Search Unity

Feedback loops

Discussion in 'Game Design' started by dusthound, Oct 27, 2018.

  1. dusthound

    dusthound

    Joined:
    Sep 18, 2018
    Posts:
    98
    Can someone explain what a feedback loop is?
     
  2. Antypodish

    Antypodish

    Joined:
    Apr 29, 2014
    Posts:
    10,780
    Trivial example will be stretched hand in one direction. lets say straight forward.
    If you put pressure down, by holding a weight, your muscle will try to apply force, to rise arm up, to keep arm steady. If your arm will be lifted by your college, again, your muscles will try apply opposite force, to make arm back to steady setpoint position, to keep arm at level.

    That is feedback loop.


    Or other example



    Imagine simple toilet water tank. It has mechanical feedback closed loop control system implemented.

    Look at Float on the right.
    Imagine you have water in the tank. Float is floating in its Steady State, with 0 cm water level error.
    This is your setpoint (target), to close the valve.

    If you release water with handle (on the left), water level drops.
    While water drops, Float drops too, increasing level error. Error in control system may be positive, or negative.
    Often in closed loop systems, you see negative feedback loop.
    What it means, if my setpoint level for Steady State is 0 cm, and my Float level is 10 cm lower, I got -10 cm error. 0 - 10 gives -10 cm. Hence you try compensate that error.
    You collect feedback level position from Float.
    While Float is dropped, Fill Valve (on the left) is open and water is entering, filling tank to compensate error.
    At this state, you have (level position) error.

    Once water level increases, and your Float start rising, the error is reducing.
    At certain threshold, near error = 0, or greater, valve closes.

    Hence you have feedback loop from Float to the valve.
    You can adjust setopint of Float water level, with Float Adjustment Screw (on the left).


    Does that makes any sense?
     
    Last edited: Oct 27, 2018
  3. JoeStrout

    JoeStrout

    Joined:
    Jan 14, 2011
    Posts:
    9,859
    leftshoe18 and Red-Owl-Games like this.
  4. Antypodish

    Antypodish

    Joined:
    Apr 29, 2014
    Posts:
    10,780
    Good explanation to the case.
    But still same principles as in control system.
    Showed examples are typically open loop feedbacks.
    As per case, it means there is no additional balancing, whether you winning, or loosing.

    But on other hand, good match making for example, has close loop feedback, trying to balance both teams.

    Other example of the loop feedback is leveling in RPG games for example.
    Lets say you keep killing same monster.
    You gain level.
    You keep killing it, and you gain less and less exp for same monster, while gaining levels.
    As the result, you need to move to higher level monsters, to get more xp per kill.
     
  5. dusthound

    dusthound

    Joined:
    Sep 18, 2018
    Posts:
    98
    I still don't get it. And yes, i mean fb loops in game design, not coding.
     
  6. BIGTIMEMASTER

    BIGTIMEMASTER

    Joined:
    Jun 1, 2017
    Posts:
    5,181
  7. MosesGameDev

    MosesGameDev

    Joined:
    Jul 27, 2013
    Posts:
    6
    Feedback loops refer to the effect a certain game mechanic might have on the player, There are two main kinds of feedback loops,1 positive and negative.

    A positive feedback loop rewards the player for performing certain actions, Example: The player gets coins everytime they use a boost pad. the loop is designed to get players to use the boost pad.

    Negative loops affect the player in the opposite way, example the player gains speed while using the boost pad but is 80% more likely to crash into the wall.

    Learn more here:
     
  8. newjerseyrunner

    newjerseyrunner

    Joined:
    Jul 20, 2017
    Posts:
    966
    Here are some examples of feedback loops:

    Negative feedback - used to make it harder for players who are doing well, mostly used to balance party games where the difference in skill between players can be extreme. In Mario Kart, when in 1st place, you mostly get bananas and power weapons are tend to become more common the further back you are. This can be frustrating to good players, so I tend to only like it in short party games, not competitive games.

    Positive feedback - used to reward good players, mostly used to create extra layers of difficulty for players who have already doing well. For example, the more kills you get in Halo 4, the more ordinances you get. It's rewarding for players who do well, but can also causes the best player to also be the hardest to kill. Putting combos together in fighting games is the same thing.

    You tend to want positive feedback to scale back and turn into negative feedback after a while so you get an oscillation effect rather than a snowball effect.
     
  9. BoredBoredBored

    BoredBoredBored

    Joined:
    Nov 4, 2018
    Posts:
    12
    The others have given good explanations for positive and negative feedback in games, and why you want one or the other, or both. However, feedback theory is a lot deeper than than that, to the point of PhD papers involving hundreds of pages of complex math. While you probably don't want to go that deep into it for a game, there are game-worthy possibilities in at least slightly deeper feedback theory.

    For example, feedback systems have a factor called gain: basically how much of a change in output you get from a change in input. The phase of the gain determines whether it's positive or negative. The interesting aspect for games is that depending on the gain (and phase), you can have effects that range from simple (defeating an enemy makes you stronger, and thus more able to defeat other enemies) to very complex (maybe defeating too many enemies too quickly makes the enemies stronger). More complex feedback could make games more interesting, if done right. You'd be constantly assessing the effects of what you're doing, and having to change your strategy. Doing it wrong would result in lots of frustration for the player.

    Consider feedback theory a powerful tool for game design. The more effort you put into understanding it and applying it properly, the better your game will be.
     
    Kiwasi and newjerseyrunner like this.
  10. dusthound

    dusthound

    Joined:
    Sep 18, 2018
    Posts:
    98
  11. Kiwasi

    Kiwasi

    Joined:
    Dec 5, 2013
    Posts:
    16,860
    In game design feedback just refers to the information you give the player. If the player does something good you give them a reward. If the player does something bad you give them a penalty. Over time the player learns to go for the rewards and avoid the penalties.

    In game design one of the important things is to constantly tell the player if they are doing well or poorly. Some of the most frustrating games to play are those which provide no feedback. Some of the best games are those that seamlessly integrate feedback into the play experience.

    Feed back takes many forms across many levels of the game. Successful games do feedback at all levels. Simple feed back like footsteps sounds or click sounds/animations indicate that the player is interacting with the game correctly (or not). Moment to moment game play feed back options include screen shakes, life bars, ammo counts, enemy reactions and so on. Goal oriented meta feedback options include battle summaries, achievements, win screens, experience/leveling and stats screens.

    A loop is a system that affects itself. A feedback loop is a system that takes an output variable and uses it to change an input variable. Classical examples of feedback systems include the water tank on your toilet. Your body also maintains its temperature by feedback loops. When you adjust the shower temperature, you are operating as part of a feedback loop.

    Feedback loops are contrasted with feed forward loops. A feed forward loop measures an input to control another input. This would be equivalent to measuring the temperature in your hot water cylinder to figure out how much to open the taps.

    Loops can be positive or negative. A negative loop attempts to maintain the status quo. If there is a deviation, the loop affects the inputs to drive the system in the opposite direction. Standing in the shower working the taps is an example of a negative loop. If its too hot, you make it colder. If its too cold, you make it hotter.

    Positive loops drive the system in the same direction as the deviation. These tend to drive the system towards extremes. Global cooling in an ice age is one example. The colder it gets, the more ice forms, the more sunlight is reflected back into space, this leads to the temperatures getting colder. Human sexuality is another familiar example of a positive loop.
     
    Red-Owl-Games likes this.
  12. BoredBoredBored

    BoredBoredBored

    Joined:
    Nov 4, 2018
    Posts:
    12
    No, you can design feedback into the game itself. For example, you can have a player's actions affect what is available to other players. The first player to claim a treasure (or powerup or whatever), means that there's one less treasure in the game world for the other players to find. That's feedback, but not 'just information you give the player'. The other players may never even know that that specific treasure existed.

    Feedback loops inherent in the rules or world offer plenty of opportunity for interesting gameplay, especially if the loops interact. They provide possibilities for unexpected, but not completely unpredictable, results. Predicting what will happen when feedback loops are involved can be challenging, but also fun, memorable, etc.