DRAFT · Updated 16 October 2026Downloads
Download the trend list, its scores against every scenario, the trend network and three years of weekly attention data as JSON and CSV, with licence, citation and versioning.
The trends
GINC 250 v2.0. The site is a view of these files; every number on a page traces to a field in one of them.
| File | Contents | Format |
|---|
| Trends | Every trend with its capability domain, description, tags and its score against each of the 20 scenarios | JSON |
| Trends, ranked | One row per trend with sensitivity and network measures | CSV |
| Trend by scenario scores | One row per trend and scenario, on the seven-point rubric | CSV |
| Trend similarity | One row per pair of trends at Moderate similarity or above | CSV |
| Trend network | Similarity matrix, themes, measures and method | JSON |
| Likelihood and impact | One row per trend: a provisional estimate, out of 100, of how likely it is to play out by 2035 and how far it would change national capability, with a one-line reason: the average of two AI models, and each reading | CSV |
| Second pass record | How the model panel revised the list: the panel, what changed and how far the panel agreed on the scores | JSON |
Attention
Weekly public attention to each trend over three years, for the world and by place.
| File | Contents | Format |
|---|
| Trend attention index | The weekly score for each trend, with the settings and the Wikipedia pages and search terms counted for each | JSON |
| Trend attention, weekly | One row per trend and week: the score and its two sources | CSV |
| GINC Share of Attention, weekly | One row per trend and one column per week: each trend's share of all the attention the list draws, as a percentage; each week adds up to 100 | CSV |
| GINC Share of Attention by place, weekly | One row per trend and place: its share of attention among the readers of the languages mostly read there, week by week, as a percentage | CSV |
| Trend attention, summary | One row per trend: level now, change over one and three years, peak and how far the sources agree | CSV |
| Trend attention by region and G20 member | One row per trend and place: level now and change over 90 days, 180 days and one, two and three years | CSV |
| Trend attention by place, full | What each place is read from, with the weekly index of every trend in every place | JSON |
| Trend attention by language | Wikipedia reading: one row per trend and language, one column per week | CSV |
| Trend attention by country | Google search interest across countries over the three years: one row per trend and country | CSV |
| Wikipedia reading by trend | The Wikipedia index in every language, readers by language and the pages counted | JSON |
| Google search interest by trend | The weekly worldwide series and the spread across countries | JSON |
| Google search interest by trend and country | The weekly series for each trend inside each country of the set, with the query asked | JSON |
The scenarios the trends are scored against, with their records, tests and results, are published at scenarios.ginc.org/data.
Licence
Content and data are published under CC BY 4.0. You may share and adapt them for any purpose with attribution. Code is published under the MIT licence.
Two sources carry terms of their own. Wikipedia page view counts are released by the Wikimedia Foundation under CC0. Google Trends figures may be reused with attribution to Google Trends; they are published here as harvested and as indices derived from them.
Citation
- The list. GINC (2026). GINC Trends v2.0: 250 trends scored against 20 scenarios. Oxford: Global Institute for National Capability. trends.ginc.org
- A trend. GINC (2026). Tariff barrier normalisation, GINC Trends v2.0. trends.ginc.org/trends/tariff-barrier-normalisation
- Attention data. GINC (2026). Trend attention index v1.1 [dataset]. Oxford: Global Institute for National Capability. trends.ginc.org/data
Stable addresses
A trend’s identifier and address are permanent. The download addresses on this page do not change between releases; the files carry their own version strings and the date they were built.
What depends on what
The trend list is the source. The scores against scenarios are part of it. Similarity, the themes, sensitivity and every network measure are computed from the list and can be reproduced from it. The attention data is harvested from Wikipedia and Google and does not feed any of the measures.
The scenarios themselves are not published here. This site keeps each scenario’s number, name and family so that the scores can be read; the records are at scenarios.ginc.org.
Versioning
The list follows semantic versioning: patch for wording, minor for rescoring or a change of method, major when trends are added, merged or dropped. The attention index is versioned separately, and each of its files carries the weeks it covers. Every change is in the change log.
