The empty room.

This article exists because the author called his own numbers unbelievable.

The message that started this, verbatim, from Mikal to the tools doing the counting: I don't believe you. Surely we miss ALL the facts and it's closed off somewhere?

That is the correct reflex, and this site is built on it. A finding you cannot bring yourself to believe is a finding that has not yet survived your best attempt to kill it. So the attempt was made, twice over: first a census wide enough that the machinery could not hide from it statistically, then a hunt through the specific doors it would live behind if the census had somehow missed it.

The rules were fixed before either began. Count both sexes with the same instrument. State whatever runs against us first. Hold every raw pull so the counts can be rerun by anybody. And if the machinery turned up, publish it prominently and gladly, because this site would rather be wrong than be right about this.

Every indicator and dataset name in five of the world's statistical systems, counted the same way.

The question is narrow and checkable: how many series in each registry are dedicated by name to one sex? Not disaggregated, which is common and good, but named: a series whose subject is that sex's outcomes.

The six at WHO deserve to be read in full, because they are the finding inside the finding. Four track men who have sex with men as a population in HIV and syphilis surveillance. One counts reported cases of urethral discharge. One measures exposure to second-hand smoke. Every one of the six tracks men as a transmission route or a symptom count. Not one treats men's health as an outcome in its own right.

Eurostat's single male entry is a historical hospital series whose female twin appears under slightly different labelling: an artefact of cataloguing, not a dedicated instrument. The OECD's public catalogue contains no dataflow naming men or boys at all.

A census like this is worthless if the method flatters the finding. Here is the method, and here is what it found for the other side.

Two rules make the count fair. First, whole-word matching with the obvious trap handled: male must never match inside female. Second, and more important, a pairing pass: any series naming one sex whose title with the sexes swapped exists in the same catalogue counts as paired, and paired series are not counted as dedicated attention for either side. A hyphen must never decide a finding, so the matching ignores punctuation, and the one bug that would have inflated our own numbers was caught by reading the residuals before storing anything.

What the fair count found against us is stated here, first and in full. The World Bank pairs systematically. Thousands of its male-named series exist as the twin of a female-named series: literacy male beside literacy female, unemployment male beside unemployment female, across the whole catalogue. That is the counter-finding, and it matters more than a defensive caveat: it proves the fair model is buildable, cheap, and already running at the largest development database on earth.

And of the female-dedicated series that remain unpaired, a visible share measure pregnancy, birth and antenatal care, which have no male counterpart by nature. The same rule this site applied to itself when its own healthy-life-years claim collapsed applies here: state it, or publish nothing.

After both corrections, the pattern still stands in every registry on every continent. The dedicated attention runs from five-to-one at the fairest system to literally unmeasurable at the OECD, where the count on one side is zero.

Then the doors, one by one: the places the machinery would live if the census had missed it.

A census counts names, and machinery could in principle exist without naming itself in a catalogue. So the hunt went to the specific addresses, the ones the system's own defenders would point to. What follows is everything found, stated as strongly as the evidence allows, because this list is the standing steelman of this entire site and it grows every time anything real turns up.

That is the whole catalogue of what exists: real documents, one completed regional strategy, one national study pending verification, and partner framing. Worth having. None of it is a standing entity, a mandate, an indicator set, a marker or a budget line.

The last door is the one you can open yourself, right now.

The World Health Organization maintains a content page for every health topic it works on. There are 247 of them. Women's health has one, carrying 14,249 characters of overview, impact and programme links. Maternal health has one. Violence against women has one. Female genital mutilation has one.

Type who.int/health-topics/mens-health into a browser and the page loads. It is real. It has the site's navigation, the regional links, the footer, the copyright line.

It has nothing else. 576 characters of chrome and no content. The template slot exists in the system, and nothing was ever put in it. Search the full 247-topic index for men and the matches are Meningitis, Menopause and Mental health.

Nobody decided this. There is no memo to find, no vote to point at, and this article claims neither, the same way it claimed neither about the OECD's contact form or the aid database's marker list. That is precisely the finding, the same one at every layer of the system for twenty-two years: categories get built from what has desks, and no desk was ever built. The page is not a scandal. The page is a vacancy.

The door exists. The room is empty. The address is the application form.

Written before anybody has to ask.

This is not a claim that the UN is bad. The UN's own studies produced half the receipts on this site. It is not a claim that UN Women is bad, and this article does not mention UN Women's work except to note that its model demonstrably functions, which is the argument for the ask, not against anybody.

The claim is exactly as large as the evidence: the machinery is missing. No entity, no mandate, no indicator set, no marker, and a topic page with nothing in it. Every count above can be rerun by anybody from the held raw pulls, and every door can be opened by anybody with a browser.

The ask has not changed size either: machinery for men and boys inside the UN, of the kind that demonstrably works, that takes nothing from women and girls. The letters asking the owners themselves whether anything was missed are already on nine desks with a reply date of 9 September 2026. If any reply points at a room we called empty and shows it furnished, that reply gets published prominently and gladly, because it would be the best news this initiative could receive.

Until then, the census stands at 47,100 names counted, and the hunt stands at one empty page, held, hashed and linked below.