A Rulebook Without a Referee: Dissecting the 90-Pull Pity System and Its Governance Lessons for Esports

The 89th Minute of a Schedule Sheet

A compiled briefing on the banner schedule for versions 7.0 and 7.1 of an open-world role-playing game has just been published. It contains 28 information points. Twenty carry no source. One cites an official publisher announcement. Three are the writer's own opinions.

I read it three times, the way I rewatch a controversial passage of play. First at real speed. Then at 0.5x. Then looking only at what the camera did not cover. On that third pass, what I saw was not a schedule. What I saw was a rulebook written, enacted and adjudicated by the party that profits from it.

The sheet lists names: Odette, Flins, Ineffa, Vesna, Vodyanitsa. Alongside version numbers 7.0 and 7.1, and phrases such as Snezhnaya, Skirk, Escoffier, Aino, Iansan, Lan Yan. I checked every official archive I could access. Nothing matched. No official document confirms that most of these names existed at the time the briefing was published.

In my trade, when a match report has 20 of 28 entries uncredited, it does not enter the file. Every passage of play is a line in the record, and I write all of them. But I only write the lines the footage confirms.

This is why I chose to write about this briefing, even though it belongs to no pitch, no arena, and contains no athlete. Because its structure — one party setting the rules, operating the system, publishing the outcomes, and collecting the revenue — is a structure esports has collided with repeatedly over seven years.

Context: the 21-Day Rhythm and a Reason to Spend

The title in question does not operate on esports logic. It has no official professional circuit. No equivalent of Worlds, The International, a Major or a VCT. No franchised club system. No transfer market in the sporting sense — no contracts, no release fees, no buyout clauses. Most importantly, its updates are not competitive balance patches. Each version is a PvE content drop.

That means the entire analytical vocabulary I normally use — rosters, form, win rates, pick-ban, paper strength — has no standing here. I say this up front so no reader proceeds with the wrong expectation.

What exists here is a release rhythm. Each version splits into two phases. Each phase runs roughly 21 days, with variation. Each phase has its own banner window — its own spending window.

In esports, a schedule exists to allocate competition. Here, a schedule exists to allocate spending.

The briefing describes phase two of version 7.0 as carrying reruns of older characters. Phase one of version 7.1 is said to debut two new characters simultaneously. Phase two of 7.1 returns to reruns. The writer calls this a good time to prepare for 7.1.

To be fair, the briefing itself admits the exact schedule is unconfirmed. That is a positive integrity signal. It is also a confession that everything preceding it is provisional.

The Core: Dissecting a Self-Contained Revenue Engine

Set aside the unverifiable names. What remains is a fairly complete description of the pricing architecture of a gacha model — one of the most sophisticated revenue machines the games industry has produced.

The Soft Floor at 90 Pulls

A five-star character is guaranteed within 90 pulls. Call it pity.

The subtlety is in the word soft. Theoretically 90 is the ceiling. In practice, drop rates usually climb as the player approaches the threshold. The briefing supplies no probability distribution, no rate table, no note on whether rates shift with pull count.

The point of a soft floor is not to cap player losses. It is to shape player expectations. When players believe a firm stopping point always exists, they are willing to walk a longer road. The clearer the floor, the higher the average pull count.

50/50: Variance Is the Product, Not a Bug

On the event banner, the first five-star has a 50% chance of being the limited character and a 50% chance of a standard character. If it lands on standard, the next five-star is guaranteed limited.

When one branch of a 50/50 is offset by a guarantee in the following cycle, expected cost is not the cost of a single attempt. It is the cost of a chain. Expected cost sits between the two extremes — not 90 pulls, not 180, somewhere in between, depending on a distribution the briefing does not provide.

More notable than the mean is the variance. A 50/50 plus guarantee system produces a cost distribution with a long adverse tail. A small group of players pays close to double the average while another group pays very little. The publisher does not need to optimise the mean. It needs to optimise total revenue, and total revenue grows when the distribution spreads.

This is the basic principle behind lottery and insurance pricing. Sellers do not profit from expected value. They profit when buyers accept variance.

Shared Pity: Lowering Friction, Raising Frequency

Pity counters are shared across banners of the same category. A player with 47 pulls banked on one character banner keeps those 47 when switching to another. The counter travels with them.

This lowers the marginal cost of switching. Imagine a reset-on-switch model: players weigh every move carefully, because switching destroys progress. Now imagine the shared model: players switch more easily, sample more banners, and encounter more spending windows within the same period.

The briefing describes the shared model. This is revenue smoothing, not a goodwill concession.

When switching costs fall, spending frequency rises. A player stays inside the ecosystem longer and passes through a greater number of purchasing windows.

No Fixed Rerun Schedule: Manufactured Scarcity

Some characters are absent for over a year. Others return within a few versions. Players cannot predict.

That is manufactured scarcity, and in game design it is a FOMO engine. The key detail is not the mechanic itself but its unpredictability. A fixed schedule would let players plan financially for the year. Unpredictability makes long-term planning impossible — and when long-term planning fails, optimal player behaviour shifts to short-term reaction to announcements rather than response to budget.

Chronicled Wish: A Secondary Harvesting Lane

A separate banner type exists, with its own rule set, typically for older characters: Chronicled Wish.

If an old character can no longer appear on the primary banner, the publisher loses the ability to monetise them. But putting them back on the primary banner sacrifices a slot that a new character would fill, and that slot is worth more. The secondary lane solves the problem — reharvesting legacy assets without displacing new debuts.

This is tiered revenue diversification: the upper tier for the highest-value new content, the lower tier reharvesting legacy assets at near-zero marginal cost.

The Pressure Peak Sits in Phase One

Put the pieces together and the rhythm is clear. Phase one carries two simultaneous new characters; phase two is reruns. The spending-pressure peak therefore lands on phase one, then cools. This is peak-then-trough design. The trough is not kindness; it is an engineering requirement of any recurring revenue machine. Without a recharge phase, payers deplete and leave.

The Transmission Map Has No Middle Layer

An esports event has a multi-tier value chain: publisher upstream, organiser in the middle, clubs and players downstream, a broadcast layer alongside. Every tier can negotiate.

Here the chain has two nodes: the publisher and the player's wallet. No clubs to pay salaries. No organisers to share revenue with. No broadcast rights to negotiate. No match days to sell tickets for. Value flows directly from upstream to downstream, with no intermediary.

As a business structure, it is near-ideal: high margin, low distribution cost, a repeating 21-day cycle. As a governance structure, it concentrates power beyond any esports ecosystem I have tracked.

The Rule-Writer Also Adjudicates

Information point 12 is the only one citing an official source. The publisher is simultaneously the drafter of the gacha rulebook, the operator enforcing it, the authority announcing it, and the sole beneficiary of it.

In any refereeing system, those four roles must sit with four separate bodies. That is the foundation of legitimacy. SAOT is a steel eye, but a human hand still operates it. Technology does not create legitimacy. Institutions do.

I am not alleging fraud. There is no evidence of it, and I will not make accusations the footage does not support. I am describing structure — the same structure that has surfaced repeatedly in esports, each time accompanied by a transparency controversy nobody could resolve with hard proof.

The Adjacent Legal Grey Zone

Under most current frameworks, gacha is not classified as gambling: rewards carry no cash redemption value and there is no reverse transfer mechanism. But it sits adjacent to the loot-box debate, where two issues dominate — probability disclosure and protection of minors.

The briefing mentions the 90-pull threshold, the 50/50 mechanic and the next-pull guarantee. Those descriptions take the shape of a rules disclosure. Yet no regulator is cited, no statute referenced, no effective date recorded. A rulebook published without an independent issuing authority is still a rulebook. It is simply not one that can be cross-checked.

Source Audit: the Work Nobody Wants to Do

On the 28 points: 20 unsourced, one citing an official announcement, three writer opinion. Official-source ratio: roughly 3.6%.

Having once spent 47 notebook pages classifying 1,208 refereeing decisions across a single World Cup, I can say that ratio is not enough for a briefing to serve as a basis for decision-making. Forty-seven pages taught me one thing: stay silent until the evidence appears.

The names Odette, Flins, Ineffa, Vesna, Vodyanitsa and the version numbers 7.0 and 7.1 could not be cross-verified at the time of receipt. That does not make the briefing false. It makes it unverified. The distance between those two states is the distance between a rumour and a fact.

The Counterintuitive Angle: Emotion Is Valid, but It Is Not Evidence

When a new character is announced, the community splits. One side says now is the time to save. The other says the publisher is squeezing players by stacking two new characters into one phase.

Both sides are right emotionally. Both are wrong methodologically. The supporters are right that phase one matters, but they conclude it from a schedule, not from power data. The briefing contains no kit, stat or team-fit information. It tells you when, not whether. The opponents are right that stacking two debuts creates spending pressure, but they call a legitimate revenue-architecture decision an act of coercion.

Emotion can lean. Footage cannot.

The real debate is not which character is worth pulling. It is that the information both sides argue over is unsourced. Readers argue fiercely about a schedule that is itself unconfirmed.

A second, more uncomfortable angle: having 20 of 28 points unsourced may not be the writer's fault. It may be structural to content produced to fill the gap between two official announcements. Game media has a recurring information vacuum — the interval when players still want to know what comes next. The vacuum generates demand. Demand generates supply. And supply, lacking verified data, gets filled with speculation.

The mechanism mirrors the football transfer-rumour market. Nobody can verify a rumour when it is published. It gets published anyway, because people read it. The difference: football has a player-registration system and a transfer window with fixed dates. Bad information gets publicly refuted and traced. Here, the only verification system is the publisher's own announcement — issued whenever the publisher chooses.

Fans remember player names. I remember where the assistant referee stood. Here, the position that matters is not the writer's. It is the position of the only party with the authority to publish the truth.

The Stopping Point: Three Things to Do Before Spending

First, separate two kinds of question. Schedule questions are about timing. Value questions are about power. Content like this briefing answers only the first. Anyone answering the second with a schedule is filling a gap with speculation.

Second, build a verification habit. Three steps: check the official channel, record the publication date, wait for second confirmation. Ten minutes. Far cheaper than emptying a currency reserve on an unsourced schedule.

Third, an industry point for esports people. The power structure visible in the gacha model — one entity writing rules, operating them, announcing them and profiting from them — is the structure esports has renegotiated repeatedly over the past decade. Every successful renegotiation added an intermediary: a players' association, an independent body, a disciplinary committee with outside members.

Referee data exists not to convict but to exonerate. And exoneration requires data from a source that can be challenged.

A system with one source of truth — which also collects the money — will never correct itself. It corrects only when someone is patient enough to recount every pull. I once counted 1,208 decisions in a single World Cup. Counting 28 data points takes fifteen minutes.

Start anywhere. Just start.