A win that feeds the next win

A feedback loop is any result that comes back as the next input. Win a fight, and you may hold a better gun for the fight after. Lose, and the next room gets harder while you are already behind. Games use both, and they feel completely different. One of them is a power fantasy. The other is a spiral.

The useful question is not whether loops are fair in the abstract. It is when this game should let success compound, and when it should put a lid on. A roguelike that snowballs is a different promise from a duel that resets every round. Leave the lid off a competitive game and the first good minute decides the match. Put a lid on a fantasy about becoming unstoppable and you have broken the fantasy. Name the promise first. The loop is how you keep it.

A feedback loop is any system whose output comes back around as an input. In a game, that means the result of what you just did changes how likely you are to succeed at what you do next. Win, and you may end up with stronger tools, more options, or better position. Lose, and you may be down resources, down power, or facing a steeper next attempt.

Pyre, Supergiant's party-based RPG, is a good place to start looking at that, partly because its core conflicts are not conventional turn-based fights at all. The cast competes in basketball-like contests called Rites, where team composition, positioning, and scoring carry as much weight as the RPG progression wrapped around them. The structure is unusual, and the unusual parts are exactly what make the loops easy to see.

Loops sort into two families: positive and negative. Neither label is a verdict. A positive loop is not the pleasant one and a negative loop is not the mean one; the words describe direction. Positive loops amplify whatever change has already happened. Negative loops push back against it and drag the game toward equilibrium.

What makes Pyre worth studying is that it runs both at once. Results are allowed to matter, and neither direction is allowed to own the campaign. That is the real work of feedback-loop design: choosing when an outcome should compound, when it should be steadied, and when the loop should be softened or cut out of the game altogether.

A positive loop turns success into more success and failure into more failure. Kill streaks in the earlier Call of Duty games, Modern Warfare among them, are the clean multiplayer version. String together a few kills without dying and the game hands you bonuses: enemy positions on the radar, an airstrike, an attack helicopter.

So the reward for playing well is equipment that helps you keep playing well. That is the snowball, the familiar case of the rich getting richer, where one good moment converts into a stronger position and then a better moment. Used deliberately, it can feel great. It pays out for mastery, it builds momentum, and it bends the match into a steep upward curve.

The same loop runs downhill just as easily. Lose a piece in chess and your tactical options narrow. Narrower options tend to produce worse positions, and worse positions tend to cost more pieces. One slip does not settle a match on its own, but the board keeps a record of it. Your weakened state is an input into every move you make afterward.

That is what makes these loops both useful and hazardous. They let a result matter beyond the moment it occurred in. A win stops being a number on the scoreboard and turns into position, resources, control, or a shorter road to the end. A loss stops being a single bad trade and turns into less agency and a longer road back.

A negative loop counteracts results instead of amplifying them, leaning against whichever way the game is currently tipping. Mario Kart's item distribution is the textbook case. Race near the front and the boxes tend to hand you weaker things, green shells and banana peels. Fall to the back and your odds improve for the Bullet Bill, the golden mushroom, or the blue shell that hunts down whoever is leading.

The outcome is straightforward. Leaders get held back, trailing players get handed a route back into the race. Where positive loops magnify a change in game state, negative loops work to undo it. They shrink runaway leads, keep comebacks live, and stop the result from being locked in early.

Not every game wants that. Drop a round in Tekken and the next round starts you with no built-in advantage or handicap; that layer of the contest is simply reset. The following round begins from a clean competitive state, and that reset is a design choice in its own right.

Elsewhere the loop is just what the rules produce. Splatoon's ink coverage is also cover and also travel speed, so the team holding more ground moves faster and hides better. The advantage falls straight out of the game state: controlling the map makes the map easier to keep controlling. Whether a designer answers that with a comeback tool or leaves the pressure alone is a decision, not something the rules settle automatically.

Both families earn their place. Positive loops are how a game gets to its ending once somebody has started to pull ahead. Strip out every compounding advantage and matches stall into deadlock, or grind through a long tail where everyone can see the result and the game itself has not caught up.

Team Fortress 2 builds that into its capture points. Hold more of them and your respawn timer shortens, so the team that is winning gets to press the advantage toward a decisive finish. The loop turns map control into tempo. It tells the match to move rather than making the leading team demonstrate the same edge over and over.

Negative loops are aiming at a different feeling. They earn their keep in party games above all, where the point is keeping everybody involved and manufacturing swings. Reversals, comebacks, and near misses tend to make a room full of players happier than a tidy demonstration of who was strongest from the opening minute.

Either one needs tuning. A positive loop can be wretched for the weaker player caught in a death spiral or chasing a leader they will never reach. A negative loop can feel like theft to the player whose wins keep getting clawed back. No loop is good merely by existing. It is good when it serves the experience the game is reaching for.

Multiplayer makes loops obvious because balance and fairness are on display, but single-player games lean on them constantly, and some genres are built out of them. The JRPG shape is a positive loop end to end: fight monsters for experience, spend experience on levels, use levels to clear monsters faster. That circuit is where the pull of always getting stronger comes from.

The rest of the game usually holds that loop in check. Enemies scale up as you advance, and a defeat generally wipes only the current attempt. Die, and you try again. Because the failure is often not carried into the long-term state, the loop mostly surfaces as growth and momentum rather than decline.

Games with persistence behave differently. XCOM does not quietly undo a botched mission. The campaign rolls on, and the soldiers who died on that expedition are gone permanently. One mission's outcome is literally the next mission's starting input. That is a large part of where the tension comes from, and it is also a serious snowball problem.

Tom Francis, the designer of Gunpoint, laid the mechanics of it out plainly. Survive a mission and your troops get tougher, better, more capable, which makes them likelier to survive the next one and improve again. Die, and rookies replace them, and rookies are likelier to die and be replaced by more rookies. Success stacks on success; failure stacks on failure.

The pain shows up at both ends of XCOM's difficulty curve. A player who is falling behind cannot conveniently retreat somewhere gentle and train replacements, because the campaign does not serve up weak enemies on request. The clock keeps running. A thin squad has to meet the threat that exists, not the one it would be ready for.

At the opposite end, veterans get stronger and harder to kill, and the longer they last the likelier they are to last longer still. That can produce a steamroller, which sounds like a reward until the tension drains out of it. Soldiers who are too safe stop carrying the danger that made them worth caring about.

Francis floated some sharp fixes for that class of problem: quiet, low-profile missions that require rookies to be deployed, or varying squad size mission to mission so the player cannot always field a full complement of favorites. None of that says XCOM is broken. It says that persistent success and persistent failure both start feeding themselves very quickly.

Any single-player game that carries consequences forward has to decide how much of the last result should shape the next challenge. Answer "all of it" and you may get a campaign people remember for years, and you may also strand some players in a spiral they cannot climb out of while quietly removing the challenge for everyone else.

Pyre was set up to hit exactly that runaway problem. Characters gain levels and grow more powerful by winning, and the campaign always advances, so a defeat does not rewind anything. Left alone, that would push a player to field the same three favorites permanently until nothing on the field could threaten them.

The reverse spiral was available too: a run of losses leaving an underleveled team with no way to climb back. Pyre answers both by running negative loops against its own positive ones. A loss still pays out some experience, so failing stings without gutting you, and a bad Rite does not wipe the roster, so the next contest is not doomed by the last one.

The counterweight on the winning side is liberation. The cast is stuck in a purgatory called the Downside, and taking a special liberation Rite lets one team leader go free, have their crimes struck from the record, and return to the Commonwealth. In mechanical terms, the player keeps sending their most powerful characters away, because only characters who have reached a high enough rank qualify.

It is permanent removal dressed as achievement rather than punishment. Your best fighters leave precisely because getting them out is the point of the story and the campaign. The practical consequence is that no single dominant lineup can carry the whole game, and the player keeps being asked to bring the rest of the roster along.

Together those systems turn mistakes into speed bumps rather than turning points. A loss costs you something real, but it does not necessarily drop you into a spiral with no floor. A win counts for something real, but it does not let one unbeatable trio walk through everything that follows.

Because the strongest units keep leaving, Pyre pushes you to spend attention across the whole team. The final and hardest Rites may well be fought by the characters you used least, the ones who never climbed high enough to be liberated. The tension there comes out of the campaign's own structure, not out of a difficulty dial being turned up behind the scenes.

That is the elegant part. The negative loop is not a slap on the wrist for doing well. It converts your success into a new constraint that lines up with the fiction, the goal, and the emotional shape of the story. Nobody tells you that winning was a mistake. You are told that winning carries a cost worth feeling.

The distinction matters, because mixed signals are the standard failure mode for negative loops. Games spend a lot of effort teaching a style of play. If the same game then penalizes you for executing that style well, it appears to argue with its own lessons, and the player starts wondering whether losing on purpose is the smarter line.

Resident Evil 4 shows the signaling risk. Its dynamic difficulty is a negative feedback loop in practice: play well and the game can tighten the screws, struggle and it can back off. Advertise a system like that too loudly and you risk teaching strong players to play worse on purpose, or telling players who are having a hard time that their progress is being cushioned for them.

Pyre sidesteps that because victory is still plainly better than defeat. Winning pays more, and liberation is never framed as a fine. It is the stated aim of the game, wired into character arcs and the drama around them. Letting a powerful ally go can hurt exactly because it is so clearly the right call.

There are smaller negative loops inside the Rites too. Score, and the character who scored sits out the next play, which hands the other side a temporary numbers edge. It echoes real basketball, where scoring means giving the ball back. The point stands on the board; the moment right after it opens a door for the opponent.

A stabilizer like that works because it never cancels the achievement. The score is kept. The player who earned it earned it. All the loop does is stop that one advantage from cascading straight into total control, which keeps the contest breathing without making success feel hollow.

The broader lesson Pyre offers is that a negative loop can neutralize the most destabilizing properties of a positive one. Killing momentum is not the aim; momentum is usually the fun part. The aim is to stop momentum from hardening into a foregone conclusion while the game still has interesting decisions left to hand out.

Civilization runs the same tension. Conquering a neighbor gets you more land, and more land is a snowballing positive loop: a bigger empire yields more resources, more units, and more chances to expand again. The counterpressure comes from unhappy citizens, the rising cost of administering a sprawling empire, and other nations turning cold on you diplomatically.

That pressure is worth the most early on, because positive loops grow in force as they run. The opening turns, the first kills, the first rounds, the earliest decisions often shape the eventual winner far more than anything later. If an early advantage gets to compound unopposed, the back half of the game becomes less about playing and more about waiting for the paperwork.

Some negative loops are not written into the rules at all. In multiplayer Civilization, the table often decides on its own to gang up on whoever is out front, simply because the leader has become the threat. No correction needs to be hidden anywhere in the system. The social layer supplies the brake, because everyone else has a reason to stop the runaway empire.

Adding a negative loop is not the only response to a runaway positive one. Often the better fix is to dampen the loop so its effects stop compounding so hard. Call of Duty did this to itself. In the earlier games, kill streak rewards could count toward the next kill streak reward, which built a chain where one bonus helped earn the following bonus, and that one earned the next.

Black Ops broke the chain: kill streak bonuses no longer fed into further kill streak bonuses. Playing well still paid, but the payment had to be earned again through direct play each time. The loop survived; the explosive part of it did not.

Mario Kart 8 takes a different route with the super horn, which gives the leader a real answer to an incoming blue shell. The negative loop stays in the game. What changes is that the player out front now has a counter to its most infuriating edge case, which makes the whole stabilizing system feel less arbitrary to be on the wrong end of.

Occasionally the tidiest move is deleting the loop. Devil May Cry pays a high score at the end of a level in red orbs, the currency for better weapons and healing items. So the players performing best are handed the tools they least need, while the players struggling have the least money for the items that would actually help. The instinct behind the reward makes sense; the effect drags the difficulty curve the wrong way.

Donkey Kong handles the same impulse better with its hard-to-reach KONG letters. Gathering the full set does not pay out coins, extra lives, or a smoother run through the main game. It opens punishing bonus stages instead. For players who have already shown they are good, that is the right prize, because it gives them more challenge rather than more raw power.

Feedback loops sit near the center of a wide range of games: JRPGs, persistent strategy campaigns, online shooters, arcade racers, party games, and plenty more. They can manufacture growth, pressure, tension, drama, comeback potential, and clean endings. They can equally manufacture death spirals, unassailable leaders, punishments that feel unearned, and signals that teach players the opposite of what you intended.

The question to answer is not whether a loop is positive or negative. It is what the loop does to the player's next decision. Does winning open a satisfying road to victory, or does it make the remaining hours predictable? Does losing create pressure worth having, or does it make recovery impossible? Does the comeback tool hold the group's attention, or does it make skilled play look pointless?

Answer that before touching any numbers. You can trim an experience reward, shorten a respawn timer, soften a comeback item, or delete a bonus outright, and none of it helps until the loop's job is settled. Decide first whether this loop exists to hurry the ending, to protect the player who is behind, to preserve long-term consequence, or to stage a reversal. Then tune its strength against that job, get the build running, and watch whether players believe the result.

Start by naming the feeling the game needs. A competitive match may want a positive loop that closes things out before they drag. A party game may want a negative loop that keeps the weakest player in the story. A long campaign may want both at once: consequence heavy enough that choices count, and correction reliable enough that one terrible evening does not end the run.

Pyre lingers because its corrections read as part of the game rather than a patch stretched over it. It lets characters grow and then asks you to give them up. It makes defeat hurt and then limits the damage so the story can keep going. It is a reminder that feedback loops are not background arithmetic. They are one of the main ways a game establishes what winning, losing, and momentum are actually worth.

In the Flockbay app, describe three fights in a row where a win makes the next fight easier, then play them. If the third fight is no longer a decision, add a cap. If the first fight does not change anything, the loop is not actually there.

An AI roguelike maker and a survivors-like template are the fast way to see a snowball. One run will tell you.