369 Cups, 15 Kilograms and One Blank Page: An Audit of the NPB Beer-Vending Model

There is a girl at Belluna Dome who sells 369 cups of draft beer in a single game. Her first game was 100 cups. Over three years that is a jump of roughly 3.7 times — a growth curve belonging not to a batter but to a server working the stands.

My eye did not stay on the 369. It stayed on the 15 — kilograms. The pack on her shoulder that Mana Kato carries up and down concrete steps for three to five hours, in humidity, on short water breaks. What the stand sees is cold beer and a smile. What it does not see is the arithmetic of a shoulder.

I keep a separate ledger for stadium labour, one that begins in the Eden Gardens concourses of the 1990s and runs to the ISL-era grounds where a water seller climbs forty flights in ninety minutes. That ledger does not record how happy the crowd was. It records what weight was on whose shoulder. The spreadsheet remembers what the stands forget — and here the spreadsheet caught precisely what the stand cannot see: the pleasure is running on somebody else's back.

The system that produces this story is not the MLB system. American ballparks sell cooled cans and bottles at aisle counters — a transaction, and it ends there. Japanese ballparks send the vendor to the seat, pouring draft on the spot. Three pillars hold the model up: freshness (draft instead of can), immediacy (no queue), and relationship (regulars who know the vendor by name). This is stadium-operations design, not sporting strategy — and it has to be read that way, because the entire economics downstream follows from it.

The workforce is mostly university students. That means seasonal supply and a hard tenure ceiling: the day the degree arrives, the experience walks out of the stadium. Mana Kato, 21, applied in 2026 after watching TikTok videos, knowing almost nothing about baseball. In October 2026 she quit, to prepare for graduation.

Here I have to log a date anomaly, because that is the job: an application in 2026, a three-year tenure, and an exit in October 2026 do not reconcile cleanly — that arithmetic runs closer to four years. The original interviews were published in The Japan Times; the dates should be verified before any of this is cited as record. I do not run a scene without a record, and here the record has a corner that is not yet clean.

On the demand side there is another ledger nobody counts in the stands. Yuichi Ito, 56, a long-time Seibu Lions fan, drinks seven to eight beers a game. His whole interview is a document of habitual consumption with no service cut-off policy quoted anywhere. Weight on one side, consumption on the other, and a single vendor in between.

The income structure is piece-rate: a vendor's earnings depend partly on the number of cups sold. What is not stated matters more than what is. The per-cup commission, whether there is a base wage, whether the classification is employee or contractor, whether the employer is the club or a concessionaire — none of it appears. No income model can be built from the source. What can be built is a measure of productivity dispersion, and it is startling.

One hundred cups at the debut, 369 three years later. That is a spread of about 3.7 times — the gap between a novice and a veteran, not a marginal improvement. In a piece-rate system, dispersion that wide usually produces two effects: new vendors exit quickly, and a small, high-earning veteran core stays. Kato lasting three years, and recruiting her younger sister, fits that pattern exactly. In other words, the real screening is not the interview; it is the body and the ability to talk — and that filter is run by fatigue, not by policy.

The biggest asset the job creates is not recorded anywhere on the books. The engine of Kato's sales is a base of regulars, held together by her self-introduction and name recall. That technique appears nowhere as trained or standardised practice — it is a habit passed down from older vendors. When she graduates and leaves, the customer map leaves with her, not with the club. The club never registered a revenue-generating relationship in its own name.

The physical side also deserves arithmetic the article never completes. A pack of roughly 15 kilograms, three to five hours of stair work, humidity, short breaks to drink water. Rough reckoning: if that pack carries on the order of 10 to 15 litres of net beer and a pour is 350 to 400 millilitres, that is about 25 to 35 cups per load — meaning roughly 10 to 15 reload cycles for 369 cups. The article records no keg-swap or reload logistics, so this is an estimate and not a fact, and should be read as such. What is on the record is this: no supervisor, no security escort, no incident-reporting channel.

That brings in the other corner of the picture, the one that put the whole matter into the news cycle. In early October 2026 a video from Yokohama went viral. A fan bought more than ten cups and did not drink them — he poured them into a disposal container. Japanese social media labelled the practice 'tarugai', a show of support for the vendor. I would not call it support. A transaction in which the purchased thing is destroyed and the real product is attention is a spectacle — and the centre of that spectacle is a service worker with no way to refuse the attention.

I want to stop myself here, because blaming the fan is the easy route and the easy route usually leads to the wrong page. The article quotes no rule, no policy, no sanction, and not a single statement from any operator or club. That absence is the real find. Ito's self-policing — no covert photographs, no following staff into the corridor — tells us such things happen; it also tells us the institutional machinery meant to stop them is not visible on the ground, so the burden has been shifted onto the self-discipline of a considerate spectator.

The incentive design and the ethical problem are the same problem here. Volume-linked pay and relationship-based selling combine to create a legitimate commercial motive for a fan to buy ten cups he will not drink. Pointing only at the fan's behaviour ignores the structural driver that converts that behaviour into revenue — and externalises the harm entirely onto the worker.

And Kato's remark that she was lucky never to have encountered a frightening or unpleasant situation should not be read as reassurance. It is a risk-management red flag. When safety is described as luck, it is luck and not safety. One vendor's good experience and one viral video are not opposites; they are the two tails of the same distribution.

My ledger carries two blanks. The first is the 15 kilograms, on which no load limit, no rotation, no mandatory rest is written. The second is the page on which a fan-conduct code or a staff-protection policy should have been written. I went looking for a labour story and found an absent policy.

The decision ahead splits two ways. If the operator tightens vendor-fan interaction, it kills the very engine that produces 369 cups a game, because that engine runs on relationship. If a protective framework takes shape — photo bans, reporting channels, section-level supervision — the same relationship can acquire a longer life. The date is not settled. The question is not why Mana Kato left; the question is which page her customer map was filed on, and when the operator's signature lands on it.