next »

[Alec S.]

So I wanted to start a discussion on what I feel is a very important concept in design:  Feedback Loops.

Feedback loops in games are mechanics that push the game-state in different ways depending on the current game-state.  For example, a game that got harder the better you're doing, or easier the worse you're doing.

Feedback falls into two categories.  

Positive Feedback Loops push values to their extremes.  Often this translates to bonuses for having high health, and penalties for having low health.  If you start doing well, the game compounds your advantage, but if you are doing poorly, it compounds the difficulty to recover.  An example of positive feedback would be the ability to shoot your sword in Legend of Zelda when you are at full health.  

Negative Feedback Loops pull values to the center.  This can translate as giving the player a boost if they are losing, or making the game more difficult if they are winning.  In the board game Settlers of Catan, the thief would be an example of a negative feedback loop.  If you have over 7 resources when the thief moves, you lose resources, however if you have under 7, you are unaffected.

Or, as Muz put it (later in the thread)
 

Feedback means that the output goes into the input (is 'fed back'). Positive here means that the output adds to the input (goes higher), negative means the output subtracts from the input (goes lower)

Feedback can exist not only in the mechanics, but also in the level design.  Say you have a Team Fortress type game.  The middle area is going to be fairly even, with no side having an environmental advantage, however, the team bases are likely to have defenses that favor defense.  This is negative feedback.  If one team starts winning, they move into the opposing teams turf, and the losing team will be able to defend more effectively.  

So basically this is a thread to talk about feedback loops in games.  What are some situations in games that use positive/negative feedback effectively?  What are situations where feedback loops make the game less enjoyable.  Have you put feedback loops into your own games either intentionally or unintentionally?

[Azure Lazuline]

Generally, I try to stay away from positive feedback, especially when it involves something happening when the player is losing. Especially in a competitive game, you would want there to always be a chance of comeback, so negative or no feedback would be ideal. One thing I tend to use in my games is giving the player a "last chance" no matter what. If something would kill the player in Copy Kitty, they are instead reduced to 1HP (unless they already have 1HP of course, then they die). I think this counts as negative feedback even though it's not a constant thing, just a single point in time. I remember a fighting game where blocked damage can never kill you even though it still hurts a tiny bit, so the opponent needs to get a clean hit to finish you off - that's kind of similar.

Going with the "last chance" thing, I make the player's damage hitbox smaller when they're on the last hit, but there's no visual indication of this. The end result is that players feel badass when they suddenly start dodging a bit better when it really matters, without feeling like they're cheating at all, and without being exploitable either (like an attack boost at low health would be).

I think I only have one positive feedback loop in the game, which is relating to the combo meter. It rises one notch for each enemy killed and drains over time, pretty standard. But when it fills all the way, for as long as you can keep it filled (which gets harder over time), you get faster firing and infinite ammo. This encourages the player to be really aggressive and keep moving, which is what I'm looking for. But when they mess up (or keep it going too long) and lose the combo, it's almost a relief, and there's a drop in the action when they can choose weapons more carefully or relax a little. (Assuming they didn't lose it when they were already doing badly and there are a bunch of enemies on-screen.)

[JWK5]

Super Puzzle Fighter 2 Turbo did a really good job at applying both in tandem.

Playing aggressively was good because you buried your opponent and the more you kept it up the faster things got dumped on them, compounding and escalating their approach to defeat. Any mistakes you make almost always compound into a chain of worse mistakes due to running out of room as your stack piles up. This is the positive feedback loop.

Playing defensively is also good and often nearly losing and nearly winning are the same thing in this game. The more that gets dumped on you the more ammunition you are building to counter attack and your opponent could have you nearly stacked to the ceiling only to find you unleash a super attack that defeats them in one shot. This is the negative feedback loop.

The way the two loops work continuously throughout the whole match keeps each match tense and exciting. You're never really totally sure if you are winning or losing, anything you do to up your odds and turn things in your favor can be used against you if your opponent is quick and clever enough.

I'd have to say it is one of the best designed games there is.

[Muz]

Positive feedback loops have their place. Sometimes the player has nearly won, like taken over 60% of the map or whatever. Instead of letting the players slog through another 7 hours to settle the game, a positive feedback loop gives the winning player an excess advantage that lets them finish the game in 1 hour or so.

In a multiplayer game, activating this would also make all the remaining players gang up against the final player for a final epic battle.. instead of spending the rest of the 7 hours fortified and waiting for the guillotine. They'll either defeat the winning player with a lot of teamwork and keep the game going, or die a quick, fun death.


Negative feedback works great in Settlers of Catan and other games where it's vital to maintain a balance of power.

But I really hate it when applied to FPSes or other skill based games. The game shouldn't be easier just because you suck; that just makes victory less satisfying.

[Alec S.]

Yeah, positive feedback loops are good for that feeling of overwhelming victory.  That point in the game where you've essentially won and all that's left is to sweep up the rubble.  Strategy games would be fairly anticlimactic without this.  

And, as has been said, a combination is good.  Say the gamestate goes from 0 ---- 100 where 100 is you winning and 0 is the enemy winning.  There could be negative feedback loops that are in effect from 10-90, and positive feedback loops that take place below 10 and above 90 which help bring the game to it's conclusion.  The game will get harder as you continue to win, until you reach that peak at 90 at which point your victory will start compounding on itself.

Bringing the level design aspect back into it, say you have a capture a capture the flag game.  The deeper you get into the enemy base, the harder it gets, until you finally get the flag and escape (in fact, most Capture the Flag games add another negative feedback loop by disabling the player in some way while they carry the flag).  Once you get back to your own base, it's almost a foregone conclusion that you will score the point, and it becomes something of a victory lap.  In terms of player experience, it imbues your base with a feeling of home, and it's a relief to reach it.  There is also still the small chance that the opponent could turn it around inside your own base, but that would be a major upset, and quite dramatic, which is also an advantage of this sort of system.

[thatshelby]

I don't really have much to add other than an example of a negative feedback loop in Mario Kart.

Basically any item block. If you're somehow unfamiliar, the players that are towards the end of the pack, losing, get better items, while the players in front, especially the player in first place, get really bad items.

Mario Kart would be incredibly boring without this game dynamic.

[Alec S.]

Oh, yeah, the Mario Kart games are absolutely full of negative feedback loops.  One of the most powerful items (blue shells) automatically target and home in on whoever is currently in the lead.  Most of the other items can also only hit people in front of you.  The more in the lead you are, the more you are vulnerable to attack.  The negative feedback loops keep the cars for the most part packed together, making the game a lot more exciting. 

I'm not sure if Mario Kart games use rubber band AI (which would be a negative feedback loop programmed right into enemy behavior), but they've managed to make a game where the mechanics themselves serve the same purpose, so that even human players will tend to be bunched up, leading to more intense games.

[Gimym TILBERT]

Player ahead don't get bad items, all his items are defensive one, same for lagging player, they get items to attack. Think about it, fake block, banana, green shell, etc... all can block attack from behind OR slow down possible rivals, lagging player would need to protect from who? Notice that three shell tend to happen for middle player, they have both offensive and defensive power. The only real imbalance item is the blue shell, it's main purpose is to break a little the monotony and give a kind of adrenalin rush, but even this has some counter measure at high level play. The idea is that better player get more challenge when they lead the game while worst player get some help.

I wish mario kart had clearly labelled rank and clearly match item probability to ranking. I would also ban the blue shell from the last lap entirely and have a more tier of items (or items use) tied to lap increase.

[Azure Lazuline]

Mario Kart kind of overdoes it, though. The best strategy in most games is to purposely get in last place to get good items, then use them in the last 30 seconds to get back up to 1st since everyone is still bunched up. If you try to play well, then there's that 30% chance or so that someone hits you in the last few seconds and now you're in 6th. (And yes, it does use rubber-band AI in addition to the negative feedback mechanics, and the computer racers can actually go faster than the player in some games if they're far enough behind.)

What would be interesting is a racing game that uses all those negative feedback loops, but grades you based on how much time you spend ahead, not entirely what place you're in at an arbitrary point in time. Kind of like King of the Hill or something.

[msilver]

I would never intentionally drive in a lower place in Mario Kart. The trick is to try to put enough distance between you and second place so that the blue shell isn't enough to close the gap. I wouldn't change that game a bit, I think its great.

I was just recently playing Civ 5 and that, like all Civs, is a pretty big positive feedback loop. More cities means more income means more buildings and research means more productive cities means more income and so on.

I usually I like my single player games positive and my multiplayer games negative.

[edh]

A better example of positive feedback would be Risk or Monopoly, or just about any RTS

[Alec S.]

Yeah, positive feedback loops are almost inherent in games of resource management.  It's the whole concept of "you have to spend money to make money".  The more resources you have, the more you can build that will give you access to even more resources.

That's why some strategy games will then add in some negative feedback to balance it out.  For example, upkeep is a form of negative feedback, and can change a game from simply amassing resources, to a game of carefully managing those resources.  If you play it smart, the positive feedback loop will out-weigh the negative feedback, but if you overextend, the negative feedback will win out.

[Gimym TILBERT]

On terminology, I like better slippery slope (positive feedback loop) or rubber band (negative feedback loop) because I had them all mixed up (positive feedback isn't a "positive" gameplay feature), also it mixed things with information feedback which is my primary use of the word feedback.

[Alec S.]

Well, the terms are preexisting and are used in science (example) as well as to describe games (they are very useful terms for any system defined by rules).  Positive just means that any change is amplified through the feedback, and negative means that any change is negated.  But slippery slope and rubber band are good ways to visualize it.   The main point of the terms is to focus on the fact that information from the system is feeding back in on itself and effecting the system.

[Muz]

Lol, yeah, I'm from an engineering background, so positive/negative feedback made perfect sense to me. The meaning applies well to control systems (maybe economics too).

Feedback means that the output goes into the input (is 'fed back'). Positive here means that the output adds to the input (goes higher), negative means the output subtracts from the input (goes lower)