How this platform knows what it publishes

The principle · you own the weights

The tool never decides what matters. It presents ten decision dimensions; you assign each a weight from 0 to 10. A founder optimising for speed weights differently from one optimising for cost or capital. A fixed editorial composite, where the tool decides clarity is worth 30%, is the anti-pattern this project exists to avoid. No country total is computed until you have declared weights. That is the red line.

Traceability per cell

Every datum is a cell carrying a value (or a gap), its primary source with a direct URL, the date the value is valid for, when it was last verified, a confidence rating, and any caveat.

Coverage, stated openly

How many of the 208 dossier countries carry a real, sourced value for each metric, as of 2026-08-20. The remainder are honest gaps, not zeros.

Source priority

When two reputable sources disagree on a cell, the dataset follows a fixed priority, highest first. A higher tier overrides a lower one; a cell whose only support is tier 4 or 5 cannot carry high confidence.

1. 1Primary law · the statute or official gazette text itself.
2. 2The regulator · the competent authority’s own fee schedules, licensing guidance and registers.
3. 3Official datasets · OECD, UNESCO UIS, Eurostat, the ITU, the World Bank and equivalents.
4. 4Reputable secondary analysis · established law-firm practice guides and recognised industry trackers.
5. 5Press and news archives · to date events, never to source a legal or numeric value.

The ten dimensions

Within each dimension a subset of subfields is scored; the rest are descriptive, shown to you but not fed into the maths. A dimension’s score is the mean of its available scored subfields, with equal sub-weight · sub-weighting would reintroduce hidden editorial judgment. A gap subfield is excluded from the mean.

Absolute anchor scales

Each scored numeric metric maps onto a fixed, published 0 to 100 scale · scored against a real-world range, not min-maxed across the countries on screen. best scores 100, worst scores 0, clamped. A few metrics that span orders of magnitude (space VC, ecosystem composite, graduates, headcount) use a log scale so a mid-sized ecosystem is not crushed to near zero because one or two giants exist.

Relative min-max made a country’s score move when a peer was added or removed, and sank a genuine extreme (the United States on fee size or VC volume) for being the extreme rather than for being worse. An absolute anchor gives a stable, defensible number.

Categorical metrics use an explicit value-to-score map (full foreign ownership: yes 100 / restricted 50 / no 0; operator liability: limited 100 / unlimited 35). Secondary-sanctions risk is shown as context but deliberately not scored · a subjective rating is kept out of the maths.

The country total

Summed over the ten dimensions; a dimension you weight 0 drops out entirely. Because the scales are absolute, a country’s score is independent of the set on screen · adding or removing a comparator never changes another country’s number. Under equal weights the United States scores ~71.6 (11th of 58); France, Portugal and Finland lead on genuinely fast, cheap, open and treaty-complete regimes.

Gap handling and hard filters

You may set knockout rules (exclude any country without a comprehensive space law); filtered countries are removed before scoring and listed separately with the reason. If, for any dimension you weighted 7 or higher, more than 40% of its scored subfields are gaps, the country is badged “insufficient data for your profile” · it still appears with its partial data, but is held out of the ranked shortlist rather than scored on thin evidence.

Coverage and review tiers · named, honest

Each profile carries an honest tier. Reviewed means a named space-law reviewer has signed it (their name, role and date appear in the tool). Researched · not yet expert-reviewed is the 66-country core: sourced, dated and confidence-rated across all ten dimensions, but not yet signed by an external expert. Skeleton covers a further 131 economies on three officially-sourced pillars only · UN treaty status (UNOOSA), OECD FDI openness and corporate tax · with every venture-specific dimension an explicit gap; they are scored only on those pillars, always flagged provisional, and never rank among the researched set. No profile is fabricated as “reviewed” · inventing a reviewer’s name would breach the anti-fabrication rule. The mechanism exists so that, when a real practitioner signs a flagship jurisdiction, their name appears.

The capability ladder · what the tier means

Every country dossier carries a capability tier, from Leading to No agency. It is an ALL IN SPACE assessment, not a rating issued by any agency or institution, and until August 2026 the country pages cited it to whichever body backed the agency record, which put this platform’s judgement in someone else’s mouth. The rungs read:

• Leading · sustained independent programme with its own launch capability and crewed or deep-space missions.
• Advanced · substantial national capability across several domains, with or without its own launcher.
• Emerging · a national agency operating satellites, without indigenous launch.
• Nascent · early native capability building, first satellites or first partnerships.
• User · a consumer of space services procured from others.
• No agency · no national space agency on record. Not the same as no institution: the dossier may still name an observatory or a university group, and says which it is.

Where the rung is a judgement rather than a rule, the dossier shows the evidence beside it, so you can disagree with the label and still use the facts: whether the country launches to orbit from its own territory, how many launch sites it holds, and how many of its nationals have flown. Several countries sit high on institutional weight and industrial contribution while holding none of the three, and the page says so rather than letting the rung imply otherwise.

The launch site roster · how it is built

Thirty-three countries hold at least one launch site and seventy-seven sites are on record. Which sites a country holds is an ALL IN SPACE compilation, assembled from national space agencies, armed forces, civil aviation authorities and the companies that operate the sites. Each site carries its own official source, shown beside it, and that source is the thing to check. Until August 2026 the roster on all 208 dossiers was credited instead to a single citation whose publisher field listed a dozen agencies at once and whose link pointed at the front page of a news site, which named nobody accountable and could be verified by no one.

An empty roster means no site was identified, which is a statement about this platform’s search and not a finding published by any agency. One hundred and seventy-five dossiers sit in that state, and the same news-site link was previously offered as the source for their absence. Absence of evidence is recorded here as absence of evidence. Where a country runs sounding rockets, hosts another state’s facility, or has announced a spaceport that has not been built, the summary says which of those it is rather than leaving the empty list to speak for itself.

Audience reach · what the percentages mean

A dossier shows a weekly reach percentage beside a public broadcaster and beside each top newspaper. Until August 2026, four hundred and nineteen of those numbers had no citation of their own: the section carried one general audience report and every individual figure rode on it. Chile showed its public broadcaster at twelve per cent, sourced to a report that states twenty-nine for that brand.

Every figure now carries a source of its own, and that source is allowed to say the figure is disputed or that no measurement exists. It is not allowed to be absent. One hundred and forty-four figures were removed because no official or industry measurement of them could be found, and twenty-nine more were withdrawn because the outlet they hung on is not the right one for the field. A further seventy-two were withdrawn for a subtler reason: the page showed one number and the source beside it stated another, the gaps ran one way with the sourced figure almost always higher, and that is the signature of a different measurement rather than a correction. Asking whether someone watched a broadcaster at all last week is not asking which brands they used for news. Publishing either number would assert something nobody measured.

Read the base, not just the number. Most surviving figures come from the Reuters Institute Digital News Report, which surveys online news consumers, not the whole adult population. Where internet penetration is sixty per cent those are very different denominators, so each figure’s note states the base it was measured on.

The national space agency record

Which body a country designates for space is recorded per dossier. Until August 2026, seventy-six of the two hundred and eight named this platform’s own database file as part of the provenance for that record, and fifty-two of those linked to a Wikidata query endpoint, which serves a query form rather than a page: a reader who followed it reached no claim at all.

The obvious repair was to promote the UNOOSA Worldwide Space Agencies directory, which already sat on the same records. It does not say what that repair would have claimed. The directory carries forty-five entries, forty-two of them national, so it is silent on thirty-eight of the seventy-six, and where it speaks it is not always about the same body: it lists the China Manned Space Agency where this record names CNSA, the civil administration, and CDTI for Spain, the delegate that preceded the Agencia Espacial Española. So UNOOSA is cited only where it independently names the same body, and where it names a different one, the citation says which. Where no source has been found, the citation says that too and counts as an open gap rather than evidence.

Which sources are admissible, and for what

A source good enough for one kind of claim is not automatically good enough for another, so the rule here is about the pairing rather than the publisher.

For a number, a rate, a fee or a deadline the platform cites the instrument that sets it: the statute, the regulator’s fee schedule, the official gazette, the agency’s own document, or a statistical authority such as Eurostat, the UNESCO Institute for Statistics or the OECD. A law firm’s summary of a fee is not the fee, and an encyclopedia entry about a tax rate is not the tax code. Six jurisdiction cells were withdrawn in August 2026 for resting on Wikipedia, and each says so where it used to sit.

For the identity and existence of a thing · that a satellite exists, who operates it, where a spaceport is · the platform uses Wikidata for 7,371 satellites, 1,085 companies, 107 spaceports and 103 ground stations, and names it as the source on every one. That is a different kind of claim: a structured, versioned, individually citable statement that a reader can open, rather than prose. It is still not a primary source, so nothing scored, ranked or presented as a measurement rests on it, and where a national register or an operator’s own filing exists that is what gets cited instead.

The honest summary is that Wikidata tells this platform what to go and look for, and something else has to tell it what the number is.