Nine Percent Against a Hundred Percent: KRAFTON's Five-Year Trust Ledger

On 23 September, when KRAFTON published the outcome of its investigation, the least discussed sentence in the document was a technical admission: two players had used outside information — an opponent's livestream — to shape their rotations and decisions during the competition. The names were Anyone's Legend's Himass and The Expendables' TanVuu. The outcome: permanent account closure and a ban from every official PUBG Esports event organised or approved by KRAFTON, including PGC, PGS and PNC. What followed was no longer a story about a punishment. Within days, more than 4.1 million signatures had been collected in support of the two Vietnamese players.

Covering track and field gave me a habit: put the split table in the first paragraph, analyse afterwards. Without knowing who was where at each ten-metre mark, the story of the final forty metres turns false. This case needs three rows in its split table — a promised 100 percent, an actual 9 percent, and a punishment whose duration is written nowhere. Read separately, the 2026 story will be misread.

Because KRAFTON is not merely a game publisher here. The same entity is publisher, tournament organiser, rule-maker and sanctioning body. In track, those roles are separated — the IAAF writes rules, local federations stage meets, WADA polices doping, and CAS hears appeals. In the PUBG ecosystem, four chairs sit at one table. That makes the demand from GAM x The Expendables — explain this clearly — procedurally hard to satisfy, because the explanation must come from the body whose decision is in question. Anyone's Legend went further and raised a more fundamental point: the players' right to be heard.

During my Tokyo Olympics coverage I learned that a final result is the sum of four separate decisions: preparation, taper, race-day course, and the courage to take risk in the last two hundred metres. No single decision settles the outcome. In esports those layers multiply: patch version, scrim workload, server, broadcast delay, and tournament-operations calls. A fault in any layer changes the result, and not visibly.

Selling a probability is a consumer contract, not a favour. That sentence sits at the centre of KRAFTON's five-year ledger.

The first page of the ledger: nine percent

In 2026, as part of the NewJeans collaboration, KRAFTON placed two paid items in the in-game store — the PUBG X NewJeans Loot Pack and the PUBG X NewJeans Premium Bundle. Players could open them for collectible items and NewJeans-inspired customisation. The problem appeared in the disclosed odds. KRAFTON stated that a player who failed four consecutive times would be guaranteed the set blueprint on the fifth attempt at 100 percent. Some players said they bought more than five Premium Bundles and still received nothing.

This is where my 2026 data notebook returns. At the London World Championships 100m final, Bolt finished third in 9.95, behind Gatlin (9.92) and Coleman (9.94). Instead of posting a fan reaction, I built a spreadsheet of reaction times — Bolt 0.183, Gatlin 0.138, Coleman 0.123. The first ten metres decided the medals, not the last forty. The gap in this loot box is the same kind of arithmetic at a much larger scale: promised 100 percent, actual 9 percent, a 91-point gap.

The community's complaints were not taken seriously at first. Players submitted screenshots and records of multiple openings, yet the in-game text was being treated as the company's authoritative statement. In 2026, the argument about 37 kilometres per hour was dismissed in a campus room despite the data. Here the reverse happened — the user's screenshot was disbelieved, the system's displayed text believed.

The PUBG team later admitted that the Premium Bundle did not in fact fall under the bad-luck-prevention mechanism, and that some in-game information was displayed incorrectly. KRAFTON corrected the information, apologised and announced a compensation package. Even so, some of those who had spent heavily demanded direct cash refunds.

In June 2026, Yonhap reported that South Korea's Fair Trade Commission (FTC) had fined KRAFTON 2.5 million won for the false odds disclosure. According to the FTC, the company had told players they were guaranteed an item after four failed attempts, while the actual rate was determined to be 9 percent. It was also found to have provided inaccurate information about the odds of other items. KRAFTON subsequently refunded about 1.1 billion won to roughly 380,000 buyers and provided in-game compensation worth about 9.8 billion won. Because the company corrected the information and paid compensation on its own initiative, the penalty was limited to a fine.

The ledger should be added up here. Refunds and compensation together come to roughly 10.9 billion won, against a fine of 2.5 million won — about 4,360 times. Anyone reading only the fine column will conclude the company was punished harshly. The opposite reading is more useful: self-correction turns a fine into a symbol, and that gap is what distorts the market for companies that do not self-correct.

The second page: the faces on the skins

In 2026 another controversy emerged from the same NewJeans collaboration. South Korean media reported that some players had combined members' face skins with revealing outfits to create and share sexually harassing images and videos. The case drew more attention because Haerin and Hyein were minors at the time. KRAFTON and the management agency ADOR said they would act against uses inconsistent with the collaboration's original purpose. KRAFTON later restricted combining NewJeans face skins with certain outfits.

That decision also drew questions. Part of the community argued the problem lay not only in user behaviour but in PUBG's customisation architecture — a system that freely pairs faces with clothing invites misuse. In an item economy, the skin is the product. Changing a product's terms after sale is also a consumer question, exactly as false odds disclosure is a consumer question. Both cases share a structure: the seller writes the rules, and after the sale the seller rewrites them. The user holds only the decision to buy, not the ability to read the terms.

The third page: the Peacekeeper Elite service fee

In 2026, ZDNet Korea reported that in its IPO filing KRAFTON had for the first time publicly acknowledged providing technical services to Tencent-operated Peacekeeper Elite and receiving service fees for them. KRAFTON's long-standing position was that PUBG Mobile and Peacekeeper Elite were separate products.

The background is May 2026, when PUBG Mobile shut down in China over licensing problems and players were moved to Peacekeeper Elite, a version similar to PUBG Mobile but altered in gameplay, visuals and content. The question is simple: was the modification enough to satisfy licence requirements, or was it a route to keep PUBG running in China? The service-fee line in the filing is a paper trail for that question.

There is a track analogy. An athlete can race in one country's jersey and train under another country's programme. No rule is broken, yet placing both accounts on one table raises the question of whose performance is credited to which federation.

The fourth page: stream sniping and the size of the sanction

The PUBG Asia Stars 2026 incident began with an accusation from Gen.G's Korean streamer Soopi, who said two Vietnamese players were watching opponents' livestreams. The organiser then removed Himass and TanVuu from the remainder of the event, adjusted scores, increased livestream delay, cancelled the third day of competition, and split the prize money equally.

In a battle royale, information is the only currency. Position, rotation, loot and timing — knowing any of these in advance changes decision quality. Watching an opponent's livestream in time means holding a map of their rotations. Stream delay is therefore a safety valve. But raising the valve breaks the broadcast product: viewers want real-time action; with more delay they watch the past.

In 2026 I built a dataset on empty stadiums — in the Bundesliga's first eighteen matches after restart, home wins fell sharply, and in Monaco's empty stadium Joshua Cheptegei ran 5,000m in 12:35.36 with pace lights and no crowd. An empty stadium and an increased stream delay do the same thing: they push the presence of the live audience out of the decision. Cancelling a showmatch day cancels whatever role the audience had.

After the investigation, KRAFTON said no further violations were found among other participants. The sanction was a permanent account closure and removal from the global PUBG Esports pathway. What was not published is what matters most: which logs, which timestamps, within which time window a match with outside information was established, and at what threshold that match became proof.

In anti-doping, WADA publishes thresholds and laboratory protocols so that sanctions can be audited — anyone can know which quantity, which method, which time limit. The method is in everyone's hands before a suspicion is proven. Here we received a conclusion, not a method. The stopwatch is a witness, not a verdict — but if the witness's times are not public, no one can audit the verdict.

What 4.1 million signatures actually measure

An uncomfortable point must be written here, against my own instinct. The 4.1 million signatures are not proof that the two players are innocent. A petition examines no documents; it counts people. Whether a decision is correct is not determined by the size of a petition.

Yet the petition measures something else: accumulated deficit. Had the odds been disclosed accurately in 2026, the 2026 ban might have been equally severe while drawing a far smaller reaction. The community might then have asked whether a mistake was made; now it asks whether the institution is trustworthy. Those are two different debates around one case, and which debate occurred was settled two years earlier.

I owe myself a caution too. Drawing a clean causal chain from the 2026 odds error to the 2026 ban is tempting — same company, low visibility of the sanction. But the opposite possibility deserves a place: perhaps the sanction is procedurally correct, and the community's anger is tribal loyalty. Both can be true. The sample here is one case, not a series, and one case cannot establish an institution's general tendency.

What can be said beyond the sample is a structural question. Where the same body writes the rules, runs the competition, investigates complaints and imposes sanctions, where is the appeal? When GAM x The Expendables demanded a clear explanation, they were not asking for a different judge — they were asking for the public basis of the same decision. Major figures in the Vietnamese community — Độ Mixi, PewPew, Rambo, DjChip, Ngân Sát Thủ — reacted publicly or stepped away from PUBG. That is not personal protest; it is a customer-relationship notice.

What to watch

First, a public standard for detecting stream sniping. Which delay, which window, and what information matching turns suspicion into proof — written down, so future complaints are judged by one measure and players know where the line is.

Second, the Korean FTC decision is not only a Korean matter. How far probability disclosure is legally mandatory is now a precedent for regulators in South and Southeast Asia. Bangladesh, Vietnam and Indonesia have no equivalent structure, so buyers rely only on the publisher's written word — and when that word is wrong, the space to question it opens late, under community pressure.

Third, the question is professional, not political. An esports career in Asia runs four to six years, so a permanent ban is the end of a profession. Track has an appeal path of this magnitude, between national federations and an international court. Esports still relies on the publisher's goodwill alone.

At a track meet my favourite moment is not the finish line but the four seconds of silence before feet settle into the starting blocks. Nothing moves in those four seconds, yet the rules and the trust of the entire race are stitched into them. Esports has almost reached the finish line; the four seconds at the starting block remain unwritten.