Starcraft II – A Study in Balancing Asymmetrical Design

There is no such thing as a perfectly balanced game.

Or at least so I’ve heard, from various sources. Thinking about it makes sense though; Chess, Go, and even Noughts and Crosses (Tic-Tac-Toe) for example, both rely on one player starting first, despite each player having precisely the same tools at their disposal, being restricted by the same rules, and all playing within entirely equal environments.

If we were to design a game like Chess, we could feel relatively safe in the knowledge that certain aspects of the game were safe from balancing problems. This is because both players use exactly the same amount of board space each, and have exactly the same number and type of units. The rules or methods of play could change, but then it is reasonable to assume that those changes would be applied to both players. Yes we could still make an overpowered chess piece, or too many of them, but our opponent would be given the same too.

Games designed with asymmetrical mechanics however, may not have the same luxury. They often have to be adjusted and scrutinised for balancing, and in the case of competitive multiplayer video games, this is important. There are exceptions to this, Zach Gage’s ‘Really Bad Chess’, flips all these balancing rules of chess on their head by populating the board with a random proportion of each chess piece, and makes them different for each player, and simply lets the player have fun with the new system.

Zach Gage’s ‘Really Bad Chess’, an interesting experiment in throwing all the balancing rules out of the window.

What is Asymmetrical Design, and Why?

Depending on how far you’d want to go with the definition, and how deep the asymmetry, I’d describe asymmetrical design as games where some or all of the tools, objectives, resources, and methods a player has at their disposal are exclusive to them, and mechanically unique.

Examples of asymmetrical design would be games like Left 4 Dead’s multiplayer that pits survivors against zombies, with both teams using different methods during play; Chris Hecker’s Spy Party, a magnificent reverse Turing test, where one player tries to act like an NPC while completing objectives and the opposing player must figure out who the real player is and snipe them; Keep Talking and Nobody Explodes is about one player defusing a bomb and all other players using a manual to tell them how to do it; upcoming DOOM Eternal’s ‘Battle Mode’ pits a player as the DOOM Guy against two other players as demons, skewing both the play-style and quantity of players on each side; and of course, Starcraft II.

Spy Party from the perspective of the party-going spy.
Spy Party from the perspective of the counter spy sniper.

But why do we use it? Asymmetrical design leads to variety, and gives us choices. It can prevent repetitive strategies. It requires that our opponent knows as much about our gameplay mechanics as we about theirs. And it allows us, in some cases, to almost feel like we’re playing an entirely different game to the other player. Essentially it prevents boredom, but it also can encourage the player to devise creative ways of defeating their opponent.

As fun as asymmetrical games can be though, it still almost always needs more balancing than a more symmetrically designed game, because exclusive dominant strategies are much more disruptive than shared ones. Dominant strategies make all other strategies practically redundant, leading to uninteresting gameplay and wasteful design. Dominant strategies therefore, are bad enough when both players can deploy them, but even worse when only one side can, leading to huge advantages for one player.

Starcraft II

Blizzard, I believe, are masters of game balancing. And what makes this most impressive is that they manage to balance games that are either filled with a variety of asymmetrical mechanics, or are huge sprawling games that contain masses of content, or even sometimes both. World of Warcraft would be an obvious example here with its diverse class system and huge world full of dungeons, bosses, and the chance to take part in PvP.

However, I want to focus on Starcraft II (2010), one of the biggest real time strategy games of the last ten years, with a legacy that stretches back to 1998’s original game (and beyond if we were to count the RTS design of Blizzard’s Warcraft games too). I’ve started playing a lot of Starcraft II again recently, and what continues to impress me is how Blizzard have managed to create a game with so many asymmetrical mechanics, yet still keep it balanced and interesting enough to become, and stay, the premier competitive multiplayer RTS in the world. Why am I looking at a game that’s almost a decade old you may ask? Well, I still think it’s relevant, and while it’s not as popular as other competitive multiplayer games anymore, the fact that it is almost a decade old and still going strong is a testament to its quality. It’s also, in my opinion, a good game and a good example of asymmetrical design without going too bonkers; so there. There are other games that are more asymmetrical yes, but given its multiplayer esports status, it’s important to recognise it as something that demands ultra tight balancing, something made more difficult with asymmetrical design.

Let’s look at a clear example of how Starcraft II utilises asymmetrical design in some of its fundamental mechanics. This is most apparent in the mechanics of the three races: Terran, Zerg, and Protoss. All three have quite distinct and unique ways of producing units and buildings.

Before we start though, it’s important to note that the way that the economy is built through the acquisition of minerals and gas is exactly the same for all three races: worker units gather minerals from patches and take those minerals back to a central base building. Similarly, once the appropriate building is placed over a gas geyser, workers collect gas from it in exactly the same way. Ok, back to the races, we’ll look at just two: Terran and Zerg, and leave the Protoss out for now.

The Terran is perhaps the most recognisable of all the races, especially in terms of how we understand traditional RTS mechanics. Worker units build structures, and a worker must continue to stay building it until it’s done. Some of those structures produce different types of units that pop out from the structure once they are built. These mechanics force the Terran player to build many structures of different types in order to quickly build up a large army.

The Zerg is probably the most unique race of the three and brings mechanics to the game that seem difficult to reconcile with those of the Terran. Its worker units still build buildings which still take time to complete, but the worker unit is consumed in the process. Building units as a Zerg player is even more left-field. Zerg units can only be built, or birthed, from eggs at the Hatchery building, the main base building. Furthermore, some Zerg units can only be created by morphing from a different, existing unit. These mechanics force a Zerg player to expand their bases much earlier than other races in order to quickly build up a large army.

Each race also has a ‘macro’ mechanic, a tool that is periodically used to assist the player. For Terran it is an ability to mine extra minerals, for the Zerg it is the ability to create more larvae at the hatchery, increasing the number of units that can be built at once.

It’s a very diverse system that has the potential to create an interesting web of unique mechanics, but in some ways seems very difficult to balance all of that. Now of course, two players can play the same race against each other, which in effect nullifies the diverse approach each could take to victory. However, I think there are enough interesting choices and strategies to be made in Starcraft II, even in the case of a mirror match.

The pictures below demonstrate the unique methods each of those two races goes about buildings some of its units and structures. There are more factors that go in to the mechanics of these races but we’ll keep things relatively simple for now.

Starcraft 2’s multiplayer has become a game that relies on tight timing and fast decision making, with games that can be over in a couple of minutes or half an hour. So, given this and what we know about the races from the above images, it begs the question…

How the #!?* do they balance that?!?

Too often our (or at least my) gut reaction to balancing a game like this would be to study individual units, identify what could counter them, and then create a new unit for that purpose, or modify an existing one. The problem with this approach though is that there is a potential chain reaction where the new or modified unit unbalances another aspect of the game as there are hundreds of possibilities and eventualities associated with every unit. To try and start imagining a scenario for every single unit in a game where the player has control of many different units, structures, and the economy of the game is a recipe for madness. In fact it’s difficult in a symmetrical game, let alone an asymmetrical one.

Here’s the disclaimer, I’m not a designer at Blizzard, nor have I ever been. I mean, I’ve messed around in the Warcraft 3 editor in my time, but that hardly counts. So what I’m about to suggest is pure theory, some might even say conjecture. But anyway, aside from playtesting, I think what effective balancing here partly comes down to, is thinking about it by looking at the bigger picture. In this case, thinking in terms of ‘production’, or ‘how much stuff can I build in how much time’.

For example, if we think about what we can achieve in the first two minutes of the game for all three races, are we able to build up a force that can match that which another race can build? After we identify what we can build in that time, we can then compartmentalise that subset of units and structures, and more easily focus on the balance of them.

In fact this timing is something players, especially at the higher levels think about a lot, and this is sometimes measured in ‘time since the game started’, or the ‘supply’ level (the maximum number of units they can field at this time). For example: The game is two minutes in, at maximum efficiency, what army composition could the opponent feasibly have in that time? Or: at twenty one supply, what kind of army should I aim for by then?

Obviously it’s not all about production, as we have to factor in the traits of the individual units themselves: how fast do they move, how much damage do they do, can they hit land or air units, etc. But we could start to apply the same formula here too. Imagine it as zooming out from the focus on ‘what unit counters what’ and start thinking about what a player can achieve within a certain amount of time.

We can use this to also determine balancing that happens later in the game when more units, buildings and technologies are built. For example: with three bases, all mining the maximum amount of resources, what kind of army could we afford in 30 seconds?

Applying this line of thinking also helps us to understand how the macro mechanic for each race is balanced too. While on the surface they seem fundamentally different (Terran’s gives extra minerals, Zerg’s gives extra larvae to build more units), both help in speeding the production of a larger army. Zerg can create more units at once from the extra larvae, and Terran can use the minerals to build more unit producing structures, thus able to create more units at once.

What can we learn from this?

Trying to look at the wider picture, rather than concentrating on what our first instinct towards balancing would be, can, I believe, help a lot. If we looked at balancing a game like Starcraft II, with the attitude of trying to band-aid mechanics only at the microscopic level, instead of also considering the macroscopic bigger picture, we could easily end up with an incoherent patchwork quilt of a game.

Identifying our universal restrictions in the game can help us get our heads around balancing for asymmetry. In the case of Starcraft II, one of our universal restrictions is time, it is the same for all players in a match. It takes time to gather resources to afford new structures and units, it takes time to build structures and units, and it takes time for those units to move across the map to attack or defend. Correctly identifying these shared restrictions in an asymmetrical game can help us navigate the minefield of balancing for one.

However, we should also remember that no game is ‘truly balanced’, or ‘truly symmetrical’, and that to try and make one such, without enough variety of choice and decision making at any one point in the game, can make the choices we do have available, meaningless.


Related Stuff

The Never-ending Challenge of Asymmetrical Design – Josh Bycer

Design 101: Balancing Games – Dan Felder

Asymmetry in Games – Keith Burgun

Understanding Balance in Video Games – Keith Burgun

The Sweet Spot of Asymmetric Design | Creative Assembly and BAFTA GamesJames Green