Methodology

What counts as a development

A development is an event that moves a nation's capability — a policy decision, a technological milestone, an economic turn, an external shock. The test is consequence, not coverage: an entry earns its place if a reader would need it to understand where the nation now stands, not because it led a news cycle.

Each entry is written as a sentence with its parts separated: who acted, what they did, when, and — in the body — what followed from it. Keeping the parts apart is what lets a run of entries be read as a thread rather than a list of headlines, and what lets the same record be queried rather than only read.

One thread, two lenses

Every development is held once, in one thread of 120,798. It belongs to the nations and capabilities it concerns by graded joins, so it can be read from either side: a nation's timeline follows what happened to it, a capability's how that capability came to be, wherever it happened. The Timeline is the whole thread.

The site this replaced kept a recent thread, summarised from reporting, apart from a historical one of established record, as independent timelines behind a switch. The record no longer draws that line: an entry's date places it, and its provenance says how it was made.

Classification

Every development carries exactly one Level-1 category, which sets the colour and icon of its mark on the thread and is what the kind filter selects. The order is the method. An event is tested against the seven categories in the sequence below and takes the first one it passes — so a NATO summit is Geopolitical rather than Security, an earthquake is Environment rather than Governance, and a chip-fab subsidy is Technology rather than Economy. Without the order the categories overlap on almost every interesting event; with it, they do not.

Level 1

Tested in this order. The number on each card is its position in the sequence, not a rank of importance — Society is last because it is what remains once an event is not primarily physical, forceful, inter-state, authoritative, monetary or technical.

1. 1Environment

   The prime mover is physical or biological rather than an actor.

   Natural hazards and disasters · climate and weather · disease and biological events · water, land and food systems · ecosystems and pollution · industrial and technological accidents.

2. 2Security

   The act itself is the use, deployment or generation of organised force, other than a discrete seizure or transfer of state authority (see test 4).

   Armed conflict · operations and posture · force development and procurement · irregular and non-state violence · intelligence and covert action · cyber and electronic operations · internal security and policing · border enforcement.

3. 3Geopolitical

   The principal relation is between states or other international actors.

   Diplomacy and relations · alliances and blocs · treaties and international law · sanctions and coercive statecraft · territory and sovereignty · foreign assistance and basing · international institutions.

4. 4Governance

   The act concerns who holds state authority, or how the state itself is constituted and run.

   Executive power and leadership · elections and political competition · constitutional and institutional change · law and regulation · judiciary and rule of law · regime stability and political violence · state capacity and administration · subnational relations.

5. 5Economy

   The act is a transaction, price movement or policy over money and markets.

   Macroeconomic conditions · fiscal and monetary policy · debt and financial stability · trade and investment · industry and corporate events · energy and commodities · labour markets.

6. 6Technology

   The content is a technical or industrial capability.

   Research and discovery · development and testing · deployment and adoption · industrial base and manufacturing capacity · technology transfer and control · standards and technical governance · infrastructure and networks.

7. 7Society

   The content is the population itself.

   Civil unrest and mobilisation · population and migration · health and wellbeing · education and skills · identity, religion and culture · crime and social order · inequality and cohesion.

Level 2

The facets listed inside each card are Level 2. They are not assigned per event — no development carries a Level-2 tag. They exist to interpret and argue with a Level-1 call: they say what a category is understood to contain, which is what makes a borderline assignment checkable rather than a matter of taste.

There is no Legal category. Legal is an instrument — statute, ruling, treaty, regulation, executive order — not a kind of event, and it cuts across every category above. A treaty is Geopolitical, a constitutional ruling is Governance, a tax statute is Economy; what makes them different is what they act on, not that they are law.

Applying the order

Five recurring decisions do most of the work. They are written down because a rule you can state is a rule a reader can disagree with.

Force or sovereignty

A military verb — invaded, struck, launched an intervention — is Security. A sovereignty verb — annexed, ceded, recognised, transferred, gained independence — is Geopolitical, even where force was the means. So Turkey's 1974 intervention in Cyprus is Security; Russia's 2014 annexation of Crimea is Geopolitical.

Force or seizure

Test 2 stops short of the one thing force is most often used for. Where the act is a discrete seizure or transfer of state authority — a coup, a putsch, a revolution that installs a new government — it is Governance, because what the event settles is who holds the state, not that force was the instrument. Turkey's 1960, 1980 and 2016 coups and Park Chung-hee's 1961 seizure are Governance; Tiananmen, Gwangju and the Dirty War stay Security, because sustained repression by a state already in power is force without a transfer of authority.

Treaty or deployment

Signing a defence pact is Geopolitical: the act is a relation between states. Committing forces to a specific theatre is Security: the act is deployment. The Indonesia–Australia security treaty is Geopolitical; the Anglo-French declaration on a reassurance force for Ukraine is Security.

Money or capability

Budgets, rates, tariffs, valuations and financings are Economy. Where the content is a named technical capability — semiconductors, satellites, a rail network, generation capacity — it is Technology. India's Semicon 2.0 is Technology; Mistral's data-centre financing is Economy.

Authority or population

Governance is for acts about authority itself: elections, constitutions, courts, executive power, regime survival. Where the content is the standing of a population — an apology, a truth commission, abolition, a health system — it is Society, even when the instrument is a law.

Cross-border trade agreements sit in Geopolitical rather than Economy for the same reason: the ordered test reaches Geopolitical first, and a bilateral free trade agreement is a relation between states before it is a market event. Domestic tariffs, prices and budgets stay in Economy.

Where the assignments live

Every entry carries its category as its own kind, and the corpus's build rejects any entry whose kind is not one of the seven. The Atlas files the record began from used an older five-category vocabulary (econ, tech, global, policy, external); every entry was re-expressed in Level 1 when the corpus was imported, keyed on each event's identity rather than its position, so that no entry was quietly mis-sorted.

How a development belongs to a nation or a capability

A development's nations and capabilities are joins, each with a grade. Anything weaker than medium is no join at all, so a capability's thread with medium included is still about the capability.

Very high

For a capability: the development is about it, and would be in a short, focused history of it. For a nation: it is the actor, or the development happens in it and is chiefly about it.

High

For a capability: it directly and substantially builds, reforms, weakens or tests it. For a nation: a direct party — signatory, counterpart, target, belligerent.

Medium

For a capability: a direct, material and specific effect on it, stated or plainly implied — never a passing mention or a thematic likeness. For a nation: materially and directly affected, and named or plainly implied.

A development has at most four capabilities, at most two of them very high, and may have none. A model gave every grade, reading the entry against the whole framework; they are a model's judgement, not a source's, and each is shown beside the entry. A capability's page opens on very high and high, and the reader can narrow it to very high or widen it to medium.

Anatomy of an entry

Every development carries the same fields. Sub-events hang off a parent entry and take the parent's category, colour and icon, so a consequence is attached to the event that caused it rather than competing with it on the thread.

who

The actor. A government, an institution, a named person, or the event itself where no actor is meaningful (“The financial crisis”). At most 80 characters.

act

What they did, written as a verb phrase continuing from who. The decisive noun is wrapped in <b> so it can be scanned out of the line. At most 160 characters.

date

Written for people, not machines: July 2026, 1944–45, 1890s, 17th century, 221 BC. See ordering below.

when

The same date as a machine reads it, written by the corpus's one date reader. Sorting is done on this.

rel

How long ago, in plain words — last month, 18 years ago — counted from the day the page is read.

body

What followed from it, in two or three sentences, at most 500 characters. This is the part that earns the entry its place.

kind

One of the seven Level-1 categories above.

tier

How much it mattered to its subject: 1 central, 2 material, 3 notable, as a model judged it.

subs

Optional sub-events: a consequence or a counterpart with its own actor, action and date.

nations · caps

The nations and capabilities it belongs to, each with its grade.

source · provenance

Where it came from and how it was made, and whether a source has confirmed it (see below).

Ordering and dating

Dates are authored for readers; the corpus's one date reader turns each into a machine date that sorts. Ranges sort on the year they began; a month or a day, where given, adds precision within the year; BC sorts negative; a decade or a century sorts at its start, before the years inside it. Entries dated alike keep a fixed order.

"July 2026" → 2026-07 · "1944–45" → 1944..1945 · "1890s" → 1890..1899 · "17th century" → 1600..1699 · "221 BC" → -221

Every thread runs newest first and can be turned round to read from the earliest. Plans and targets dated after today are left out unless the reader asks for them.

Sources and verification

The record began as the nation and capability timelines of the GINC National Capability Atlas, model summaries of English Wikipedia with no citation per entry. Everything added since says how it was made: recalled by a model, written from a news search result, or read from a text or dataset — Wikipedia's Current Events portal, the Library of Congress Country Studies, the CIA World Factbook, the UN Treaty Collection, SIPRI, the IMF, UCDP and others.

An entry is verified only once a web search found a page stating the same event, by the same actor, in the same year, and the words were found letter for letter on it; 2,326 are. Every other entry a model produced is marked unverified wherever it is shown, and a source on it is a lead, not a citation. Only a verified entry links to its source. An entry a source appears to contradict is marked disputed.

Coverage

120,798 developments across the 197 nations, 71 territories and other entities and the 288 capabilities of the National Capability Framework. Coverage is not even: China was read across every capability, the largest economies have the longest threads, and the record is thinner the further back it goes.

Build

The record is kept in a database, with every source that reported each development, and is written out as one file per nation or capability. The build checks every entry — its fields, its date, its category, its joins, the length of its text, that it is held once — before anything is published, and writes the indexes this site reads. A page reads its entries from those files as it shows them. The same record is reachable from any LLM through the GINC MCP connector.

Limitations

• Few citations. Most entries were produced by a model and are unverified; the original entries carry no source at all.
• Model judgements. Categories, tiers and grades were given by a model. Tiers are not comparable between nation and capability timelines.
• Selection is editorial. What is on the record is a judgement about what matters, not an exhaustive record; a different editor would choose differently.
• Coverage is uneven. A nation with a short thread is under-recorded, not quiet.