreme case, but how about this: most players only play 2-4 characters or so. That means if you have two players who play against each other, at best (assuming no character overlap between the two characters), these two players are only experiencing 25% of the game. I would guess that most of the time, for most players (obviously serious tournament players excluded here), 80-90% of a fighting game’s characters simply don’t even get used. That doesn’t sound like very efficient design to me.

To re-state my point succinctly: if the “variety” granted from asymmetry is so crucially important, why can it be so easily ignored? Why isn’t it just a rule, the way health (a feature we all know actually is critical) is?



This problem is not necessarily inherent to videogame-style asymmetry, but in practice, the temptation to add too much content seems to be too great for most designers. Today’s games have dozens, sometimes hundreds or even thousands of asymmetric characters, each with their own “learn-my-rules-too” tax. Each of these characters also comes with an exponential rise in the difficulty of balancing the game, too.

One problem is that something about the way this design pattern works makes designers think about characters as something other than rules. So, some work goes into developing the system, but then there’s a ton of work on characters – the “content”, if you will.

This propensity to see content as separate from the system is harmful, incorrect, and I think largely aided by videogame-style asymmetry (as well as a good dose of thinking like a toy designer). Strategy games don’t have “content” – they have the system: a core mechanism, a goal, the supporting mechanisms it needs, and nothing else (read more about this in my book Clockwork Game Design, or watch episode 5 of 3 Minute Game Design).

Of course, the annoying irony here is that many of those games – Magic: The Gathering probably being the best poster-boy for this problem – really don’t have much going on in terms of their actual gameplay system. They need a constant influx of new content in order to stay alive – so you get a “10,000 in 1 Fun Pak” of extremely shallow games, rather than one deep game (I’ve referred to it as “width” before).



Before I get into solutions, one thing that I do see happening often in criticism of my writing is that readers tend to imagine my solutions and suggestions sort of slapped onto existing designs. To be clear: what I am going to advocate wouldn’t work for StarCraft, Street Fighter, DotA, MtG, etc. What I am advocating would be something that would need to be there from the ground up and it would need to be designed around. The kinds of systems we would end up with ultimately as a result of this suggestion would necessarily be significantly different in many ways from any of the kinds of games we have right now.

Without further delay, here is what I recommend designers do with asymmetrical games in the future.

Asymmetric Game Design Guideline, Part 1
Forced random characters

Here’s how it works. You hit a “play the game” button, a character is chosen for you randomly, and the game starts. For a single-player game, leaving (because you don’t like what you rolled) is a loss (which of course means you need some kind of single-player elo system, but to be honest, you need that anyway). This has a lot of benefits.

  • We avoid the situation where players have to choose between something that’s strategically strong and something that would be fun.
  • Players are forced to experience 100% of your game, rather than some small percentage of it.
  • No need to make up weird tournament rules about counter-picking (this would also be helped by my second part of the guideline).
  • It’s faster. Even if you only spend a few seconds picking a character, if you play 1000 matches, that adds up!

Forced random characters is quick and elegant. It’s really not totally unprecedented either. Boardgames like Agricola deal you random cards at the start of the game which sort of create a “character”. League of Legends also has a popular mod called “ARAM” – All Random All Middle, which gives you a random character that you’re forced to play. They give you a little bit of wiggle room with it (you can sometimes “re-roll” and you can also trade with other players), but even without the wiggle room it’d work just fine – and League wasn’t even built with ARAM in mind!


Asymmetric Game Design Guideline, Part 2
Few characters

To clarify: only as many characters as your system actually needs to function. Of course, you should consider that this number might be “one” – don’t commit to asymmetry just for its own sake. But start out with just three or four, and then scale up as necessary. If you’re reaching numbers higher than 20 or so, something might be wrong. Less than ten seems like a decent guideline. Here’s why

  • With < 10 characters, the chances of you actually balancing your game go from “impossible” to “still not guaranteed, but possible with tons of effort”.
  • Learning the entire game is pretty easy! You don’t have to play the game for months before you stop losing games because you didn’t even know a card existed.



Before I conclude, I’d like to mention a rebuttal that a colleague of mine gave me the other day on this topic. Basically, his argument was, “if you force random, it makes the game way harder to learn, because now players have to learn to play all the characters before they can play competently”. There are at least four methods that I would employ that would stop that from being an actual problem:

  • Few characters (guideline 2) limits the number of things players have to learn. If you have a game with 4 characters, it’s really not that bad that you have to learn all four.
  • Elegant design in general. People are used to thinking about a “character” as a set of 20-50 unique moves with tons of unique hitboxes, all of which would amount to probably 5-10 pages of rules that you have to learn. We can design simple-yet-deep (elegant) characters that are easy to learn, but difficult to master, and that’s what we should be doing.
  • Low execution barrier. I’ve talked before about execution in games and why it’s something to be concerned about. I can see why, if you’re thinking about forced random in Street Fighter, you’d be intimidated. Now you have to learn all these weird different inputs that are physically hard to do and require training of your hands. This entire concept – that some moves should be hard to input correctly – is something that needs to be slammed in the trash bin of videogame history immediately.
  • Tutorial Mode. You can have the game start a player out by only giving him access to one character, or two, then three, then four, and so on.


If designers create games with these two guidelines in mind (as I plan to, going forward, and actually already have done with Auro), we will end up with games that are easier to learn, possible to balance, and without any of the weird “counter picking” or “playing designer” problems that asymmetric games currently have.

Even if I haven’t sold you on the idea, I’d love for readers to do a thought experiment with it for a moment. What if you had to design a game with, say, four characters, and forced random? What other kinds of decisions would you make to support that? My feeling is, these guidelines will help designers do what they already should be doing: creating deep, elegant systems of gameplay that are easy to learn, but difficult to master.


Enjoyed this article? Consider becoming a supporter of my work on