How Games Use Feedback Loops

Pyre, the party-based RPG from Supergiant, looks unusual at first because its central conflicts are not normal turn-based battles. The characters face off in basketball-style contests called Rites, where positioning, scoring, and team composition matter as much as the RPG progression around them. It is a strange structure, and that is part of why the game is so useful for thinking about feedback loops.

A feedback loop is a system where the output is fed back into the system as an input. In game design, that usually means the player's success or failure changes the likelihood of future successes and failures. A win might grant stronger tools, more options, or a better position. A loss might remove resources, weaken the player, or make the next attempt harder.

Those loops come in two broad forms: positive feedback loops and negative feedback loops. The names are not moral judgments. A positive loop does not mean "good", and a negative loop does not mean "bad". They describe the direction of the system. Positive loops amplify change. Negative loops push against change and try to keep the game closer to equilibrium.

Pyre stands out because it uses both kinds at once. It lets success and failure matter, but it also keeps either direction from taking over the whole campaign. That balance is the heart of good feedback-loop design: deciding when a result should compound, when it should stabilize, and when a loop should be dampened or removed entirely.

Positive feedback loops reinforce successes with more successes, and compound failures with more failures. A clear multiplayer example is the kill streak system in earlier Call of Duty games, especially titles like Modern Warfare. Getting several kills in a single life grants special bonuses, such as revealing enemy positions on the radar or calling in an airstrike or attack helicopter.

That means the players who are already doing well receive tools that help them do even better. The loop creates a snowball effect, the familiar "rich get richer" outcome where one strong moment turns into an even stronger position. In the right context, that can feel exciting. It rewards mastery, creates a sense of momentum, and gives the match a sharp upward curve.

Positive loops can also work in the other direction. In chess, losing a piece means having fewer tactical options. Fewer options can lead to worse positions, and worse positions can lead to more lost pieces. A single mistake does not automatically decide the match, but the board remembers the loss. The player's weakened state becomes an input into every move that follows.

This is why positive loops are powerful and dangerous. They make outcomes matter beyond the instant where they happen. A win is not just a point on the scoreboard. It can become better positioning, stronger resources, more control, or a faster path to the end. A loss is not just a lost exchange. It can become reduced agency and a narrower route back into the game.

Negative feedback loops do the opposite. They balance out successes and failures by pushing against the current direction of the game. The classic example is item distribution in Mario Kart. Players near the front of the pack tend to receive weaker items, such as green shells and banana peels. Players near the back are more likely to receive powerful tools, such as the Bullet Bill, the golden mushroom, or the blue shell that targets the leader.

The effect is simple: winners are restrained, and losing players are given a way back into the race. If positive feedback loops compound changes in the game state, negative feedback loops counteract those changes. They reduce runaway leads, invite comebacks, and keep the contest from becoming decided too early.

That does not mean every game needs them. In Tekken, losing one round does not give the player an automatic advantage or disadvantage in the next round. The game resets that layer of the contest. The new round starts from a clean competitive state, and that is part of the design.

Other games have loops because the rules naturally create them. In Splatoon, the team with more ground covered in ink also gets more hiding places and faster travel. The advantage comes directly from the game state. The team that controls the map can move through it more easily, which can help them keep controlling the map. Whether the designer adds a comeback tool or lets that pressure stand is a design decision, not an automatic rule.

Positive and negative loops both have real uses. Positive feedback loops can help push a game toward its conclusion once one player or one team starts to win. Without some form of compounding advantage, a match can get stuck in stalemate or drag through a long ending where the result is obvious but the game has not caught up yet.

Team Fortress 2 uses this idea in its capture-point structure. A team that controls more capture points can receive shorter respawn times, letting it compound its victories and push toward a decisive win. The loop helps convert map control into pace. It tells the match to move forward instead of asking the winning team to prove the same advantage forever.

Negative feedback loops are useful for a different feeling. They are especially valuable in party games, where the goal is often to keep everyone involved and to create dramatic swings. Comebacks, reversals, and near misses can make a group game more exciting than a clean demonstration of who is strongest from the first minute.

Both kinds of loops need careful balance. Positive loops can be miserable for weaker players who get trapped in a death spiral or cannot catch a runaway leader. Negative loops can feel unfair to successful players who keep seeing their victories taken away. A loop is not good because it exists. It is good when it supports the experience the game is trying to create.

Feedback loops are easiest to notice in multiplayer because fairness and balance are so visible. But single-player games use them constantly. Some genres are built around positive loops. In many JRPGs, fighting monsters gives experience points, experience points increase the player's level, and higher levels let the player defeat monsters more efficiently. The loop creates the addictive feeling of always getting stronger.

That loop is usually balanced by the rest of the game. Enemies become harder as the player moves forward, and failure usually resets the immediate attempt. If the player dies, they retry. The failure is not always carried forward into the long-term state, so the positive loop mostly expresses itself as growth and momentum.

Persistent games are different. XCOM does not simply erase a failed mission. The campaign continues, and any soldiers who died during the failed expedition are gone forever. The result of one mission becomes the input for the next mission. That persistence is one reason the game feels so tense, but it also creates a severe snowball problem.

Gunpoint designer Tom Francis described the issue clearly. If troops survive a mission, they get stronger, tougher, and more capable, which makes them more likely to survive future missions and become stronger again. If they die, they are replaced by weak rookies, who are more likely to die and be replaced by more weak rookies. Success compounds, and failure compounds.

That positive loop becomes painful at the extremes of XCOM's difficulty curve. A struggling player cannot easily stop and train new troops in a safer place because the campaign does not simply offer low-level enemies on demand. The game keeps moving. The weak squad has to face the current threat, not the threat it wishes it were ready for.

At the other extreme, veteran troops become stronger and harder to kill. The longer they survive, the more likely they are to keep surviving. That can produce powerful steamrollers, which sounds satisfying until the tension fades. If the best soldiers are too safe, the game loses some of the danger that made them valuable in the first place.

Francis suggested several smart ways to address that kind of problem, such as low-profile missions that force the use of rookie soldiers, or making squad size vary by mission so the player cannot always take the maximum number of units into every fight. The underlying point is not that XCOM is broken. The point is that persistent success and persistent failure can become self-reinforcing very quickly.

When a single-player game carries consequences forward, it has to decide how much of the previous result should shape the next challenge. If the answer is "all of it", the game may create a memorable campaign, but it may also push some players into an inescapable spiral and others into a state where challenge quietly disappears.

Pyre has a structure that could have created the same kind of runaway problem. Characters level up and become more powerful when they win. The game always moves forward, so losing a match does not simply rewind the campaign. That combination could have encouraged the player to use the same team members forever, watching them become so strong that nothing remained challenging.

It could also have pushed losing players into the opposite spiral. A few bad matches might leave the team underpowered and unable to recover. Instead, Pyre counteracts those positive loops with negative feedback loops. Losing a match still grants some experience, so failure is painful without being too destructive. There is no permanent death for the whole roster after a bad Rite, so the player is not massively damaged for future contests.

On the success side, Pyre uses its liberation system. The game's characters are trapped in a purgatory-like place called the Downside, and winning special liberation Rites lets a team leader leave, have their crimes forgotten, and return to the Commonwealth. Mechanically, that means the player periodically sends away some of the most powerful characters, because only characters of high enough rank can be liberated.

It is a kind of permanent removal, but framed as success rather than punishment. The strongest characters leave because that is the goal of the story and the campaign. The result is that the player cannot simply lean on the same dominant team forever. The game keeps asking the player to level and use more of the roster.

The combined effect is that mistakes become road bumps, not tipping points. Losing hurts, but it does not necessarily send the player into an inescapable death spiral. Winning matters, but it does not let the same strongest units solve the rest of the game without pressure.

By constantly removing top units from the roster, Pyre encourages careful use of the whole team. The player may end up fighting the last and most difficult battles not with the obvious best characters, but with the less-used characters who have not yet been liberated. That creates tension from the campaign structure itself, not from a sudden artificial difficulty spike.

This is the elegant part of the design. The negative loop does not merely slap the leader for being successful. It turns success into a new constraint that matches the fiction, the goal, and the emotional arc. The player is not told that winning was wrong. The player is told that winning has consequences that matter.

That distinction is crucial because one big problem with negative feedback loops is mixed signals. Games usually work hard to encourage a certain style of play. If the game then punishes someone for playing well, it can seem to contradict its own teaching. The player starts asking whether they should deliberately make mistakes to avoid triggering the system.

Resident Evil 4 is a useful example of that signaling problem. The game uses dynamic difficulty, which is effectively a negative feedback loop. If the player is doing well, the game can become harder. If the player is struggling, the game can ease off. Revealing that system too openly would risk telling good players to make mistakes, or telling struggling players that their progress is being softened behind the scenes.

Pyre avoids that problem because it is still better to win than to lose. The player receives more from victory, and liberation is not presented as a penalty. It is the stated goal of the game, tied to character arcs and drama. Sending away a powerful character can be painful precisely because it feels like the right thing to do.

Pyre also uses negative loops inside the Rites themselves. When a character scores, that character sits out the next play, giving the losing side a temporary numbers advantage. This is similar to real basketball, where the team that was scored on gets the ball. The scoring team gets the point, but the next moment gives the other side a practical opening.

That kind of stabilizer works because it does not deny the success. The point remains. The scoring player still achieved something. The loop simply prevents the same advantage from immediately cascading into total control. It keeps the contest alive without making success feel pointless.

The larger lesson from Pyre is that negative feedback loops can cancel out the most destabilizing properties of positive feedback loops. The goal is not to erase momentum. Momentum is often fun. The goal is to keep momentum from becoming inevitability before the game has finished creating interesting decisions.

Civilization shows a similar dynamic. Conquering another nation gives the player more land, which can create a snowballing positive feedback loop. A larger empire can produce more resources, more units, and more opportunities to keep expanding. But the game pushes back through unhappy citizens, higher costs for running a large empire, and diplomatic pressure from other nations.

That pressure matters most early, because positive feedback loops often magnify over time. The first few turns, kills, rounds, or decisions can have an enormous impact on the eventual winner. If the game lets an early advantage compound without resistance, the later parts may become less about playing and more about waiting for the result to formalize.

Some negative loops can even emerge from players rather than rules. In Civilization multiplayer, everyone may decide to gang up on the leader because the leader is becoming too dangerous. The system does not need to hide the correction. The social layer creates a stabilizing force because every other player has a reason to stop the runaway empire.

Negative loops are not the only answer to runaway positive loops. Sometimes the better move is to dampen the loop so its effects do not compound as aggressively. Call of Duty changed this in later entries. In earlier games, kill streak bonuses could count toward future kill streak bonuses, creating a wild chain where one reward could help trigger the next reward and then the next.

That was changed in Call of Duty: Black Ops, where kill streak bonuses no longer counted toward more kill streak bonuses. Players still had rewards for doing well, but they needed to keep earning them through direct play. The loop still existed, but its most explosive property was reduced.

Mario Kart 8 offers a different kind of correction with the super horn, which gives the leading player an effective way to destroy blue shells. That does not remove the negative feedback loop from the game. It gives the successful player a way to answer its most frustrating edge case, which can make the stabilizing system feel less arbitrary.

Sometimes the cleanest solution is to remove a feedback loop altogether. In Devil May Cry, a high score at the end of a level can grant more red orbs, which are used to buy better weapons and health items. That rewards the best players with tools they may not need, while struggling players have less money for the items that could help them. The reward is understandable, but it risks pulling the difficulty curve in the wrong direction.

Donkey Kong handles this kind of reward more cleanly with its difficult-to-collect KONG letters. Collecting them all does not grant coins, extra lives, or an easier path through the main game. It unlocks extra difficult bonus levels. That is a suitable reward for players who have already proven they are good at the game, because it gives them more challenge rather than more basic power.

Feedback loops are a key part of many types of games: JRPGs, persistent strategy games, online shooters, arcade racing games, party games, and more. They can create growth, pressure, tension, drama, comeback potential, and decisive endings. They can also create death spirals, runaway winners, unfair punishments, or mixed signals that teach players the wrong lesson.

The central question is not whether a feedback loop is positive or negative. The question is what the loop does to the player's next decision. Does success create a satisfying path toward victory, or does it make the rest of the game obvious? Does failure create useful pressure, or does it make recovery impossible? Does a comeback tool keep everyone engaged, or does it make strong play feel meaningless?

That question has to be answered before tuning numbers. A designer can lower an experience reward, shorten a respawn timer, weaken a comeback item, or remove a bonus entirely, but those changes only help if the loop's job is clear. First decide whether the loop should accelerate the ending, protect the losing player, preserve long-term consequence, or create a dramatic reversal. Then tune the strength of the loop around that job, and watch whether players trust the result.

Good loop design starts by naming the feeling the game needs. A competitive match might need a positive loop that ends the game before it drags. A party game might need a negative loop that keeps weaker players in the story. A persistent campaign might need both: enough consequence for choices to matter, and enough correction that the campaign does not collapse after one bad night.

Pyre is memorable because it makes those corrections feel like part of the game rather than a hidden patch over the game. It lets characters grow, then asks the player to let them go. It makes losing matter, then softens the damage enough for the story to continue. It shows that feedback loops are not just math in the background. They are one of the ways a game decides what success, failure, and momentum actually mean.