Anti-Boost: How Riot Games Pursued 296,416 Boosting Accounts

Three in the morning at a PC bang near Konkuk University in Seoul, and only four screens are still lit. I sit in the last row, one ear out of my headset, close enough to hear someone read a string of characters to the person beside them. A row of letters and numbers, a nod, and the chair changes hands. Twenty minutes later, the monitor to my left has moved into a different ranked match: same player name, same profile frame, but a different pair of hands on the keys. Those hands are faster, more confident, and the win streak begins.

I used to think that was a small story about a few net cafes near a university. Then Riot Games published its official communication on the Anti-Boost system, carrying a number that made me sit back down: 296,416 accounts actioned for rank manipulation across VALORANT and League of Legends. Three in the morning in Konkuk suddenly came with a figure attached, and the chair that changed hands became a row in a dataset.

The law beneath the ladder

Anti-Boost is an automated system operated by Riot Games to detect and penalize rank manipulation. Four categories of behaviour are targeted. First, boosting, where a higher-skill player logs into someone else's account and climbs on their behalf. Second, buying, selling or transferring accounts. Third, intentional deranking to enable boosting or to find easier matches. Fourth, climbing on an alt account with a clear skill mismatch.

The most notable element is the safe harbour Riot builds. Alt accounts that a player creates and operates themselves are treated as normal activity. The system targets the intent to manipulate rank behind the account, not the existence of the account itself. That is a narrow standard, deliberately so, and it is also the hardest part of the whole story.

The penalty structure has four tiers. On detection, ranked points and rewards earned through cheating are cancelled, the account is returned to its pre-manipulation rank, and a temporary suspension applies. Repeat offences lengthen the ban. Buying or selling accounts, or intentional deranking, can lead to a permanent ban. And the broadest clause of all: the booster's main account, along with players who frequently queue with them, may also be actioned.

Looking ahead, Riot says it will expand the system and add detection of match-level signs of boosting, rather than stopping at account-level signals. The stated goal is a fairer ranked environment. This entire announcement sits at the layer of ladder governance. It touches no team, no professional player, no tournament and no balance patch.

One account, two pairs of hands

In traditional sport, cheating usually attaches to a specific person: a player paid to miss, a referee paid in envelopes, a coach fixing a scoreline. Here, the unit being penalized is not a person but an account. An account is a name, a match history, a rank visible to anyone who types that name into a search field. When another pair of hands sits down at that chair, the history still belongs to the old name. The name keeps getting credit for shots it never took.

The name I mispronounced years ago now rings out like a song, and every time it does I remember that behind a string of characters is a person with their own way of pressing keys. Anti-Boost's problem sits in exactly that place, only in reverse: how do you know that a different person has sat down behind the string of characters.

Nothing on the screen announces that today's player has changed. The same account, sometimes the same device, a broadly similar skill band. The difference lives on a layer that is very hard to observe: decision tempo, mouse movement habits, how a one-on-one is handled, when a fight is taken. That is why the system has to work from behavioural signals rather than physical proof. And when a rule must rest on signals, the boundary between a correct accusation and a wrong one becomes the central question of the entire mechanism.

Intent as evidence

Riot frames the problem in a very particular way: self-created, self-operated alts are normal. What gets penalized is the intent to manipulate rank. In principle, this protects legitimate players. Many people keep a second account to try champions, to play with lower-skilled friends, or simply to start over after an exhausting season. A blanket ban on every alt account would wipe out a large share of ordinary community behaviour.

But that narrowness creates a second problem. Intent is not something you can see; it has to be inferred from behaviour. Inference means a judgement model sits behind the rule, and every model has a grey zone. Riot's own communication concedes that not every case of alt-account use is actioned. That sentence sounds gentle, but it describes a boundary drawn by judgement rather than by a hard threshold.

My point is not to criticize the method. It is to point at the structure of authority behind it. In football, integrity machinery is split across several hands: federations legislate, investigators gather evidence, disciplinary panels rule, and a sports court sits above as an appeal layer. Here, the publisher writes the law, runs the detection system, and issues the final ruling, with no independent body described anywhere in the account. That arrangement has clear advantages: speed, data consistency, no dependency on a slow third party. Its cost is the absence of an external check.

Listening before commenting is how I corrected my own mistake, and I think the same principle applies to a system of rules. A system earns trust when it has a channel for the accused to speak, not only a channel for it to announce.

Four tiers and a moral hierarchy

The four-tier penalty structure is more than a sentencing table. It is a statement about the moral hierarchy the publisher wants to install in the community.

The lowest tier covers behaviour treated as momentary: cheating-derived points and rewards cancelled, rank restored, a temporary suspension. The second tier covers repeat offenders: bans lengthen. The existence of an escalation mechanism says something in itself: recidivism is anticipated, and a single strike is acknowledged as insufficient.

The top tier covers two commercially driven behaviours: buying, selling or transferring accounts, and intentional deranking. This is where Riot says the most about its own values. Selling an account means trading an identity, and that identity is owned by the publisher. Intentional deranking means deliberately damaging other people's experience to serve another transaction. Both are punished far more heavily than a weak player asking a friend to play a few games.

Between those poles sits boosting, the emotional centre of the story. It is not automatically grouped with permanent bans, yet it is the behaviour that brought the whole system into existence. A player paying to climb is not breaking the rules the way an account seller does, but they are buying something the ladder promises cannot be bought.

Joint liability and the innocent bystander

The clause that made me stop longest is the extension clause: the booster's main account and players who frequently queue with them may also be actioned.

In a game where duo queue is routine, the phrase frequently queue together is a very low bar. A friend who queues twice a week with someone else has no way of knowing that the other person is being paid to log into a third person's account. Joint liability is an efficient tool for closing the perimeter around organized boosting networks. It also creates a zone in which a correct accusation and a wrong one look almost identical.

The communication mentions no appeals mechanism, no threshold for shared matches, and no way to separate those who knew from those who did not. Those gaps matter far more than what was written. In any disciplinary system, the hardest part is always the part about the person swept up without fault.

The blind spot of aggregated numbers

296,416 accounts, pooled across VALORANT and League of Legends. One number, two games that share almost nothing. One is a first-person shooter where skill is measured in milliseconds of reaction and precision of aim. The other is a MOBA where rank is climbed through macro decisions spread over thirty minutes per match. Boosting operates differently in each environment, is bought and sold differently, and faces different pressures.

Pooling the two titles into one line has a clear communications benefit: a bigger number. It also erases the ability to read each title's boosting economy. What the aggregate hides is the economic mechanism behind it, and that is precisely the part you need in order to judge whether a campaign actually reaches the root.

One further layer: there is no regional breakdown. Demand for boosting is unevenly distributed. It is higher where account markets are developed, where rank carries social prestige, and where enough players pay to sustain a network of boosters. Without a regional split, the reader cannot tell which market is being squeezed and which remains wide open.

And the final layer, which I want to state clearly: a total is not a trend. The announcement carries no prior-period figure, so the familiar line that Riot is tightening further is the reader's inference, not a conclusion drawn from data. That inference is reasonable, since Riot itself says it will expand the system. But reasonable is not the same as demonstrated. The data is self-reported by the publisher and not independently audited, which is worth remembering when reading the number.

Points are returned, memory is not

Anti-Boost is designed as reactive with rollback. Cheating-derived points and rewards are cancelled after detection, and the account is returned to its original rank. Between the manipulation and the detection there is always a lag. Inside that lag are real matches.

People queued that night, met an account being played by someone else, and lost. Their points are not part of the penalty package. Their ladder position may be technically corrected, but the feeling of that evening is not. The rollback mechanism works on the cheater's ledger, not on the opponent's memory.

The empty chair that day said more than any crowd, and in this story the empty chair is the matches that were taken away and never returned.

This is where I think Riot could do more on the communications side. A figure of 296,416 actioned accounts is a message aimed at violators. It is not yet a message aimed at those who were harmed. Those two messages need two different kinds of storytelling, and the second is almost entirely absent from what has been published.

The race between hunter and evader

Two details in the announcement say more than any number. The first is the escalating penalty mechanism, which exists because recidivism is expected. The second is the plan to add match-level detection of boosting signs, an admission that the current detection level is not enough.

That is the nature of every integrity fight in esports: a loop between detection and evasion. When the system tightens at account level, activity moves to the organizational level. When the system can read match-level signals, activity moves to external communication channels, to coordinated deranking rings, to splitting behaviour across devices. Each tightening raises the cost by one notch. No notch removes the incentive, because the incentive is rank, and rank is a public currency.

Why this story reaches beyond one game

At the publisher layer, Anti-Boost is an investment in maintaining trust. The ranked ladder is the daily engine of the whole ecosystem. It sustains daily active players, and daily active players are the foundation of everything above: viewership, sponsor attention, and the amateur talent pipeline.

At the grey-market layer, heavy penalties on account buying and selling strike directly at the supply side. That is a sensible place to start, because supply is visible through transactions. Demand is harder, because a buyer simply sits still and pays.

At the scouting layer, a cleaner ladder has higher signal value. Academies and teams still use solo queue rank as one of the first indicators when searching for talent. If rank is diluted by accounts played by someone else, screening costs rise for everyone downstream.

Based on my experience following matches and governance communications, this is the structural difference between esports and traditional sport. In football, integrity is a machine of separate organizations with divided authority and an independent appeal tier. Here, the publisher writes the law, enforces it and judges it inside one body. That concentration explains why integrity campaigns in esports can appear so quickly, and it also explains why they always need close scrutiny for transparency.

The counterintuitive angle: what is being squeezed is not the root

The common community reaction to this announcement is predictable: Riot is getting tougher. That is a comfortable reading, and I think it misses the hardest part.

The hardest part is demand. A player who pays to climb does not do it because they enjoy cheating. They do it because rank is the only thing in the game that publicly tells the world who they are, and because the gap between what rank promises and what it delivers keeps widening. When a player lacks the time to climb to the tier they want, a market appears to sell them that exact tier. Squeezing supply makes the service more expensive, not extinct.

The heaviest penalties in the ladder fall on commercial behaviour: account trading and intentional deranking. Those are the behaviours Riot can demonstrate through transaction signals. Boosting, the behaviour that brought the whole system into being, sits in the harder-to-prove zone. The structure accidentally creates a paradox: the heaviest penalty does not land on the most common behaviour.

And the most overlooked element in community discussion is joint liability. If that clause is applied broadly, the first person to suffer is not the booster but the unaware duo partner. A legal system lasts only if it can withstand the pressure of wrongful cases. This announcement does not yet show that the pressure-bearing layer exists.

I also want to say something about how the community reads the number. Every player name is a short poem, if we take the time to read it carefully, and every actioned account is the same. Behind 296,416 rows of data are 296,416 stories, some of them correct and possibly some of them wrong. Reading carefully means accepting both possibilities at once, instead of choosing only the one that feels reassuring.

What to watch

Four signals are worth tracking. The first is Riot's next data disclosure, because it would enable comparison over time, which this announcement cannot. The second is the first publicly documented wrongful punishment, because it will test the durability of an intent-based standard. The third is any document clarifying the joint-liability clause, especially a shared-match threshold and an appeals route. The fourth is how other publishers report their own integrity data, because only with comparison points can a reader judge whether this figure is large or small in the wider context of the industry.

Three in the morning in Konkuk, and the chair changes hands again. Next time a name logs into a ranked match, another layer of rules will stand behind it. The question I take home is not whether that layer exists. The question is whether it can tell apart the person who deliberately changed hands and the person who merely sat nearby.