Companion data set · August 2026

Geostore comparative archive: 10,936 declarations, 2024 next to 2026

The 2026 binary answers the question "which values?". The March 2024 documentation answers "what does it mean?". This archive puts the two side by side: on one side the structure read in the binary, on the other the 2024 prose when it exists, and the tags removed from the client scope rendered as numbered holes rather than hidden.

loading the index...
Download everything Read the full study

Terms are combined with AND. Prefix a term with - to exclude it, wrap it in quotes for an exact phrase, or aim it at one field:

0 entries
unpacking the index...

Take the whole thing and dig on your own

The archive is served from one file, and that file is downloadable. No account, no form, no API key. Everything the page displays is in the dump, in the same shape: one object per declaration, plus a meta block with the counters.

Everything, one archive

ZIP · 567 KB

JSON and CSV together, plus a README that documents every column and every state. This is the one to take if you are not sure which format you want.

Download the ZIP

JSON

JSON · 4.2 MB

The full data set, indented, one object per declaration. Load it in a notebook, a script or a graph database and keep the parent / child structure intact.

Download the JSON

CSV

CSV · 2.2 MB

The same rows, flattened, semicolon separated, UTF-8 with BOM so Excel opens it without mangling the accents. Sort, pivot, filter.

Download the CSV

What each column holds

kindtype for a message or an enumeration, attribute for a field, signal for an enumeration value.
subkindmessage, enum, enum_value, optional, repeated, required, reserved, leak_model, leak_field.
name / parentthe declared name, and the message or enumeration that owns it.
package / filethe protobuf package and the schema file the declaration belongs to.
tagthe field number, or the number of the enumeration value. It is the stable identifier, not the name.
typethe declared type of the field, dotted names pointing at other messages of the corpus.
statusone of the six states described in the left column of the explorer.
model_2024name of the matching 2024 model when the two sides could be paired.
doc_2024the March 2024 prose, verbatim, never rewritten.

The corpus is declarative. It describes what the schema declares, not what the system does. A field that exists is not a field that is populated, and a signal that is declared is not a signal that is weighted. Enumeration values carry binary_only by construction: the 2024 documentation describes models and fields, not enumeration values.

Other sources and studies

Nov 2026 Google Shopping, from the SERP to AI Mode: what a product must prove to be shown, then bought, from four catalog scores (two of them drawn from clicks) and 82 filters to the five sheets the AI model reads and the universal cart Nov 2026 Twiddlers: what still moves a result after ranking, from the fan-out search plan, a Bard filter and a relevance oracle to E-E-A-T expertise, topic authority, spam and DMCA, with 120 twiddlers Oct 2026 Knowledge Graph: how Google feeds it, maintains it and makes it grow, from the passport every fact carries to the 194 access requirements that name its suppliers and the 8,679 collections that decide which requests an entity can fill Oct 2026 AI Overviews and AI Mode: what Google’s AI reads before it answers, from 138 result types to Knowledge Graph facts that reach the answer with no URL, with 9 internal schemas to browse Oct 2026 How Google matches pages to search intent: Intent Brain measures twenty broad intents from clicks for every query, language and country, Orbit sorts every page into 86 categories, with 15 internal schemas to browse Oct 2026 KScorer: how Google decides what a search is after, with the 56 signals annotated by its engineers (clicks that predict the rater score, the weight of the top 10, entities, context) and 15 internal schemas to browse Oct 2026 How NavBoost and Glue really work: everything Google remembers about every search, field by field (time on screen, skipped results, reformulations, AI Overviews), with 137 internal schemas to browse Oct 2026 Date a Knowledge Graph entity from its MID (calibrated on nearly 27,000 dated entities), and Google's original codec, rebuilt, that turns a MID into a 64-bit integer Oct 2026 Google BotGuard: what our tests reveal about access to results, session state, the checks run in the browser (up to four VMs on one results page) and SERP reliability Oct 2026 How Gemini searches and chooses its citations: 222,753 Google searches analysed across 76,450 public conversations More stuffs...
RESONEO