Method
Every number on the board is either copied from a named source or computed from public pageview counts. Nothing is estimated by us, and nothing is weighted by anyone's opinion of a cause.
Deaths a year
Each toll is copied from the cited publisher, almost always a World Health Organization fact sheet, with the exact phrase quoted on the cause's page. A weekly job fetches every source page and checks that the quoted phrase is still there. If a publisher revises a figure, the row is marked under review until the new figure is entered; the old figure is not silently kept as if confirmed. Where a source gives a range, the board uses the midpoint and shows the range.
Rows overlap. The tolls come from different methods and years (2000–2019 average, 2019, 2021, 2021 model; likely underreported, 2021, total mortality burden, 2022, 2023, 2024, 2025, annual). Some count deaths caused by a condition, others deaths attributable to a risk factor, and AMR counts deaths associated with resistant infections. The same death can appear in more than one row: a child who dies of diarrhoea is in both "Unsafe water, sanitation and hygiene" and "Diarrhoeal disease in young children". The board compares attention per death within each row. It never adds rows together.
Attention
Attention is the number of views of English Wikipedia articles about the cause, by human readers (the Wikimedia
user agent type, which excludes identified bots and crawlers), on all devices, over the last 12 complete months:
November 2025 to October 2026. Each cause sums up to four articles, listed below and on its page, so that a cause is not penalised for
having its readers spread across several titles. When an article is renamed, both titles are counted. Data: the
Wikimedia REST API, refreshed weekly; a new month appears after
Wikimedia publishes it.
Honest limit. English Wikipedia is a proxy for anglophone attention, not world attention, and it undercounts causes discussed mostly in other languages. Article choice matters at the margin. The range on the board spans several orders of magnitude, which no plausible proxy error closes, so read the ranking as directional.
The reference row
The benchmark is shark attacks: 12 confirmed shark-related deaths worldwide in 2025, according to the International Shark Attack File, against 176,427 views of the Wikipedia article "Shark attack". It shows how much attention the world can give a rare death. An earlier prototype used a recent armed attack as the benchmark; that read as ranking the victims of a political event, so a neutral, famously over-covered cause of death replaced it. The reference row is never assigned.
Views per death and the neglect index
views per death = views in the 12-month window ÷ deaths a year. The board is sorted by this, lowest first.
The neglect index = log₁₀(deaths ÷ views) − (the same value for the most-watched cause) + 1, so the
most-watched cause scores 1 and the least-watched scores highest. It is logarithmic because raw deaths-per-view spans
several orders of magnitude: on a linear scale the single most neglected cause would win every assignment, whatever the
person could do.
The assignment
You rate nine capabilities (capital, build, manufacture, policy, distribute, clinical, field, science, convene) from none (0) to decisive (3). Each cause carries a rating of how much it needs each capability. Then:
score = neglect index × fit² × rarity × scale
- Fit is the cosine similarity between what you declared and what the cause needs: a match of shape, not of volume, so declaring more does not raise your score, only declaring the right things does. It is squared so that what you can do dominates and neglect breaks ties between things you can do.
- Rarity (thousands 1, hundreds 2, dozens 4, a handful 8) and scale (myself 1, a team 2, a company 4, capital 6, an institution 8) multiply your score. They are shown on the assignment, but they apply to every cause equally, so they never change which cause you get.
- The cause with the highest score is your assignment; ties go to the more neglected cause. The three runners-up are shown. If nothing you declared matches anything a cause needs, the board says so instead of forcing a fit.
Who rated the needs. The need ratings and the "what would move it" summaries were written by the people who built this board from the cited WHO material. They have not been reviewed by public-health experts or by the organisations listed. Corrections are welcome.
The falsifier
If no assignment issued here produces a resolution condition that comes true within 180 days of launch (22 November 2026 to 21 May 2027), the board is a feed with extra steps. The public record shows the running count and will say so.