Sheet 1 · Edition 2026-08-30 02:52 UTC
Which governments publish what open data
Read directly from 31 national government catalogues and counted by subject. 25 of them can be counted that way; 6 cannot. Five of them publish no subject classification at all, so no subject figure can exist for them, here or anywhere. A further catalogue is registered and has not been read. The sheet draws each of those differently, because they mean different things. Nothing here is estimated.
- 785,080+
- Datasets counted
- 31
- Governments read
- 25
- Countable by subject
- 6
- Read, not countable
- 1
- Registered, not read
Coverage by subject and government
Sorting orders subjects within one government’s column. Catalogue sizes differ enormously between governments, so this is never a league table.
Cell states
A SOUNDINGA figure read from that catalogue. Where the figure has a page behind it the cell is a link and Enter opens it. Below 960px the exact figure is replaced by the sounding mark, because a six-digit count cannot be drawn truthfully in a narrow cell; the readout carries it in full at every width.
- 0
MEASURED ZEROCounted, and there are genuinely none. Printed in full ink on plain paper, exactly like any other sounding, because it is one. It is not a link: there is nothing to list.
UNSURVEYEDBlank paper inside a pecked boundary. Nobody has measured it, so nothing is claimed either way. It is not a zero, it never sorts as one, and the catalogue’s own size in the column head is still real. Where a whole catalogue is unsurveyed the boundary runs the length of its column, lettered with which absence it is.
NO SUBJECT VOCABULARYThe catalogue publishes no subject classification, so no figure can exist for it, here or anywhere. Its datasets are real and are counted in the column head.
Depth, drawn against the same column and no other
A sounding’s tint is its share of that government’s own classified catalogue, stepped one band per half-decade below the deepest share in the same column. Every column reaches its own full tint, so tints are never comparable sideways: a full tint in a small catalogue and a full tint in a large one say the same thing about their own columns and nothing about each other. The numeral is always the figure, and the readout gives each column’s datum.
- two decades or more shallower
- a decade and a half shallower
- a decade shallower
- a half-decade shallower
- the deepest sounding in that column
- A dataset can carry several subjects, so a government’s column can sum past 100% of its classified total.
- Shares describe the catalogue, not the government’s output: catalogue size mostly reflects how finely a country splits its publications.
- The UNCLASSIFIED row is not a subject. It has no row number, it is never ranked, and it carries no tint: its share is taken against the whole catalogue rather than the classified part, so it does not sit on this depth scale.
Read the matrix by row or by column
Read it by subject on the theme pages or by government on the country pages.