Dynamic Balance

Dynamic balance covers the opening, midgame, and endgame of classic game analysis on a much finer scale. Rather than treating the game as three discrete phases, which is fine for postgame analyses, we have to consider the fully continuous spectrum of play, from start to end. This differs from the static balance, because we have to consider the interaction of the player or players with the statically balanced system. We are concerned with not only static balance, but also dynamic balance ”how the balance changes with time and player interaction.

We have to consider passive balancing , that is, keeping the system in balance with the player, without actually moving the equilibrium point. We also need to consider active balancing , shifting the equilibrium dynamically in response to the player's actions, either to increase difficulty or to adapt the game to the abilities of the player.

What Are We Balancing?

The word balancing suggests the act of restoring a system to an equilibrium position. Consequently, our discussion of game balance revolves around the hypothesis that a game is a system (or collection of systems) that needs to be restored to equilibrium. This is, in fact, the case, but these systems need to be balanced in slightly different ways, depending on their nature and function.

In some ways, this is a moot point ”part of balance is the player herself ”and we don't necessarily want to have to implement handicaps for a good player so that a rank newbie can hold her own. You can take play balancing too far. If a player is willing to work fairly within the bounds of the system in order to gain a competitive edge (such as those players of Starcraft who memorize statistics and use the user interface shortcuts to the fullest to gain an advantage, as shown in Figure 8.14), she should be allowed that privilege.

Figure 8.14. Starcraft.

The objective of balancing a game is to provide a game that is internally consistent and fair, without allowing players to exploit flaws and weaknesses to gain advantages. The other aim (of course) is to make sure that the game is fun.

A game system should initially be in a state of static balance, but once it is set in motion, a different form of balance, the dynamic balance, is maintained . The success or failure of the game designer to manage dynamic balance defines the gameplay. A good dynamic balance provides the impetus of the gameplay.

There are several ways that the player can interact with the dynamic balance, depending on the aims of the game. (Note that these interactions do not have to be at the global level; the player can be assigned different interaction models for different subgames.) The following three interaction models are available. The player can:

• Restore a balance.

• Maintain a balance.

• Destroy a balance.

Restoring a Balance

If the task of the player is to restore the balance, the object of the game is to move the game system back to an equilibrium point. The gameplay is derived from the player's attempt to restore the initial unbalanced state of the system back to a more ordered state.

The opposing unbalancing force is not strong enough to counteract the player's attempts to restore order. Either the force has stopped interacting with the system before the player intercedes, or else it interacts so weakly that the player is able to force the system back to a balanced state.

An example of a simple game that uses this particular interaction model is a sliding block puzzle. The system starts in a chaotic (unbalanced) state that must be restored to an ordered (balanced) state. The win condition is when the system has been restored.

Maintaining a Balance

If the task of the player is to maintain a balance, the object of the game is to prevent the opposing unbalancing force from overrunning the system.

The difference between this interaction model and the previous one is that the unbalancing force is still very much active. For each action of the player, the opposing force attempts to provide an (at least) equal and opposite reaction to counteract and defeat the player's attempts to force the system back to an ordered state.

If the player was to stop interacting with the system, the unbalancing force would win. The gameplay is defined so that the ideal state is a position of equilibrium between the two opposing forces. There is no win condition in this sort of game. Given a steadily increasing opposing force, it is only a matter of time before the player loses.

An example of a game that uses this interaction model is Tetris (see Figure 8.15). The player must attempt to keep the playing field clear of blocks, and the opposing force relentlessly tries to fill the playing field with blocks. Success depends on maintaining the balance between the opposing forces. As the difficulty level increases, the speed with which the opposing force attempts to fill the playing area with blocks increases , and the player must work faster and harder to maintain the balance.

Figure 8.15. Tetris.

Destroying a Balance

If the task of a player is to destroy a balance, the object is a reversal of the first interaction model. In this case, the player takes on the role of the unbalanced force, and the opposing force attempts to balance the system.

Actually, this is almost identical to the first interaction model. It's really just a matter of semantics. It depends on who is defining whether a system is balanced or not. It is, however, still useful to retain the distinction. Note that destroying a balance does not necessarily mean that the system has to be plunged into chaos ”although that is a valid interpretation and has been used in a good number of games . Additionally, it could also mean that the player has to shift from one equilibrium point to another.

An example of a game that uses this interaction model is X-Com: Enemy Unknown (see Figure 8.16). The Earth has been invaded by evil aliens (aren't they all?) who seem intent on using it for their own nefarious purposes. This is the first equilibrium point. The task of the player is to rid the Earth of the evil aliens , hence moving the system to a new alien-free equilibrium.

Figure 8.16. X-Com: Enemy Unknown.

As we've already hinted, this could also fit the description of the first interaction model. However, we prefer to use that for game systems where there is no discernible initial equilibrium point from which the player is transitioning.

Balanced Systems

The aim of game balancing is to set up a balanced system for the player to experience. Consequently, gameplay should be set up to do the following:

• Provide a consistent challenge

• Provide the player with a perceivably fair playing experience

• Avoid stagnation

• Avoid trivialities

• Allow setting of difficulty level (where appropriate)

Let's examine how we can achieve these aims in our game design. The following sections describe each point in some detail and discuss methods of implementation.

Providing a Consistent Challenge

The game should scale in difficulty smoothly as the player progresses into it. A number of games seem to miss this particular point, and in some cases, the midgame experience turns out to be substantially more difficult than the endgame.

One of the major things to be checking for is that the game's difficulty increases smoothly and does not peak or spike irregularly. This is definitely something we need to avoid; the damage that it does to the pacing of the game is irreparable, because the player will feel that anything after that point is anticlimactic. In other words, if you show off your strongest hand too early in the game, anything after that is a disappointment. This highlights the importance of thorough play testing.

Providing a Perceivably Fair Playing Experience

A major factor in whether a player enjoys a game is whether she perceives it to be fair or not. Note that it does not actually matter whether the game is fair. What is important here is the player's perception of fairness.

For example, if you are going to allow the computer opponents to cheat, you should do it subtly. Blatant cheating is a throwback to the days of minimal processing power; the only way the player could expect a decent opponent was if the computer cheated, and there was not enough spare processing power to even attempt to hide it. Nowadays, of course, we cannot use this excuse . Blatant cheating by the computer is seen as a sign of laziness on the part of the designers and developers.

A number of measures can be used to help ensure that a game is fair. This is integral to good design technique. For example, no good designer would knowingly design a game where a player destroys all chances of winning by taking an action earlier in the game and not finding out until later in the game.

A classic example of this is Monty on the Run by Gremlin Graphics (see Figure 8.17). This was a maze-based platform game for the 8-bit ZX Spectrum computer released back in the '80s. The object of the game was to guide the hero to freedom and to escape the long arm of the law. One of the unique selling points of this game was the "freedom kit." When the player started the game, he had to choose five items to take along. These items would help get past various obstacles throughout the game. The problem was that the player was given no clues as to which were correct and which were not. The only way to discover this was by trial and error. Thus, the player was effectively doomed from the start unless the correct choice was made at the beginning of the game. This is clearly not fair. This is an extreme example; not many games so blatantly flout fairness in this way. However, even though this is an old game, there are still more recent games that do similar things. How many times have you forgotten to pick up an item that is necessary later in the game? The only choice is to restart the game or to painfully pick your way back to retrieve the item.

Figure 8.17. Monty on the Run.

This example also breaks another of the cardinal rules of game design fairness: informing the player of everything needed to play the game and not using unknowable or unguessable information. The choice of the freedom kit items is not intuitive. The player is given no information of which is the best choice to make. In fact, the only way to be sure of making the correct choice is to cheat ”either by reading it in a magazine or finding out from friends . Any game that requires reading a strategy guide or searching on the web, rather than the player's natural ability to play it successfully, is fundamentally flawed.

Another important consideration for ensuring fairness ”particularly in multiplayer games ”is to protect new players while they are finding their feet in the world. Here, we can take a lesson from nature. Most animals protect their young in some form or another. We need to ensure that our new players are protected in a similar manner. Nothing is more discouraging than joining a new game only to be slaughtered for fun or profit by an experienced player. This protection can be provided in a number of ways. The opportunity to practice in a single-player game is the best way to enable this. First-person shooters tend to implement this well; the single-player game provides a good training ground for players to prepare for joining the multiplayer online m l e. Purely online games, such as EverQuest and Ultima Online , need to take a different approach (see Figure 8.18). Here, special training areas should be set aside for new players. There should also be a guiding principle that says that an 8- foot 300- pound barbarian giant cannot be killed by an ordinary rat. Where's the balance in that? Anarchy Online does provides specific training areas for new players ”where you can get killed by a rat!

Figure 8.18. EverQuest and Ultima Online .

The question of balance in an online game is an ongoing one, as you have the opportunity to respond to "real play" situation in a way that a standard single-player game can't ”until the expansion pack, at any rate. Asheron's Call does this on a regular basis ”and yet it's still fundamentally unbalanced in favor of magic users. It seems to be that way because that's what the magic users want, and obviously the publishers want to hold their audience. Suddenly, game balance is more about ongoing sales and politics than it is about the game play ”but it's something that you have to consider as a designer.

Nothing is more frustrating for a player than having to repeat actions over and over again. This is a cardinal (and unfortunately common) sin for a computer game. Surely everyone has screamed in frustration at the save-die-reload cycle that causes the player to replay a section of the game already completed. Worse still are those games that send you back to a distant "checkpoint" on death. Game designers need to learn an important lesson here. Nobody wants to be forced to repeat completed sections of the game. Give them the choice, by all means, but don't force it on them.

Last, but certainly by no means least, the concept of fairness extends to how the player's avatar dies. At no point in the game should the players feel as if events and their consequences are out of control. The players should be made to feel as if every event in the game is under their control. If they fail to control it, they should feel that it was a failure to act on their part, and not just a random arbitrary event that they had no way of avoiding. Instant death syndrome is the most concrete manifestation of this. Deathtrap Dungeon , published by Eidos, was guilty of this (see Figure 8.19). Many of the traps were completely unavoidable. The only way to even know that the traps were there was to trigger them and die. Then your next incarnation would know better. Good game design? Maybe not.

Figure 8.19. Deathtrap Dungeon.

The worst thing about fairness is that everyone has a different concept of what fair is. The task that the game designer faces is finding a common ground that will keep as many people happy as possible. The old adage is true: You can fool some of the people all of the time, and all of the people some of the time, but not all of the people all of the time. Harsh though it may be, as a game designer, you are trying to fool as many people as possible into thinking that your crude simulation of a real world cares enough about the players to ensure that they get a fair experience.

Avoiding Stagnation

Stagnation is generally as unpleasant as it sounds and smells almost as bad in a game design as it does in water. Stagnation occurs when players are playing a game and reach a point where they appear to be stuck, with no way to go on. There is nothing worse than running round a level of the latest and greatest first-person shooter trying to find that last hidden switch that opens the level exit. Of course, it's not just that type of game that is guilty of this offense (although it is a persistent offender). Any game that leaves the players in a position where they simply do not know what to do next is stagnating.

In some cases, this is very difficult to avoid. A sprawling action-adventure has so many different combinations and configurations that it is difficult to anticipate exactly what the player may or may not try and do. However, it is still possible to give the players positive and negative feedback as they progress. The problem of stagnation can be tackled passively ; that is, the designer can make sure that the clues about how to proceed are hidden in plain sight. The other alternative is to tackle stagnation actively: Have the game work out whether the player has been wandering around aimlessly and provide a few gentle nudges to guide him in the right direction.

The key point is never to let the player feel bewildered. The players should always feel as if they know what their next move should be. It is no fun to bang your head up against a brick wall simply because you are completely and utterly stuck in a game. This ties in with our earlier piece of advice about making sure the player is adequately provided with information. If a player has to resort to outside assistance ”whether by cheating, reading a strategy guide, or looking up the answers on the web ”the game designer should view that as a failure of the design.

Avoiding Trivialities

Not many players actually want to be bogged down in the minutiae of myriad trivial decisions when they can be directing the big decisions at a higher level. Forcing the player to decide where the gold is stored when she is trying to build an army and plan a grand strategy is a form of slow torture. Who cares where the gold is stored ”just store it.

A trivial decision is a no-brainer decision. Any decision that has only one logical outcome, or where the outcome has no real effect on the game world, is trivial. The player should not be bothered with these. Let the computer handle it and, if necessary, inform the player afterward.

Sid Meier's Alpha Centauri handled this magnificently (see Figure 8.20). In this game, the player can choose to handle every decision in the game, from grand strategy all the way down to unit production and direction. In addition, the player can let a computer-controlled manager control the bases, and the player's units can be set to automatic control. The player has the choice to micromanage every aspect of the game, from the movements of an individual unit all the way up to the overall control of the planet. The important thing here is that the player is given the choice . Other games force the player to do all the micromanagement, whether they want to or not.

Figure 8.20. Sid Meier's Alpha Centauri.

Worse still, there are games that force completely trivial decisions on the player. If the player wants to choose whether to wear the blue tunic or the red tunic, then fine, but don't force that decision on him as a gameplay choice, unless it has some sort of direct relevance to the gameplay. For example, if the player's avatar needs to be disguised as a guard in the enemies' Red Guards, the blue tunic wouldn't help matters.

To be fair, this sort of trivial decision doesn't crop up as blatantly as this; usually it appears in a more subtle form, disguising the fact that the decision has no value. Note that if this is done well, the trivial decision can actually add to the gameplay. It can add depth and flavor to an otherwise shallow game. If you do choose to use it, though, make sure that you use it with care. If you are caught, it will undermine your gameplay.

For an example of how you could use trivial decisions, consider a fictional cops and robbers game. Your officer is patrolling the city as usual, on the lookout for crime, when suddenly he spots a group of suspicious-looking characters on the corner. He stops the car and they immediately run down an alleyway and vanish . Behind the scenes, these people never had any part in the gameplay; they were just flavor to give the impression of a bustling city. The player is not led too far down the wrong path ”she is just given the impression that there is more to the city than meets the eye.

Setting the Difficulty Level

The first time the players of your game come across balance is when they select the difficulty level. The standard for difficulty levels seems to have evolved into three (or sometimes four) distinct settings: easy , normal , hard , and nightmare (or similar assignations), popularized by the original Doom from id Software.

Traditionally, not all games have difficulty level settings. For example, adventure games tend to have only one difficulty level ”for no good reason, as far as we can see ”as do online-only games, although for more sensible reasons. After all, how do you assign a difficulty level to a world made up of avatars for real people? You could segregate players with different experience levels into different areas, graded according to their abilities with tougher monsters and tougher spells, but this doesn't really solve the problem ”it just sidesteps it.

Other games have taken a more original approach to the difficulty level problem: self-adjusting games. These games tailor themselves to the player; the more skilled the player, the harder the game gets. Max Payne by Remedy Entertainment is a game that claims to implement this (see Figure 8.21). The only problem that we can see with dynamic difficulty level adjustment is the possibility of abuse of the system. After all, what is to stop a skilled player deliberately playing badly just before he gets to a really tough section of the game so the game will go easier on him when he gets there? After he's through the softened-up section, he can resume blasting with his prior skill level until he comes up to another tough section.

Figure 8.21. Max Payne.

Generally, the standard difficulty levels are implemented by making the enemies tougher, and they are usually applied as a global modifier for all enemies. Another commonly used approach (sometimes in tandem with increasing toughness) is to make the enemies more numerous . For example, on a normal level, the statistics and number for an enemy might be equivalent to the easy level plus 15 percent, the hard level might be equivalent to the easy level plus 30 percent, and the nightmare level might be equivalent to the easy level plus 50 percent. That is to say, on the normal, hard, and nightmare levels, the enemy toughness or density increases by 15, 30, and 50 percent, respectively. Bear in mind that these are just arbitrary numbers plucked out of thin air; they may not be suitable or even applicable to your game. If you are designing a quiz game, for example, the idea of making the opponents 50-percent tougher is absolutely meaningless. Your only option is to grade questions for difficulty ”which is not as easy as you may think, because of subjectivity ”and group them into sets based on the difficulty levels that they have been assigned.

Yet another mechanism is to increase the intelligence of enemy creatures by one means or another. In AI research conducted by Dr. John Laird at the University of Michigan, he was able to demonstrate that shortening the intervals between "looking around" on the part of Quake- bots contributed more to their success as players than any other tactic. They didn't have to have smarter strategies; they just had to have faster reaction times.

Usually, deciding how to design and implement difficulty level settings is not particularly tricky. In most cases, there is either a de facto standard set of guidelines in place that the game designer can draw on, or it's obvious to the designer based on her knowledge of the design. In any case, choosing how to implement difficulty level settings is by no means the most taxing task the game designer faces.