Before I was the marketing guy here at Antlion I spent 14 years running a marketing and consulting company for games. In that time I consulted closely on a number of fascinating sides of the game industry, including being the guy who picked which games got published and which got rejection letters (All the power was mine to hold!), the monetization strategy of games, I even got to be lead designer on a few titles by some strange twist of fate later in my career.
One thing I always loved consulting on were game balance issues. Understanding game balance is really a niche field, and even veteran developers can sometimes overlook the very basics. Today, Antlion fans, I share with you the kind of conversations I got to have with these online game devs so that you can understand what challenges a game like League, DOTA, Overwatch, World of Tanks, and many others face with each update in their attempt to chase balance.
Before we begin, let’s be clear about one thing: Balance is not about making things equal. Creating a set of options where the bonuses or differences you get are minor or non-existent (Ken & Ryu in Street Fighter 2) is BORING. Balance is about creating the feeling that your choices are affecting the style with which you play and the paths you take without impacting your chances at victory in a meaningful way.
Iconic, but not "good" balance.
Single player games have a huge advantage here, which is victory is measured in completing the game as opposed to defeating another player. I can complete Fallout 2 with a stealthy sniper or a power armor wearing melee killer, they’re balanced because they feel completely different and I can win equally well with both. Multiplayer games, on the other hand, may not feel as though I am on equal footing between these two.
Step 1: Defining the Objective
So with that in mind, a balance meeting begins with this question: What is the objective of this update?
The key word here is “objective.” As some examples, you may balance to make the game easier for new players or harder for veterans. You can balance to increase or decrease the use of a character/ability. You can balance to increase or decrease the repetition of a level. You can even balance a game to get people to pay more... sick as that is.
Notice that balance goes in both directions for nearly every option, and usually in a mutually exclusive fashion. Making a level more repetitive is not necessarily a bad thing, as it provides structure to create a high skill cap on micro managing a strategy. Making a level more open gives a higher skill cap on a more macro view of a battlefield. They’re different objectives, but neither is right or wrong inherently.
Most people think that “making a strong character weak” (AKA: a nerf) is what balance is all about, even devs sometimes fall into that trap. Instead we need to look at the goals of the product first.
For instance, creating a new easy to use champion in Paladins (Hello Lex, welcome to the game) is a great way to help balance the skill gap between veteran players and new ones.
This guy comes with an auto-aim ability. Just sayin'...
Similarly, the balance really COULD be to reduce the reliance on a certain character or ability via a nerf. There’s a reason you see a lot of Kodiak-3s in Mechwarrior Online. The goal here is to give a wider variety of play options and play styles, which will increase the longevity of players at the cost of upsetting the players who rely on the play style being reduced.
You'd probably get similar results searching for "nerf" and any mech/hero/champion
Notice in both of these examples there’s a benefit and a cost. The holy grail of any change to balance is to find a way to achieve the objective with minimal cost, but it is incredibly challenging to do so.
A simple way we often approach balance of “characters” is not to make A = B, but rather A > B and B > C and C > A. Or to put it another way, to ensure there are ample amounts of counters to every character, style, and strategy. However, if the counter is too narrow or too wide we similarly run into balance issues. There's no point in running a counter to a strategy that only comes up every 10-20 games, but having a counter that must be in every match isn't balance at all, it just becomes a requirement of play.
But I digress, the point is, that every update for balance reasons should have a clear cut objective in mind from the start, whether that is creating more play diversity, helping new players, getting old players to come back or stay longer, or even just make more $, the goal needs to be clearly laid out from the start.
Step 2: Implementing the Objective
The next question we ask in a balance meeting is How do we determine what needs changing?
Once we know the objective that usually doesn't mean we know how to achieve it best. Making a character more widely used could be done by buffing the character, nerfing other characters, or even just changing the maps. It's not always obvious which path is best.
To determine the path to take we start by listing out all the possible methods to achieve the objective, then divide information up into two groups, data and anecdotes.
Data is rarely made public, but you’d be AMAZED at how much data is recorded every game session. Everything from character selection to a heatmap tracking where you go, where you kill, where you die on every map. How you use your abilities in what rotation, at what timing interval… if a game is doing it right, the quantity of data is staggering to look at. This is a small clip of ONE spreadsheet containing JUST data on some weapons in a real game, sent direct to me from a developer contact of mine. We blocked out certain information that would reveal the game or sensitive data.
You like numbers? You may have a career in game balance!
The first thing we need to do on the data side is return to the above question and ask what our objective is and, as a result, what data is going to inform that decision. This could be a very simple set of data: In the example of “We need a new character that is easier for new players” we’ll start by pulling the data for all the player stats from their first 10 hours vs. players with 500+ hours and look at the comparison numbers. The goal is to isolate the key stats that tell us why new players are bad. Maybe their ability to aim is worse, maybe their time to conflict is shorter (IE: They are running straight at the enemy), maybe they are not using their abilities as often, maybe they are being matched poorly, and so on. This will inform the potential directions we can go to make the game easier for new blood.
Another example could be map balance. Below are two images, which the developers at Mechwarrior Online were kind enough to provide publicly. Do your best to come to a conclusion before reading the analysis I’d give (just based on this information in a vacuum and several hundred hours in MWO).
Image #1: Heatmap of player movement
Image #2: Heatmap of Kills
The conclusion you likely should have come to is that despite there being a LOT of movement on this map in a variety of directions, ALL of the conflict simply came down to two points: The tunnel on the left and the map-center (it's hard to see in this image, but the center is basically a ridgeline with a crashed ship on it that provides cover). This map played out the same way every single time. And that was BORING. This map needed an overhaul, badly, and this heatmap tells you the whole story of how people were playing your map design.
These are just a couple examples of how data can inform balance, but data can only take us so far. If you rely only on data you're going to miss more often than you hit. There's simply too much data and too many variables for numbers alone to parse. One thing our brains are good at, which computers are not, is creating causal syntax. Or to put it another way, once we have a bit of data and a lot of experience we're pretty good at guessing possible cause and effect, which is where the other half of this balancing act comes in.
Anecdotes consist of two things. Instinct of the developer and player feedback. This is where all the bitching and moaning comes in, and the reasons sometimes devs pretend they don’t read the forums. First, dispel that myth: The devs are reading the forum. They may not reply, they may not even censor assholes, but they’re paying attention.
The first thing we do is flag people we don’t like. Toxic players and players with a clear bias in their post towards certain things. You’re flagged and your comments are purged before analysis begins. If I’ve ever written an important line to make you think twice before making a toxic post, this line may be it. If you want the devs to listen, do not be toxic in game or on the forums! The reason for this is simple, toxic players have unclear motives and are unreliable. There's plenty of data without their input.
I was going to put a picture of a toxic post here, but then it would be full of profanity, so use your imagination.
Next we remove players under a certain play threshold. Nice as you may be, without X hours into the game you’re probably not useful to us for this. X of course varies with the objective. If we care about making new player experience better X may be very low. If we’re focused on end game skillcap stuff, X may be extremely high.
Then we hunt for relevant comments and suggestions that highlight our problem and objective. New player complaints, suggestions on what maps are boring, whatever it is. We remove the player name from the suggestion to eliminate any bias, and basically have a meeting where these results are tallied, read, or somehow go from words back into data. For instance if, over the last year, the map with the greatest complaints is NOT the map we expected to be the problem, like the above image would suggest, we need to figure out why that is before going further. It could be that people LIKE repetition or it could be we have overlooked some bigger issue on another map. We won’t know until we listen to the players.
Sometimes there just isn’t consensus, data, or anything else to go on. In these cases, as the last resort, good developers trust their instinct. These basically all start with “I think that…” and among the game dev team we determine what it is we think is the issue. I think that new players are being matched against veterans too often. I think that the systems we built are not as accessible to new players, even though it's obvious to veterans. I think the game should go Free to Play (ugh, the worst conversations all start there).
One thing I can say is devs are in a tough spot when it comes to talking about balance with the players. Everyone has an opinion, but they don’t want to post data because it will be used by the players to either justify a position that is potenially wrong (because they will only have some of the data) or used by the best players to game the system as far as possible. Can you imagine what a professional Overwatch team would be able to do with accurate heat maps of every level? Basically posting data will create anomalies in the data! At the same time they’re objective may not be the same as YOUR objective. If their objective is attracting new players they may have to do so at the cost of the happiness of existing players like you! Or they may simply be nerfing the hero you love so dearly. Seriously, are you going to tell me you’ve never been against a nerf because you use that hero? It’s human nature, and that is ok, but your complaint about it creates a lot of noise that devs have to learn to ignore.
What devs CAN do, however, is be more clear with the objectives behind each update. We’re very good at giving patch notes these days (this was not always the case), but as an industry we have a ways to go with communicating motive.
And again, balancing a game is an imperfect science. Devs are going to get it wrong pretty often in every game, just remember that what you THINK may be the reason for a change may not actually be the objective. Ergo, what you think is a bad change may actually be exactly what the devs set out to do, even if it may come at the cost of your personal experience.