In Depth: How to Define and Discuss Balance

Ok gents, it's time for a little discussion. There has been a lot of talk recently about balance. The issue is that balance means different things to different people. As a result, people often misinterpret the meaning behind other's posts, and then spend a lengthy amount of time arguing at cross purposes.

To be honest, this irritates me a little. I like to see rich and detailed discussion on the forums, not have it get cluttered up with the same misconceptions time after time. Hence, I would like to take the time to release a series of posts containing detailed analysis of some of the more complex concepts that underlie these discussions. I feel that I’m the person to do this because analysis of complex interacting systems (with extra waffle) is my speciality. Without further ado:

What is "balance"?
It seems a simple question, bordering more on semantics then actual game design. Commonly, it means that things are in a state of equilibrium or parity, yet in games, this one innocuous word hides a multitude of different interpretations. Let's break them down, and look at what they mean.

Competitive Balance
The first I'm going to talk about is the most obvious, and is simply that players have balanced opportunities to win. I'm going to refer to this as competitive balance or CB. Simply put, CB is about ensuring that all players in the game have an equal chance to win the game, based upon their own skill and strategic planning. If one player in a game had access to a special unit which was extremely cheap and powerful, then this would be a violation of CB, as that player would have access to an unfair advantage. CB is important because it appeals to our sense of fairness. One of the reasons that humans play games is to test and measure themselves against each other (although I would like to stress that this is far from the only reason). If we feel that the test unfairly biases itself towards one player, then the validity of the test is questionable, and this reason why we might play the game becomes invalidated. If you play to test yourself, then there is no reason to play if the test already has a pre-disposed outcome.

So how does this apply to PA? Well in one sense PA is easy to balance from a CB perspective. All players have access to the same basic tools in terms of units and structures, meaning their options are symmetrical. This is due to a lack of factions. However, this does not mean that CB is assured. The nature of map generation means it is possible for players to be positioned in awkward locations, have little starting metal, or to be placed between other players leading to being attacked on two fronts. All of these are possible. However, as PA goes forwards, these can be improved by refining map generation, spawn placement algorithms and the creation of spherically symmetrical maps (for those who wish to use them). The competitive balance will never be perfect, but it can certainly be good enough that most people don't complain. For many, it already is. However, a lot of the time, this isn't the kind of balance that people are talking about. That is something a little more subtle.

Content Invalidation
Let’s move onto the next possible interpretation of balance. This is probably the one that people use the most on the forums, although it often gets mistaken for CB. This second interpretation is the statement that no part of the game should completely invalidate a different part of the game. I'm going to refer to this as Content Invalidation, or CI. For example, if nukes were buildable straight by the commander, and cost 100M, then they would invalidate almost every other part of the game. The winning strategy, would almost certainly be "build nukes, scout enemy commander, win". To be sure, this would still be a game that is playable, and may even be enjoyable to number of (every odd) individuals. However, the majority of the game would be pointless if we assume that the player's first and most immediate goal is to win. The reason why this form of balance violation is bad is a little more complex to break down (although it is intuitive to see that it is bad). Fundamentally, we play games for interest and enjoyment. However humans also have a tendency to get bored if they are exposed to the same situations repeatedly. The amount of repetition required to produce boredom is subjective, and depends heavily upon the amount of variability that is perceived in a game. For example, some people find hours of enjoyment on a racing game where they continually try to shave milliseconds off their lap time on the same racing circuit. To me, such an experience seems the antithesis of fun, but i do empathise that some are able to perceive the minute variations between successive laps as fun. If we have a game which seems to offer a lot of different qualitatively different experiences, yet only a small number of these present a chance for victory, we are forced to choose between engagement and competitiveness. This conflict is something that makes us grumpy and bored.

In order for something to be a source of CI, we need to answer the question: "What part of the game does this thing invalidate?". On the surface, this question seems to be simple, yet on closer inspection it is actually far more complex. For example, one could claim that Shellers currently invalidate all laser turrets. In isolation, a single Sheller can destroy an infinite number of turrets without losing a single hit-point's worth of damage. However, despite this, I think most people would agree that turrets are not invalidated in the current version of PA. In fact, many would propose the opposite. The flaw is in the first two words of the argument above - "In isolation". In reality there are many things which can destroy Shellers, and there are many scenarios where Shellers are not available (although they are not yet perfect). It is these constantly shifting webs of interactions that make good RTSs such a joy to play, so it is unreasonable to suspect that any one thing will always be a valid response.

So how do we distinguish when something is merely an appropriate counter, and when something invalidates something? This is not an easy answer, as invalidation is a soft concept. We can at least start by saying that if something invalidates something else at all times and circumstances, then it is definitely violating CI. However, even in PA today, this is quite a rare thing. Most things can be used sometimes. Furthermore, it isn't unreasonable to suggest that some units should be very effective at dealing with other specific units. We have to think of the invalidation as a more analogue concept rather than a binary one, and ask how much of the game does this thing invalidate, and when? There is no concrete answer for this, but it is my hope that people reading this will be a little more grounded in the theory underlying balance, and will break down their arguments a little further before screaming "[x] is OP!"

Before moving onto the next section, I would like to leave with one general rule of thumb. We all know that at any point in the game, players will have a number of options they can perform. Some of these options will be valid, and some of these options will have been invalidated by enemy action. As a general rule of thumb, the most interesting kind of game is had when about half of those options can be carried out as valid at any one time. There is a lot of thought, logic, and even a bit of maths that goes into backing up this statement, and if I do more in this "In Depth" series of posts, then I will try to elaborate that statement further. I would suggest it may be reasonable to frame a discussion along the lines of "does this feature (or collection of features that can be reasonably expected to be used at the same time), at a given point in the game, lead to more than half of the opponent’s possible options being invalidated". If the answer is a conclusive yes, then that feature (or collection of features) is possibly a source of content invalidation.

Those are the two main concepts surrounding balance. In the next section, we will talk about some ancillary concepts surrounding balance, that never-the-less often get confused for balance.

The act of Balancing
We have discussed the two main ways that the word "balance" is used here. Expanding on that, I would like to pose a simple question: “If we know what balance is, then what is the act of balancing?". It sounds like a painfully obvious question, yet time and time again, i see people tripping up on this issue. We might say that the act of balancing is the process of bringing about balance, in one of the ways that we have already discussed. However, the context this phrase is often used in is often interpreted as "just making small changes to unit stats" in a distinct discipline from other game development skills such as coding up entirely new features, or creating new art assets. Balance is often described as a form of "fine tuning", or a layer of polish which is applied over the top of existing components. However, I feel that this view is overly simplistic, and misses some of the important details.

Regardless of the balance change, whether it is a slight decrease in one stat of a single unit, or the addition of an entirely new unit to counteract a perceived overly-powerful strategy, the overall goal is still one of the two forms of balance outlined above. Either the goal is to improve the parity between players, or it is to ensure that no part of the game is completely eclipsed by another. The important component of this is that we can view balance as having a core direction, and the fine tuning of how it is executed. For example, people may like the core direction of a unit as "A short range, high damage flame tank with absurd durability", however the fine tuning of such a unit may be something that many are unhappy with. The inverse may also be true. This may seem obvious to some people, but it is a core cause of misconceptions on the forums. Many people may be criticising (with varying levels of constructivity) the core direction of a unit, and yet people interpret such criticism as a being about the execution of such balancing. This is often where the "it's not done yet" arguments get wheeled out. Saying "It's not done" implies that the original point was a critique expecting the fine tuning and polish of a finished product, when in reality, it may be a perfectly valid, but more basic point about the core direction of something. Thus, when discussing such matters it is important to try to understand what the actual point being argued is.

This also raises another interesting question. Can a unit of any core direction be balanced such that it does not completely invalidate other parts of the game? Honestly, I don't know, but I would be tempted to suggest the answer is "Yes, anything can be made to work, but some things would be much harder to make work than others". It drives me mad when a suggestion is innocently made, then shot down by cries of "No, that would be OP", with little to no analysis of why such a suggestion could not be balanced. So, how can we tell which ideas would be hard to balance, and which would be quite easy? One needs to think in terms of the degrees of freedom of the unit. In other words, "what parts of this unit can we change without altering the core direction of the unit?". For example, let’s look at the contentious issue of megabots/experimentals. Megabots are frequently cited as being hard to balance. The reason for this is that the core direction of a megabot is the assumption of immense cost, but overwhelming power. With other units, a balancer always has the option of ramping the cost and power ratio up and down in order to ensure that nothing else is invalidated. However, a megabot has to be big and terrifying. If it isn't, then it simply isn't a megabot. This added constraint makes the unit more difficult to get right, as it is quite likely that the unit will be too cheap and will substantially invalidate other parts of the game, or be too expensive and is itself an invalid choice in most circumstances. This is not to say that it is impossible to balance such units, only that it is more difficult (and hence requires more developer effort) than a "regular" unit. Please note, this is just an example. If you wish to resurrect the megabot argument again, please do so in a different thread. The take home message is that ideas should be broken apart and the different levers that make them up should be examined, before knee-jerk cried of "TOO OP!" should be raised.

I think we now have a pretty good idea of what balance is. Now let’s take a little bit of time to ask what balance isn't.

Balance is not the same thing as fun
This is a bit of a tough nut to crack, but it is very important to get this point across. Time and time again, I see the assumption being made that balance and fun are basically the same thing. To illustrate how they are not, let us look at the textbook case of rock-paper-scissors. In RPS, both players play in synchronicity, and have no advantages that the other player doesn't have, meaning that there are zero issues with competitive balance. All three of the options are equally effective or ineffective, meaning that none are invalid, and no sensible person would start shouting "Rock is OP! Nerf Rock!". RPS is a perfectly balanced game. Yet, I would say that anyone who gains hours of enjoyment playing RPS is quite possibly deranged. RPS is balanced, but it really isn't that fun.

So why is there the underlying assumption that an unbalanced game is not fun, while a balanced game is fun? In order to answer this fully, we would really need a definition of what fun is, and why humans experience it. Why is it that photocopying 500 manuals for a paycheck is deathly dull, yet killing 500 boars on an RPG to grind out some loot is oddly compelling? To describe this, we would need to spend a long time discussing, psychology, neurology, philosophy and evolutionary biology (at least). I doubt people would care for such a discussion right now, thus I will boil it down as concisely as possible. I believe that fun stems from presenting a person with a novel challenge that can be overcome with an application of either skill, calculation or creative planning. The more novel situations that can be presented to the player, the longer a game will hold the player's interest. Balance can help maximise the amount of novelty that we get out of a limited set of tools, by ensuring that all of those tools get used at some point. However, if that toolset is very small, then no amount of balance will make the system engaging. In order for a game to be fun, we need lots and lots of tools for the players to play with. Unfortunately the more tools which are added, the more difficult the game becomes to ensure that all have their place, and none invalidate each-other. However, this does not mean that we should shy away from adding such tools. The important part is the quantity and quality of the tools available. If adding three tools invalidates one existing tool, then that is still a worthwhile trade as it still enriches the game. This obviously isn't as good a case as adding those new tools while preserving the utility of the old one, but it may still be a valid trade off to make. This is something that we should always bear in mind, particularly when considering that Uber's relentless iteration may temporarily invalidate one thing by adding another, before there is time to fix and tune. This is why any suggestions about balance that posit an unbalanced feature or unit should be completely removed are almost always extremely premature.

In Closing
I hope this has given a detailed analysis of how we discuss balance on the forums. In closing, I would like to suggest that when people want to point out that something is overpowered, they may wish to ask themselves the following questions before posting:

1. What does the unit/feature invalidate, when does it invalidate it, and how much of the game is it invalidated for?

2. What is the core direction of the unit and how is it fine-tuned? Which is more responsible for the problems of the unit?

3. Have I played enough games in order to say this is true across a broad set of conditions?

Finally, when suggesting a fix for a prospective imbalance, it is always best to try simple solutions first, as they are much easier to test. It would be far more likely for a dev to wander across a simple solution and think that it may be worth testing, than it would for a complex solution to be worth the time. Complex and more detailed solutions should only be developed if a simple solution is not functional. If the suggestion is that a particular feature should be removed (lots of people have suggested this for nukes), then it is always better to ask "how could this feature be fixed?" before going for removal. Remember, more tools results in a more engaging and re-playable game.

I hope this can move along the discussion around balancing a little, and cut down on some of points which haven’t been fully analysed.

Applying this methodology, I and other members of The Realm have been working on an example of a PA balance direction which aims to minimise content invalidation [Link to thread and direct link to documentation]. Particular effort has been made into determining a unique role for every unit with as little overlap as possible, whilst ensuring that advanced units are not automatically more efficient then basic units. Feel free to take a look, and let us know your thoughts.

I just quickly wanted to assemble a list of available balance 'levers' at the top of this discussion for reference, if I may. In no particular order (will reshuffle and add things in a bit):

• Cost
• Health
• Damage Per Shot
• Fire Rate
• Range
• LOS
• Accuracy
• Speed
• Acceleration
• Projectile Speed
• Projectile Tracking
• Firing Arc
• Splash Damage
• Turn rate (unit)
• Turn rate (turret)
• Energy Cost Per Shot
• Hard Counters
• Targetable layers

(if I've missed something give me a shout)

i would go and add a sub to cost such as
unit productioncost and maintainancecost the later as in f.e. energy per second for radar

I always enjoy your posts, MadSci, thanks for the thoughtful discussion.

THANK YOU.
This is my biggest beef with the 'crazy T2' approach Uber's been testing out. It's incredibly binary even if it is balanced, one player either kills the other before they get to T2 or the one trying for T2 gets there first and wins because of it.
Thankfully they'll probably be testing out other balance options regarding T2 eco now

I most definitely agree with the "balance does not equal fun", but I would argue that the T2 rush counter style being tested can still be fun in it's implementation, as the T1 centric player attempts to successfully harass the T2 rusher, and the T2 rusher has to protect his juicy advanced tier before getting destroyed.

Also size of energy pool used for shots and speed of said pool refilling.

OMG, you know so little about Racing Games!! That really is not that the point in racing at all

Anyway, you can put all the stats that you want, but if you fail to deliver a game that's challenging, enjoyable and fair, you're likely gonna to fail on the market.

You know why? Because our logic mind is a fraction of the whole. We evaluate and enjoy things mainly at a subconscious level. Like a slice of pizza, a kiss or a video game.

I program since the age of 12, now I'm 41. When I can't solve a problem, I try to simply go to sleep. When I wake up, often I've the solution, and it's simple and it works. It always amazes me.

How do you explain that?

Too much math not always help. You have most of the answers already, just sit, observe and enjoy the process. Answers will come. Let the whole mind work around the obstacle. Unlikely we can solve very complex problems with the tiny conscious/logic side of it.

And your point is?

My point is pretty clear to the one who can see it.

You're basically telling us that a game will be fun to play (people will buy it and play it repeatedly) if we put together all the numbers in a logic structure. And that's likely your professional approach to complex problems (and it's quite proven that often it brings to erroneous conclusions - there is a precise reason why).

Of course manipulating stats are a part of the process. But a game will succeed only if players will find it challenging, enjoyable and fair (specially in multiplayer games).

I have the feeling that in general we are forgetting the whole picture in here. A human mind will never be able to put together all those stats, and forecast the outcome.

But even so, as you said yourself, a well balanced game is not guaranteed to be fun to play.

So, I ask you, what's the point of your post?

Edit: are you telling me that if I do not entirely understand the stats behind scene, I can't judge if the game amuses and entertains me?

So, can I just clarify, are you suggesting that things are made worse by examining them rationally? Because that is a pretty bold point to make.

Where did I say that?

I like this statement of yours.

3. Have I played enough games in order to say this is true across a broad set of conditions?

But it's misleading 'cos you put it as the last one of three. We should start from there (actually we usually do). By playing. If something doesn't feel right, then we should move to numbers and stats in the effort to tune it. But those stats alone will never tell us if it will ultimately feel right in game. That's our subconscious mind who'll tell us. Period. No math will make a fun game. Forget about. Our whole mind is much more complicated than that, and we find amusement in what we feel, not in what we think.

So, if you're start from the numbers, you will never get the whole picture. Your judgement will be affected by your theories. And you will feel the wrong things. Other players will feel differently and you will not fix their problems. In fact, that may be what's happening to the Devs right now.

You've surely seen that at the moment there is a lot of emotionality and subjectivity about the current balancing of PA. Wonder why... because that's how our mind works, we get no motivation by logic. We get motivated by emotions.

Ultimately what I'm saying is that everyone can tell you if they're having fun playing the game. Even if they do know nothing about the stats behind it. They're entitled to give their feedback as much as you do. Telling us that there is only your way to analyze complex problems is a bit arrogant.

You said "To be honest, this irritates me a little." And that's fine, but you have to accept that people may see the problems differently than you do.

People do not agree on your logic, 'cos even your logic (as much as everyone's logic) is biased by emotions. It's really tragically funny how good people do not agree on logic facts. Nevertheless, you have to accept that people may follow different paths to solve their problems, and disagree with yours, or your conclusions. Actually, their conclusions and solutions might be better than yours.

Is it more clear now?

ps. Do not expect me to spend more time on this

It's a good job science doesn't work this way, or we would still be in the dark ages. To truly progress in science, you have to forcefully ignore your emotions (often the cause of so called "confirmation bias") and look at things objectively, changing your opinions as new evidence emerges. As Neil DeGrasse Tyson once said, "The good thing about science is that it's true whether or not you believe it".

I don't see how applying this principle could harm the balancing of PA, and actually think that trying to ignore our emotional reaction to balance changes would improve the ability to judge perceived unfairness.
For example, it's fine to say that people won't play the game and be thinking of every single stat that is creating balance, but it is wrong to suggest that the people making the game shouldn't be considering all of these things when balancing it.

But yes, I understand that science doesn't exactly match game development, so there will be people who think emotions should be a big part of the process. Meh, creative people.

He's not saying that's how you should play racing games, he's saying that's how some people derive enjoyment from these games.

Ah. To put this in manageable terms for those in the community who are having trouble reading that much text, logic is a great tool, but applying it in a vacuum will result in a game that is more about the grind than the awesome.

I agree with him.

Ignorance is bliss. Unfortunately, I cannot ignore an unbalanced video game, nor am I usually completely rational about it.

I am usually just blunt, with a hint of rationale. We play it. Something isn't working. You adjust it. Sorry if that sounds primitive but that is what I do. At least unlike some CoD games, I don't bipolar balance, if something needs an adjustment I think of the best one and apply mildly. I know I do good by that, I watch other things get left unnerfed, or overnerfed, and I always say to myself "no no no, its obviously this way not that way".

I agree with your post however MadSci. The facts are undenyable. I just support a system that is fun, varied, not "arranged" in an advantageous "go-to" setup that every competitive practices (rushing t2 in current build, practicing it until you get it fastest ensures you win because the winner is whomever gets it fastest), but sometimes a "turtle" beats you and sometimes a "rusher" beats you and sometimes "air" beats you sometimes "land" beats you sometimes "structural" beats you sometimes "orbital" beats you, and sometimes put up against one of those it doesn't because you did the superior hit versus it. If they go orbital and you react to it fast and limiting, and you win, you got their number. If they went orbital and tied you up with a land war and you decided you could win a land war before orbital was an issue, just to lose due to orbital, then they got your number.

This stuff hurts my head, I do not lie, but as long as there are lots of fun tools to pull suprise clutch-moves and attrition-wins off with, I think this game will be alright.

Definitely avoid the grind, dont want to practice fastest basebuilding or timing my t2 teching. 4 min 2 seconds, damn, stuart can do it faster, try it again trophy! nah, hell with that...

But surely applying logic outside of a vacuum is what MadSci is suggesting? Don't use purely emotional responses to what is essentially, an objective problem?
Who knows, I think I lost track several posts ago!

1. Subjective appraisal
2. Objective, logical appraisal
3. Balance
4. PROFIT

I would also like the concept of Competence level invalidation. Usually when balance is concerned it is taken to the limit of skill that as if a ideally skilled players would play it. However in the real world there are a lot of noobs and other people that are not ideal. And a significant part of the player base will not reach a level comparable to ideal. A game that has a lot of competence level invalidation will bounce from players discovering one OP feature in their current play environment until someone other finds another "slammer" with no even games inbetween. A game with competence level concervation will provide quality matches to player with small differences in skill. They might not be able to use all the game features but the features they can use make up a good game. However when game mechanics are introduced they potentially impact all levels of play. Usually the impact to high level play is deemed more valuable. This can wreck low level balance where easy-to-access upsides are not held back by their hard-to-use counters. Often finding the orthogonality is key in this aspect.

If you could make a choice that would leave you with excatly 50% chance of winning either way it would be fine with regards to Competive Balance, Content invalidation and Competence level invalidation. However it would be dull in a balance sense because there is no way to win, you don't contribute to your competence. A good balance would constantly give the opportunity for the player to make winning choices (or in constrast make losing choices). This way a game that has 1 significant choice and 9 insignificant choices is poorer than a game that has the winning distributed over 10 slightly significant choices (or more moving parts tends to create a more interesting system).

The extent of content invalidation can also be significant. Over how many kinds of mex distributions are we supposed to balance over? What kinds of biome variations would be outruled because of balance? Do we rule over 20 player games as not worthy of being severly balanced? Do we put an upper limit on how far a 1v1 spawn distance can be (ie if lnog distance games play poor or funny is that a problem to be fixed or reason not to play long distance)?

What I don't like about balnce considerations is often people assume that what makes or doens' t make a unit OP is obvious. Balance includes a whole lot of opinion no what kinds of mechanism should be in place and what their acceptable impacts should be. It is often not interesting to know whether a unit is balanced but whether why is it.

Idk about player skill. I agree sometimes players are unaware to use absolutely everything. In other situations they are not as fast. Honestly, every game has a point where two players can be complete mismatches. At least this game has single player. I think this game is pretty good at strategy being ballpark per person's peer skill level, and the speed of gameplay is adjustable theoretically although that would give a hardcore player a headache playing slowly. There are also lots of tools to help players get better or manage a game without expert speed.

Overall, I think this game is well on it's way to exceeding any skill gaps or imbalance caused by a learning curve. I feel the difference in economy in t1 and t2 is one thing that can affect skill gap, but that being fixed this game is neigh perfect at introducing players and making them fair and fun against each other. This game also has "go" mechanics, like adding more players against less players, giving players an economy multiplier, giving a player an AI teammate, ect. Unbalanced, but in the intentional way to attempt to counterbalance an existing skill imbalance, so one weak player is given an advantage against a stronger player and it is expected that under those circumstances the outcome could end up either way and becomes a game of interest.

So this game has some sweet fun packed in it, is all I know.