Monday, May 10, 2010

Game Design: Positive and Negative Feedback; Flow Control

 As I've said a few times in the past, I have been reading Rules of Play: Game Design Fundamentals. Lately I have also been applying what I've learned into judging StarCraft mapping contests, like this one, and weaving them into my own map like this one. I've decided to take some of the knowledge and tools I've gained and write a post here about them. First, I want to talk about positive and negative feedback back loops in two different contexts.

The first context I want to talk about them in, is in the study of cybernetics.
Cybernetics deals with the ways a system gauges its effect and makes necessary adjustments. The simplest cybernetic device consists of a sensor, a comparator, and an activator. The sensor provides feedback to the comparator, which determines whether the machine is deviating from its established norm. The comparator then provides guidance to the activator, which produces an output that affects the environment in some way. This fundamental process of output-feedback-adjust-ment is the basis of cybernetics.—Stephen Littlejohn, Theories of Human Communication

 As this quote explains, cybernetics is a study of how a system reacts and makes adjustments to the current state of the system. This should sound rather relevant to video games. This type of automated reaction is used all the time; have you ever played a game that automatically tiers to the user's level of skill? There are two primary types of adjustments that are made upon a system, these two are positive and negative feedback reactions, usually used within a loop.

For example, take a room with a thermostat and a heater. When this room's sensor, the thermostat, reads a certain temperature it can trigger the heater to react. This system can be rigged to perform a negative feedback reaction in which the heater heats the room when the temperature is less than or equal to a certain amount. Assuming the room naturally cools, the room's temperate will oscillate between a few degrees and stay that way, all the while the room's specific state is constantly changing. A positive feedback loop would be if the heater turns on when the temperate is equal to or greater than a specific temperature. In a positive feedback loop, the temperature of the room will spiral up and up once it hits a certain activation point. Both positive and negative feedback loops are essential tools to be used in game design.

Too often I see games designed in StarCraft maps that are too hard or too easy, or too hard or too easy once the player reaches a certain point. I also see games made in which the leader has no way of letting the other players catch up to him, because he has such a great advantage over the other players. I also see the losers in a game have no means of catching up to the leaders. What's the point of continuing play if the outcome of the match has already been determined in the middle of the game because the leading player has a tremendous advantage over the rest of the players? These types of design flaws can be controlled with positive and negative feedback loops. Usually positive feedback loops are harmful to a game's flow (flow will be explained more in depth soon), and should be switched out for negative feedback loops. If a leading player seems to be constantly destroying all the other players in an unfair way, create a negative feedback response in which the system (the game) reacts and hinders the leading player. Similarly, you can give advantages to the players in last place to encourage and help them catch up to the leaders.

A great example of a negative feedback response is in the game Super Mario Cart. In Super Mario Cart there are weapons you gain from running into question marked boxes with your cart. Once you hit one you obtain a random weapon. You have a much greater chance to obtain a powerful weapon when you are near last place in the race, and you have a much greater chance to obtain a weak weapon when you are in first place. This sort of automated reaction the system generates creates a much more compelling form of gameplay, in which all the racers are usually closely pitted against each other neck to neck until the very end of the race. This generates an exciting play in which the outcome of the overall match remains uncertain till the very end.

The second meaning of positive and negative feedback would be rewards and punshiments to a player. Simply put, rewards are great to use to encourage players to make certain choices, and punishments are used to deter certain behavior. There is also a third category known as negative feedback. Negative feedback is not necessarily a punishment, but is used to deter a player from making specific choices. For example say you do not want your game to be focused on fighting, although you want to have an occasional enemy in your game. It turns out that too many players are focusing on finding and fighting enemies rather than experiencing your game how you designed it to be experienced. You do not want to punish the player as to deter them from playing at all, so what you could do instead is implement a clever negative feedback mechanism. You might be thinking that you could just make enemies stronger and therefore make the player want to avoid them -this might just make finding and killing enemies an interesting challenge for the player. Instead you could give the player absolutely no reward for killing the enemy (no points or anything), you could also make enemy encounters not worth the risk (as in making the player restart the level if an enemy kills them, or have enemies be "thieves" where if they hit you you lose items). If your goal is to make enemies something the player wants to avoid, give the player a reason to avoid them. If you find that fighting these enemies is truly fun for the players, you could also switch your game to a fighting one; it's your design, you decide.
Flow; flow would be the fine balance between difficulty in a game and the player's skills. Skills do not just include physical coordination; skills also include knowledge of the game's workings, knowledge of the other player's skill level, physical coordination, intelligence level, and response time. There are more than likely many many more forms of skill which I did not mention, but these should give a general idea. The difficulty of the game is pretty self-explanatory. The flow channel of a game is a narrow isle where the difficulty of the game is perfectly balanced and matched to the skill level of the player. Here is a diagram of the flow channel graphed with challenge over skill level:

 As you can see, it can be hard to create a flow in gameplay for all different types of players. One way that game designers achieve balanced flow is to apply positive and/or negative feedback loops to a game's processes.

Here is an interesting excerpt from the book Rules of Play that talks about the design process for a game made by two designers:
In a wonderful essay published on, Jesse Schell and Joe Shochet of Disney Imagineering write about the process of designing Pirates of the Caribbean-Battle for the Buccaneer Gold, a game "ride" where a group of players stands on a motion-platform pirate ship surrounded by video projections. During the game, one player steers the ship while the other players operate a number of cannons, firing at monsters, forts, and enemy vessels. Pirates of the Caribbean is designed as a condensed five-minute experience, and it was essential that players feel properly challenged at every moment of the game.
In their design analysis, Schell and Shochet detail a number of design problems that had to be overcome in order to maximize player enjoyment. For example, during playtesting they identified as a problem the fact that the player steering the ship could take the ship to what they call "dull places," leading to a less engaging experience for all of the players. In the selected quotes below, Schell and Shochet outline some solutions to this problem:
Architectural Weenies: "Weenie" is a phrase coined by Walt Disney himself. It refers to the technique used on movie sets of guiding stage dogs by holding up part of a sausage… In the case of Pirates, [there are] three main "weenies," one for each island: a volcano, an enormous fort, and a plume of smoke coming from a burning town. No matter which way the boat is facing, at least one of these "weenies" is in view. Since the coolest action takes place at the islands, [we wanted] to guide the captains to go there.
Guide Ships: Since the short-term goal of the game is to fire on other pirate ships, captains strive to get near these ships so that their gunners can get a clear shot. Many of the ships in the Pirates world are "on their way" to the islands mentioned above. Many captains, in just trying to stay near these ships find that just as they have destroyed the ship, they have arrived at one of the islands, without even trying to get there.
Sneak attacks: What if the captain ignores the guide ships? Even if he heads toward one of the "weenies" it might mean as long as a minute during which the gunners have little to shoot at. For this reason, [we] created special "sneak attack" ships that "magically" appear behind the players' ship, and quickly pull up alongside, when no other boats are in range.
The Waterspout: This was [a] nickname for [a] "last ditch" forcefield that surrounds the game play area.If a captain tries to sail out of the main game play area and out to open sea, they hit the forcefield, and the ship is "magically" pointed back to where the action is. The few guests who see this don't even realize that anything unusual has happened. They are just pleased to have their boat going somewhere cool.
Schell and Shochet are thinking in very experiential terms, using clever techniques to subtly guide player action in meaningful directions. At the time of its release, Pirates was a very high-tech production, featuring real-time 3D graphics, physically engaging cannon-firing interfaces, and a large motion platform to simulate a pirate ship rocking on the waves. Often in these instances, a desire to "properly" simulate a coherent 3D space or "correctly" output logical behavior for computer-controlled characters overshadows the design of the actual play experience. But Schell and Shochet had no hesitation in making pirate ships "magically" appear to guide the player, or abandoning "realistic" physics to have the player's ship turn on a dime to facilitate navigation. As they put it,"By choosing to be less concerned with reality and more concerned with what was fun, we created an experience that…is easier to adapt to, quicker to learn, and is a better show." In game design, player experience should always trump so-called "realism."

Boredom and anxiety, as game design watchwords, are wonderful because they speak directly to player experience. As you shape and sculpt your players' pleasure, you are guiding them between the Scylla and Charybdis of anxiety and boredom.This task is made all the more difficult because, as we know, the experience of play can only be indirectly designed. How do you create a set of rules that maximizes the play of pleasure for your audience?

The designers of this game implement two negative feedback loops. Did you see them? Onewould be the sneak attack ships. These ships react to the players being all alone at seas, and then appear out of nowhere to create an exciting experience. The other would be the water spout; the water spout reacts to the players entering the boundaries of the map, then spins them in the correct direction, thus keeping the game in a specific state. The guideships and weenies would be more a form of positive feedback in which encourage the player to make specific decisions. The guideships encourage the players to travel to the key focul points of the game while the weenies create enticing visual scenes that players are likely to travel to. Both of these provide the players with a reward and a reason to take specific actions.


  1. I think you're quite right about the dangers of positive feedback loops; in a game with a "strong" positive feedback loop, there's simply no point in playing any longer once one player has taken a decisive advantage. I think you're a bit too keen on negative feedback loops, though. Just as it is frustrating to have no chance at a comeback, it is also frustrating if all the effort that you put into gaining an advantage over your opponents comes to naught because the game balances it out with negative feedback. Equally frustrating is watching the game give some advantage to a player who has done nothing to earn and deserve it; seeing inferior play rewarded generates a feeling of unfairness. Could you imagine how awful chess would be if the positive feedback loop created by attaining a material and positional advantage were somehow removed and replaced with a system that rewarded the loser?

    In my experience, the most fun games are ones with "weak" positive feedback loops. With a "weak" loop, the prospect of your own power spiraling upwards to victory gives you something to fight for, while the prospect of the same happening for your enemy gives you something to fight against, creating excitement and tension, respectively. The sides fight in an effort to gain "momentum" leading to victory; where excessive negative reinforcement automatically cancels out such momentum, there's nothing to fight for. In a properly balanced game, there are sufficient options for gaining positive reinforcement that even a more powerful player cannot completely dominate a weaker player; the weaker side almost always has the ability to shift the game's momentum if it succeeds in out-playing the stronger side. Gaining strength gives you a real and significant advantage, yes, but not one that is deterministically decisive. It's only when the game is designed such that there is *nothing you can do* against a stronger opponent, that is, such that a brute strength advantage automatically closes off all options for fighting back, that the positive feedback becomes a problem.

    Do note that these remarks are primarily directed towards games with players in opposition. There may be more room for negative feedback in games where there is a single player, or where the players are on a team, so your milage may vary in applying the above comments to such projects.

  2. You are absolutely correct. There is no magic formula when designing video games, and each game is different. I may have sounded keen on negative feedback loops, though this is only because these seem to be more commonly used. Interesting games can arise when positive feedback loops are employed in a meaningful way. For example, StarCraft 2 melee was designed so that 1v1 matches wouldn't last much longer than 20 minutes. I was told from watching some analysis videos that players were designed to be able to exploit the other player's weaknesses in strategy and use those to quickly gain victory. This in a way could be seen as a positive feedback loop where the player who makes a major mistake immediately becomes much less likely to pull a comeback and come out victorious.

    It is important to understand the tools used in game design, and often times challenging paradigms leads to interesting design innovations. Thanks for the comment.


Note: Only a member of this blog may post a comment.