Perfectly fair matchmaking drives players away
For years, the rule was simple: pair players of equal skill, and everything else takes care of itself. That rule is now cracking. A new study published in Management Science shows that perfectly balanced matches can quietly drive players away. The real problem isn’t a single loss — it’s the streak that follows.
The research team, led by scientists at Carnegie Mellon University and New York University[1][2], analyzed 5.4 million chess games from the online platform Lichess[3]. What they found turned the old playbook upside down. Skill-based matchmaking, or SBMM, was supposed to protect players from frustration. Instead, it created a hidden trap: when every opponent is equally strong, a losing streak becomes almost inevitable, and that streak pushes people out of the queue.
How the Old System Worked
Skill-based matchmaking emerged in the early 2000s as online gaming exploded. Before that, most games used simple random matching. You could face a beginner or a world champion on your next click. The experience was chaotic, and many players quit after getting crushed.
The solution seemed obvious. Use a rating system — similar to the Elo system invented by Arpad Elo in the 1960s for chess — to estimate each player’s ability. Then pair people with opponents close to their own rating. The match becomes fair. The game becomes competitive. Everyone should be happy.
That logic powered platforms like Xbox Live[4], PlayStation Network[5], and Steam[6] for nearly two decades. Games from Call of Duty to League of Legends adopted SBMM as the gold standard. Researchers at Microsoft and Riot Games[7] published papers showing that balanced matches increased player satisfaction in the short term.
But the new study reveals what those early papers missed: the long-term behavioral cost.
The Sequence Problem
The core insight is simple but powerful. A single loss is part of competition. Players accept that. They understand that in any fair contest, someone has to lose.
The problem is what happens next. When every opponent is equally skilled, the odds of losing two, three, or four matches in a row become surprisingly high. A player who wins 50% of their matches — the exact target of SBMM — will experience a losing streak of three or more games about 12% of the time. A streak of five or more happens roughly 3% of the time.
Those streaks change behavior. The study found that after a losing streak, players were significantly more likely to stop playing for the day, or even to stop playing altogether over the following week. The system designed to keep matches fair was actually creating the conditions for players to quit.
The researchers call this the “fairness paradox.” The fairest possible matchmaking, measured by skill balance, can produce the worst possible experience, measured by engagement.
What the Data Showed
The team analyzed 5.4 million games from Lichess[3], one of the largest free chess platforms in the world. Chess is ideal for this kind of study because it has a clear, objective rating system and a large, active player base.
The results were striking. When the researchers simulated an optimized matchmaking system that accounted for recent wins and losses — not just overall rating — engagement increased by 4% to 6% compared with standard SBMM. In theoretical scenarios where the system was tuned perfectly, the gains reached as high as 50%.
The optimized system worked by adjusting matchmaking based on recent outcomes. A player who had just lost three matches in a row would be paired with a slightly weaker opponent, even if that opponent had a lower rating. The goal wasn’t to give the player an easy win, but to break the streak before it became a quitting signal.
This approach mirrors findings from behavioral economics. Daniel Kahneman and Amos Tversky showed decades ago that people feel losses more intensely than equivalent gains — a principle called loss aversion. A losing streak amplifies that effect. Each loss hurts more than the previous one, and the pain accumulates faster than the pleasure of wins.
The Real-World Impact
For game developers, these numbers translate into real money. The Global Games Market Report projects that the gaming industry will generate nearly $188 billion annually by 2025. Even a 1% improvement in player retention can mean millions of dollars in revenue for major platforms.
But the implications go beyond pure economics. The study raises deeper questions about what “fair” really means in competitive systems.
Consider the experience of a casual player on a platform like Call of Duty: Warzone or Fortnite. They log in after work, hoping to relax. The SBMM system pairs them with opponents of similar skill. They lose three matches in a row — not because they played badly, but because every opponent was equally capable. The experience feels less like fun and more like work.
Now consider the same player with a slightly different system. After two losses, the matchmaker finds them a slightly weaker opponent. They win. The streak breaks. They play for another hour. Over time, that small adjustment keeps them coming back.
The study’s authors argue that this is not about manipulation. It is about recognizing that human psychology does not follow a simple rating curve. People are not machines. They have emotions, patterns, and breaking points.
Where This Gets Tricky
The research also opens a difficult door. If matchmaking systems can be tuned to maximize engagement, where does that stop? At what point does “optimization” become “manipulation”?
The study found that paid advantages — like buying stronger characters or weapons — actually improved engagement in some cases. Players who spent money on in-game advantages were more likely to keep playing, even when those advantages made matches less fair.
This creates a tension between competitive integrity and retention goals. A system that keeps players engaged might also encourage them to spend more money, especially if losing streaks can be broken by purchasing upgrades.
The researchers are careful to note that their findings do not endorse pay-to-win design. But they acknowledge that the line between fair matchmaking and behavioral manipulation is not always clear.
Parallels in Other Fields
The fairness paradox is not unique to gaming. Similar patterns appear in sports leagues, job markets, and education systems.
In professional sports, leagues use draft systems and salary caps to create competitive balance. The NBA and NFL both aim for parity, where every team has a roughly equal chance of winning. Yet research from Stanford University[8] shows that too much parity can reduce fan interest. Fans want close games, but they also want narratives — underdogs, dynasties, and redemption arcs.
In job markets, equal opportunity policies aim to level the playing field. But studies from Harvard Business School[9] suggest that too much leveling can reduce motivation. People work harder when they believe their effort can create meaningful advantage, not just when the starting line is fair.
In education, grading curves ensure that students are compared to their peers rather than absolute standards. But research from University of Chicago[10] shows that curved grading can discourage collaboration and increase anxiety, especially among students who fall behind early.
The common thread is clear: fairness is not a simple number. It is a psychological experience that depends on context, sequence, and expectation.
What Comes Next
The Management Science study is already sparking conversations among game developers and platform designers. Several major companies, including Electronic Arts and Ubisoft, have internal teams studying matchmaking algorithms. The Lichess[3] team has expressed interest in testing the optimized approach on their platform.
But the research also raises a deeper question that no one has fully answered: What do players actually want from competitive games?
Is it fair matches? Is it winning? Is it the feeling of improvement? Or is it something harder to measure — a sense of agency, of control over their own experience?
The study suggests that the answer is not simple. Players want fair matches, but they also want streaks to end. They want competition, but they also want relief. They want to feel that their skill matters, but they also want to feel that the system is on their side.
These are not contradictions. They are the normal, messy preferences of human beings. And any matchmaking system that ignores them will eventually drive players away — even if every match is perfectly balanced.
The Open Question
The University of Warwick team that published the Astrophysical Journal study on planetary rings faced a similar challenge. They had a beautiful theory — that dust rings around young stars could reveal the masses of hidden planets. But theory is cheap. The hard part is testing it against real data.
They chose PDS 70, a star system 370 light-years away. It had two known planets, PDS 70 b and PDS 70 c, and it had been imaged directly by the Atacama Large Millimeter/submillimeter Array (ALMA) in Chile. The team ran their simulations, compared them to the observations, and found that their technique worked. It estimated the mass of PDS 70 c at about 7.5 times the mass of Jupiter — matching existing measurements.
But the real discovery came from the dust. The simulations showed that forming planets could trap as much as 20 times the mass of Earth of dust within their rings. That is a huge amount of material, enough to potentially form new planets. Yet ALMA has not detected any new planets in those rings.
The question is not whether the rings exist. They do. The question is what they are hiding.
Ralph Pudritz of McMaster University put it plainly: “Our results suggest that the dust is sufficiently abundant and concentrated enough to potentially kick-off planet formation. This is an important insight that will initiate further observations and theory.”
That is the open question. Not what was solved, but what is now possible to ask. The rings are there. The dust is there. The planets should be there. So where are they?
The Bridge Between Two Worlds
The matchmaking study and the planetary study share a deeper pattern. Both start with a simple model — skill ratings for players, dust rings for planets — and both discover that the model is incomplete. The real system is more complex, more psychological, more dynamic than the original theory predicted.
For matchmaking, the missing piece was sequence. Players do not respond to individual matches. They respond to streaks. A single loss is fine. Three in a row is a problem.
For planetary formation, the missing piece is concentration. Dust rings are not just passive markers. They are active environments where material accumulates, potentially triggering new rounds of planet formation.
Both findings point to the same conclusion: the visible structure is only part of the story. What matters is what happens between the rings, between the matches, between the moments we can see.
Amena Faruqi of the University of Warwick called the dust rings “planetary fingerprints.” The matchmaking researchers might call losing streaks “player fingerprints.” Both are signals that we are only beginning to learn how to read.
The Future of Fairness
The Management Science study is not a final answer. It is a starting point. The researchers are already planning follow-up studies on other platforms, including first-person shooters and real-time strategy games. They want to know whether the fairness paradox holds across different genres and different player populations.
The Astrophysical Journal study is also ongoing. The team is applying their technique to other protoplanetary disks observed by ALMA and the James Webb Space Telescope. They want to map the hidden planets of the galaxy, one dust ring at a time.
Both teams are asking the same question, in different languages: What are we missing?
For matchmaking, the answer might be emotion. For planetary formation, the answer might be time. But both answers lead to the same place — a deeper understanding of systems that are more alive, more dynamic, and more surprising than we ever imagined.
The rings are beautiful. The matches are fair. But the real story is what happens between them.
Sources
3. Lichess
4. Xbox Live
6. Steam
7. Riot Games