GDC 2010: Design in Detail XI

So how do you recognize strength when you see it? How can you train yourself to appreciate strength?

The first way to develop a sense for strength is to change the balance constantly. People hate it because it resets their competence, but it will prevent them from optimizing their skills and their strategies. One theory about Beginner’s Luck is that when you first attempt a new skill your brain is very engaged and thinks the entire action through very carefully. On subsequent tries, your brian gets lazy and tries to take shortcuts, so you are more likely to be successful on your first try than on subsequent tries. Also, development is hard, and deadlines are approaching, so the temptation is to lock things down as soon as possible. Resist temptation and keep the balance changing until you find the true strength of the game elements.

As you strengthen an element, the other elements become relatively weaker. After a pass through the elements, you will find that the first one can no longer compete, and must be strengthened again. Keep doing this until all the elements feel powerful.

This guy is really good at Halo…

Pro-players often complain that “The guys making decisions suck at their own game” and it’s true! I’ll admit that I’m not very good at Halo. I’ll even claim that I’m not good on purpose! The problem is that the dopamine released for being a good player is the same as the chemical reward for being a good game designer. Since you can’t tell the difference, you may mistake the thrill of winning for the satisfaction of balancing the game. You should always feel like you are learning about your game, and if you start to feel like you have mastered it you need to change something so you aren’t good again.

You must acknowledge your own tendency to optimize and ignore problems once they have become familiar. Don’t work on the same element for too long, don’t become comfortable. If something feels so familiar you stop noticing it, change it.

At the same time, if something is right, if it is just perfect and you don’t want to lose it, you need to play it so much that it becomes part of you. I can still, after years and years, drive a new Warthog and tell you if it is tuned correctly or not. I’m known in the Animation Pit as “Three Frames” Griesemer, because if they added a single extra frame to a Halo melee attack I could tell immediately. You need to hone your sensitivity by playing with a finely tuned element over and over until it is ingrained in your muscle memory.


Balance of Power II

Tank Beats Everything

Even in the case where an element cannot be limited by role, it can be limited by availability.  Sometimes the Player should feel overpowered, either as a reward for a good player or as a temporary boost for an average player.  A game that is nothing but a relentless competence test can become monotonous, especially once a player has reached the limits of their physical ability to get better.  Giving them an occassional taste of unmitigated power relieves the pressure to perform and is an excellent palate cleanser.  An element that is fun and fulfills a player aspiration but can’t be limited to a single role is an excellent candidate for this experience.

No mana?  No problem!

Take that, Rock, Paper and Scissors

A successful transition from “balanced” to “overpowered” usually requires some changes, though.  It needs to be exaggerated so the sounds, effects, even the fiction, match the new power level.  It should become a featured element on a single level or two, the player should feel special for getting to use it, and the enemy resistance should be ratched up as well to highlight how strong they have become.  The element should also be made as modal as possible, so the player knows when the “overpowered” experience begins and it is clear when it is supposed to end and resume normal gameplay.

In a Corner

Let’s say an overpowered element doesn’t fit in any of these categories.  It isn’t just a perceived imbalance, the other elements can’t be strengthened to an equivalent power level, the role can’t be limited or enforced, and it isn’t appropriate for an over-the-top set piece.  If nothing else can be done, it has to be weakened somehow, right?  Let me tell you about The Needler

The Needler is a weapon from the Halo series.  It fires a stream of neon pink projectiles that actively track enemy targets.  On impact, each needle embeds itself into armor or flesh, remaining there for a few seconds before detonating and causing further damage.  If a character ever has more than 5-6 needles attached at any given time, they all explode in a particularly lethal chain-reaction known as the “Pink Mist”.  It first appeared in Halo 1 and was the bane of my existence for 10+ years because it was incredibly fun to use, but equally impossible to balance.

Like this, only a couple feet lower

I feel a headache coming on

It would start out too powerful, so we would weaken it.  Which would make it useless, so we would change the mechanics.  Which would make it effective in too many situations, so we would limit its placement.  We ended up drastically weakening it right before we shipped to prevent it from wrecking the game.  Three times!  The amount of effort, thought, debate, programming and art resources, playtesting surveys, data analysis and sleepless nights that were poured into the Needler far outpace any four other weapons combined.

So, even if an element cannot be balanced in any other way, the answer is still not to weaken it, but to cut it completely.  It is too difficult to weaken an element without destroying what made it fun in the first place, and the end results won’t be worth the hassle.  It’s best to just save it for another game where it will fit without compromise.

Balance of Power

“Iron sharpens iron, and one man sharpens another”

-Proverbs 27:17  [ESV]

“The whats-it is too powerful.”  It could be a weapon, an RTS unit, a character in a fighting game, a multiplayer class… it doesn’t matter because all chronic balance problems follow the same general pattern.  The game starts to revolve around a single dominant element, which is inherently overpowered and reduces the game’s strategic complexity, ultimately limiting its longevity.  Nobody notices when a single element is too weak, because they just avoid it.  And nobody complains when a dominant element increases strategic complexity, because that makes the game better and more fun.  And a simple problem, like a damage value that is just set too high, usually has a simple solution that is quickly applied.  But for chronic, fundamental balance problems, the designer is repeatedly faced with the same decision.  Should the dominant element be weakened for the sake of balance?  (Hint:  The correct answer is always “NO!”)

Perception is Reality

First, it is important to recognize that the only “balance” that matters is the balance that players perceive.  The goal of balance is longevity, and if players continue to play because they believe the game is fair, it doesn’t matter if it is objectively balanced in some measurable sense.  In fact, if a large enough community perceives the game to be balanced, but your metrics claim it isn’t, then the metrics are wrong.

Nice Gear

Sometimes power is deceptive

So often the problem isn’t that an element is too powerful, but that it feels too powerful.  Maybe its a gun with a really great firing sound.  Or a new unit that players haven’t figured out how to counter yet.  Or perhaps it appeals more to skilled players or everyone is using it because it is new and novel.  These problems usually fix themselves if they are left alone.  I changed the perception of an “overpowered” weapon during Halo’s development just by announcing that I had fixed it (even though I hadn’t actually changed anything.)

The Tooth Fairy is Overpowered

In almost every case, there is no such thing as an “overpowered” element; there are well-tuned elements in a crowd of underpowered, ill-tuned alternatives.  The Sniper class is implemented and all of the sudden you can’t take three steps without being headshot from across the map.  The problem is not that the Sniper is strong, but that the Medic and the Engineer are weak.  It’s easier to make a potent Sniper, so it immediately outclasses the rest.  Instead of spending time figuring out how to cripple the Sniper, focus on making the other classes equally awesome.  Or better yet, ditch the ones that will never feel as powerful as the Sniper and choose different classes that have their own natural strengths.

Get it?  Because neither one exists!

His only weakness is his terrible agent


Too Powerful, or All-Powerful?

In the paper design balance pass, every element should have been given a role to fill.  Sometimes an element breaks the balance by breaking its designated role.  An anti-vehicle missile that can be used against a crowd of infantry.  A “glass cannon” that can hold his own in a melee fight.  A long-distance weapon that is just as effective at point-blank range.  The solution in this case is not to weaken the element, but to restrict it so that its strength cannot be applied in as many situations.

Often it takes ingenuity to limit a weapon without weakening it, but it isn’t as difficult as it might seem initially.  A global weakness will affect the player in every situation, so a heavy-handed global weakness will be a constant irritation.  But a specific limitation will only be felt when an element is being used outside its role, which a player can learn to avoid, eliminating the annoyance entirely.  Nobody complains that their fancy sports car doesn’t work underwater, they just stay on the bridge.

[Continued in Balance of Power II]

GDC 2010: Design in Detail IX

Ok, now you have a flowing Sniper Rifle and all the other weapons are fun by themselves. How do you put them together?

This slide is for the Engineers. Design needs to start doing rough balancing in the middle of production, probably before you hit any kind of Code Complete milestones. Properly supporting the designers this early in the project is going to be mean violating several best practices, but we need as much time to iterate as possible.

First, we need a solid build at all times. Easy for me to say, right? But stability is important. If the build is broken it interrupts the balance process and I have to start over. And if you don’t maintain the gameplay systems the whole time, the game will not have time to get fun. It sucks, but that’s why you are working on games and not productivity software.

I know the theory is to “optimize at the end”, but it is impossible to balance a game with poor performance. Not everything has to run at a playable framerate; you can turn off lighting or textures or whatever it takes, but designers need a responsive platform on which to build. Imagine coding if you couldn’t see what you typed until two seconds after you typed it. That’s what it is like trying to tune a game with bad framerate.

For example, I have seen this in playtest after playtest. If you a level doesn’t have good lighting in a playtest build, the AI will score lower. People will think it looks stupid for some unknown reason. I don’t know why, but it shows how performance problems make it hard to balance the game.

During this pass, you are balancing strength.

In What the Dog Saw, Gladwell tries to figure out why there are 50 kinds of mustard, but just one kind of Ketchup. He concludes that Heinz is the best because it has all of the tastes in balanced proportions.

Heinz Ketchup has every flavor your tongue can taste. Here’s some of the ingredients: tomatoes (bitter), vinegar (sour), corn syrup (sweet), salt (uh… salt) And then you put it on french fries (umami). Every flavor is strong (ie has a high amplitude) but they are still balanced against one another.

Halo is like ketchup! It has lots of very strong elements, but since they are all strong they blend together into a balanced whole. In fact, without strength balance is much more difficult, because random factors destabilize the experience. A game with weak elements is like ketchup with weak flavors; if they become weak enough you start to taste the plastic from the bottle and the rat poop from the bottling plant.

[To be continued…]

Shorting the Fun

Betting Against the Bank

A dizzying array of factors fed into the current global financial meltdown.  Sub-prime mortgages, anti-market regulation, credit default swaps, a real estate bubble, global conspiracy, global warming, alien sabotage… the list is endless.  But when it comes to the collapse of specific commercial institutions like Bear Sterns, it is pretty clear that short-selling accelerated or even guaranteed their failure.

A short sale (or shorting for… uh… short) is when an investor sells stock borrowed from a third-party with the intention of re-purchasing the stock and returning it at a later date.  If the stock goes down during that time, the investor gets to keep the difference as profit.  On the other hand, if the stock goes up they have to buy back the stock at the higher price and risk losing a great deal of money.  In March 2008, Bear Sterns (a global investment bank) became strapped for cash because it had over-invested in mortgage-backed securities and the bottom fell out of the housing market.  They might have been able to borrow the money to cover themselves, or at least gotten a bailout, but so many investors were short-selling Bear Sterns stock that it looked like collapse was inevitable.  This self-fulfilling prophecy ended with the company being sold to JP Morgan Chase a few weeks later.

Down with arrows!

Winning the race to the bottom

Shorting the Fun

Designers do the same thing, only instead of bringing down financial institutions, they bet against their mechanics and doom their game to mediocrity.

Take a Sniper Rifle, for instance.  Long and sleek, capable of putting a bullet through the head of a distant enemy faster than they can hear the sound of it being fired.  It has everything going for it.  Lethality to make it feel potent.  Precision to give it a high skill-ceiling.  A steady cadence with plenty of anticipation; each shot counts, but a miss isn’t a disaster.  Usable beyond the range of any counter-attack, even the hanging contrail it leaves behind is a testament to the thrill of each shot.  A designer couldn’t ask for a more perfect example of flow.

Then why do the Sniper Rifles in so many games suck out loud?  It just isn’t that hard to make a satisfying Sniper Rifle, so how do so many designers mess it up?  The surprising answer is that most Sniper Rifles probably started out fun, but at some point in development a designer got nervous about game balance and began to hedge their bets.  They made it less accurate, or decreased the damage, or added some weakness like an unstable crosshair or a really long reload time.  They were so afraid that they wouldn’t be able to make the Sniper Rifle work with the rest of the game, they shorted it and ended up losing a great deal of potential fun.

Still not a good idea...

Not to be confused with "fun shorts"

That’s how you get a Sniper Rifle that takes three shots to kill someone.  Or a tank that drives two miles an hour.  Or a spell that costs so much mana your mage can never use it in a fight.  Or any awesome and fun element with a crippling weakness that makes it all but useless.  Designers sabotaging their own games because they lost sight of what made something worth including in the first place!

  • Adding a Weakness – It may seem like the best way to balance an element that is too strong is to give it an equally large weakness, but it just doesn’t work out that way.  Either the weakness will not be enough, in which case you will have an element that is still too powerful and no longer fun to use.  Or it will be more than enough, in which case you will have designed an element that has a weakness as its defining feature instead of the original aspirationI would play as the invincible ninja character, but he has taken a vow of pacifism and doesn’t have any punches or kicks.
  • Creating a Counter – Another common way to address an overpowered element is to design a second element whose sole role is to counter the first.  Not only is this second element probably not fun on its own merits, but everyone will be saddled with choosing an option that they secretly hope they don’t have to employ, just to prevent everyone else from choosing the fun option they desperately want.  I will choose the unenjoyable anti-tank mine so that nobody else can have fun using a tank.
  • The Old Switch and Bait – Many games, especially those looking for a sense of progression, will initially introduce the player to a hobbled version of an element, and then unlock the truly fun version as a reward.  Of course, this ignores the fact that players usually won’t invest in a game that isn’t fun, and even if they do eventually earn the right to have fun, a vast majority of their experience will have been struggling through a crippled, unenjoyable game.  If I kill 300 more rats with this blunted shovel maybe I’ll get to use a real sword!… in this game that is ostensibly about swords.
  • Death of a Thousand Tweaks – Perhaps the most outwardly reasonable (and therefore the most nefarious) way of shorting fun is to make an endless series of minor changes, each one leaving the element a little less fun than it was.  Slightly less accurate, a touch less damage, a little longer reload, a smidge more recoil, a fraction less range, etc.  In this parody of the tuning process, the element gets gradually less fun until it is “unfun enough” to be included.  The problem is that tuning is more delicate than balance, so the element will lose the qualities that made it fun long before it becomes fair.  It will also leave players frustrated; if only this gun had been tuned better it would be really fun!

So, ignore the analysts and the pundits!  If an element or mechanic has the tiniest spark of fun, a designer’s job is to protect and nurture it, not smother it because it might out-shine the rest of the game.  No matter how tempting, there are always better options, so don’t short the fun!

So what are the alternatives to shorting the fun?

GDC 2010: Design in Detail VI

After paper design, we move to initial settings. Usually there is a period in between while the programmers build the system into the game.

Some of the best producers I have worked with really understood the fact that this stage needs some breathing room!

How many of you are familiar with the concept of Flow? Lots of talks have been done on this, I’m going to assume you know what I am talking about, in general.
The problem is, he wrote in the ‘70s, he doesn’t address video games. So what does flow look like in a video game?

Smiling makes you happy; laughing makes you healthy. Certain finger movements make you have flow. We call that Cadence! This timing for the Sniper Rifle is very specific, because the cadence determines the type of flow created. It is different for different weapons or different parts of the game, but cadence is important in all kinds of flow.

Verisimilitude: the quality of seeming to be true. For controls this means that the physical action of your character in some way corresponds to (or at least doesn’t clash with) the action taken by the player. Remember the Paper Design for Halo 1? It speculated that zoom was on a trigger. If I could go back in time, it probably would be.

The sniper rifle doesn’t unzoom as soon as you fire your last shot (even though that would be more efficient). It unzooms with a slight delay so you can see the results of your last shot. Why is this important?
Studies have shown that blue sleeping pills work better. They are chemically identical, but the placebo effect is enormous. Slow-mo explosions look more “real”. Ask any action director, authentic explosions look fake because we have been looking at slow ones for so long. Your brain has expectations and when those expectations are met, the effect is amplified and that encourages your brain to maintain flow.

Let me tell you a story.

The Sniper Rifle looks great on YouTube because you can really follow the action.  Even someone that has never played Halo can figure out what happened.  Maintining a thread from beginning to end is a key component of staying in the flow state.  For more information, check a specific post I wrote on the topic.


GDC 2010: Design in Detail V

The second way to develop a sense of balance about paper designs is to look for anticipation. (Anticipation really isn’t the right word for it, maybe “Imperfect Predictability”?) If you make players guess, they won’t see the point and will quit. If you don’t give them a chance, they will feel controlled and quit. Balance means probable, but not inevitable, future events. This allows the player to anticipate them.

Don’t your players into ridiculous movie villains.

David Sirlin ( is a SF Champion and a game designer (and just happened to be in the audience when I gave this talk at GDC.) He calls this concept Yomi.
In an action game like Halo you don’t want players to engage in this kind of second-guessing. It will paralyze them from acting and end up in random guessing. The Sniper Rifle has one purpose; you know what someone with a sniper rifle is going to do. The role prevents you from guessing by allowing them to anticipate what is going to happen. If they are expecting an event, they can process it more quickly and follow the action better.

Successfully sniping confers no benefits. As designers, we throw around the term “Feedback Loop” a lot, but they are not always good because they lead to a game being overly predictable. Anticipation requires uncertainty, and feedback loops work against that.

The best thing about roles is that they help you keep things manageable by breaking the design into workable parts.

If you are using roles, make sure players always have other options. The Sniper Rifle is never your only option, so the entire balance never rests on one weapon. By using multiple gameplay channels you reduce the difficulty of your balance problem.

If you break up the balance into groups that do not require cross-balancing your problem becomes simpler. I won’t say that monolithic is always bad, just harder.

The Sniper Rifle is not on every map. This limits the number of possible interactions that need to remain balanced.

Imagine somebody is holding a gun to your head… There is too much to do, what are you going to cut! What are you going to do? How can you get control of your scope? By making the hard choices…

What are the hard choices in the paper design pass? Which mechanics will you emphasize? How will you engage the player’s imagination?

Producers, please insert your fingers in your ears at this time. Some of the best advice I ever got was “Once you are done, cut half” and I have never regretted doing this. I’m not going to say “Kill your babies” Now that I am going to be a dad, the term doesn’t seem appropriate anymore. How about “Put your babies up for adoption”?

To be continued…