This page exists so you can check one thing: that making the console readable did not lose anything. For every card it lists each fact that was on it before — every number, every real piece we quoted, and every one of your own banked words — and says where that fact is now.
Carried means it is still on the face of the card. Moved into the fold means it is still there, one click away under “the detail”. Cut means it is gone, and every cut carries the reason it went. Nothing can be cut without a reason written down — the page refuses to build.
1 · The mechanism & the proposals
What sits above the type labels
Before
Every piece we save carries one label that says what kind of thing it is — a brand film, a tutorial, a press release. This card decides what sits ABOVE those labels. The answer you leaned toward on 16 August (“I want to take your recommendation, but a few call-outs”): nothing new gets written on any piece. The shelves are just how the label list is organized for browsing, and your five worlds become saved views the product computes. Since then every call-out has landed — your shelf names are applied, three types are off the list, the identity shelf is named. The tree at the top of this console shows the whole before-and-after in one look, and the right side moves zero pieces. Confirming here makes it final.
The registry today: 74 content-type rows · 53 active · grouped by 10 parent values (the tree view, left pane)
Stored per piece: one content_type value · 36 distinct values in use · 6,393 of the 13,938 live pieces untyped
What this option stores: nothing new per piece — 64 registry rows re-pointed to your shelf names
The industry pattern: type stored, family as grouping, rich views computed (the evidence page, rules 1·2·7)
◮: The biggest number in the whole registry sits under the family everyone calls empty: social_native_community parents social_post (2,518) + social_clip (943) — 3,461 pieces filed under a family with zero ratified members. That is the mixed-carve debt this ruling retires.
★: Confirm the mechanism: type on the piece · family as registry shelf · worlds as lenses.
if you approve: Closes proposal 1. What changes: 64 registry parent pointers, in the pipeline lane. What does NOT: no piece is touched, nothing re-classified, $0 today. What it unblocks: the shelf cards below become executable — and the pipeline's switched-off second routing step (which still keys on the old family names) finally learns what to key on instead.
Now
Nothing new gets written on any piece. Shelves are only how the label list is grouped for browsing, and your five worlds are views the product works out.
Today: 74 label rows, 53 switched on, 52 of them under 10 family names
Your carve: the same 64 labels, regrouped into 12 shelves
Written on a piece: one type label — unchanged
Pieces that move: none
The industry pattern: type stored, family as grouping, views computed
Nothing about the way pieces are tagged today changes.
◮: The family everyone calls empty is the biggest one stored: 3,461 pieces sit under social-native.
★: Confirm the mechanism: the type on the piece, the family as a shelf, the worlds as views.
if you approve: Twenty-one labels get re-pointed and 43 arrive new, all in the build lane; no piece is touched.
folded · If the tree at the top of this page reads right, you have already pre-read the twelve shelf cards. You leaned toward this on 16 August — “I want to take your recommendation, but a few call-outs” — and every call-out has landed since: your shelf names are applied, three types are off the list, the identity shelf is named.
folded · Stored per piece today: one type value and 36 distinct values in use. Of the 13,938 pieces that are not deleted (13,944 rows in all), 6,393 carry no type at all.
folded · The empty family in full: social-native shelves the two labels stored as social post (2,518 pieces) and social clip (943) — 3,461 pieces under a family with no ratified members at all. That is the mixed-carve debt this ruling retires.
folded · What it unblocks: the shelf cards below become executable, and the pipeline's switched-off second routing step — whose twelve rows still key on the superseded mixed list of categories — finally learns what to key on instead.
The social-native family
Before
On your approved list, the family called “Social-native & community” is empty — you never approved a single type under it. But in the database it is doing the opposite of nothing: it is the shelf over social_post and social_clip, the two biggest stored labels in the whole library (2,518 + 943 pieces). Those two are already headed for a split — “social” will move to the where-it-ran fields and each piece will get its real type (the leftover cards seal that). So the honest question is what remains: is anything genuinely BORN social? Two candidates qualify — a creator collab and a meme format. An editor would call a piece “a meme” even with the platform hidden, which is the test every type on the list passed.
Ratified members: zero
Registry members (live): social_clip · social_graphic · social_post — 3,461 pieces
A real social_post: “GPT-5.5 is here! Have been daily driving this model…” — c334aa5a
A real social_clip: “Generated by Sora, without modification” — 3b8b98e3
★: Name Creator Collab and Meme Format as types; drop the family (they shelve under your existing shelves).
if you approve: Two candidate types join the list on your mark; the family name retires with the migration. The 3,461 split-bound pieces are handled by the leftover cards either way — nothing here blocks them.
Now
You never approved a single type under “Social-native & community”, yet it shelves the two biggest labels we store. What is left to decide is whether anything is genuinely born social.
Types you approved under it: none
What it shelves today: the labels stored as social post and social clip — 3,461 pieces
Where those pieces go: each gets its real type; “social” becomes a where-it-ran field
What could stay: a creator collab and a meme format
Stored as social post: “GPT-5.5 is here! Have been daily driving this model…” — c334aa5a
Stored as social clip: “Generated by Sora, without modification” — 3b8b98e3
Approving both would take the ratified set from 64 labels to 66.
◮: Both candidates pass the same test every label on the list passed: an editor would still call it “a meme” with the platform hidden.
★: Name Creator Collab and Meme Format as types, and drop the family.
if you approve: Two types join the list on your mark, and the family name retires with the migration.
folded · The two labels split 2,518 pieces stored as social post and 943 stored as social clip, which is where the 3,461 comes from.
folded · Those two are already headed for a split: “social” moves to the where-it-ran fields and each piece gets its real type. The leftover cards seal that, so the 3,461 split-bound pieces are handled either way and nothing here blocks them.
Which five worlds are the lens set
Before
Two versions of your own carve exist. July (ratified): Brand Identity & Systems · Campaigns · Content · Digital Products & Experiences · Physical Products & Experiences. August (dictated at the walk): “brand and identity, campaigns and content, events and experience, digital…, physical…” — Campaigns and Content merged, an events world added. The recommendation keeps the July five as the lens set: Content is your always-on world precisely BECAUSE those pieces belong to no campaign or other container, and events are already a ruled container kind that any world's pieces attach to. Amending is genuinely yours to make — it amends ratified ground, so say it plainly.
Ratified July worlds: Brand Identity & Systems · Campaigns · Content · Digital P&E · Physical P&E
Your August dictation: campaigns+content merged · an events world added
Events today: a grouping kind (event/season), ruled — campaigns attach to events
★: Keep the July five as lenses; events stay a container kind.
if you approve: Lenses are computed, so this costs nothing either way — the worlds are saved views, and renaming or recutting them later moves no data.
Now
In August you dictated a different set of worlds from the five you ratified in July. The lean is to keep the July five.
Ratified in July: Brand Identity & Systems · Campaigns · Content · Digital · Physical
Dictated in August: campaigns and content merged, an events world added
Events today: already a container kind — campaigns attach to events
Content is the always-on world precisely because those pieces belong to no campaign.
★: Keep the July five as the lenses; events stay a container kind.
if you approve: Nothing moves either way — the worlds are saved views, and recutting them later costs nothing.
folded · Your August dictation, word for word: “brand and identity, campaigns and content, events and experience, digital…, physical…” — Campaigns and Content merged, and an events world added.
folded · Amending is genuinely yours to make. It amends ratified ground, so if you want the August set, say so plainly and it becomes the ruling.
The two parked family names
Before
Families 9 and 10 (visual & identity assets · event, retail & interactive) had a parked naming question. Under your carve it dissolves: their member types regroup as Brand & Identity, Events & Experiential, Retail & Commerce and Interactive & Digital Experience, and the old names retire with the family layer. Listed so nothing disappears silently — a mark on proposal 1 closes this automatically.
Status: rides proposal 1; no separate decision left if P-01 confirms
★: Confirm — closed by proposal 1.
if you approve: Nothing separate executes.
Now
Families 9 and 10 had a parked naming question. Under your carve their names retire with the family layer, so proposal 1 closes this.
The two names: visual & identity assets · event, retail & interactive
Where their types go: Brand & Identity · Events · Retail & Commerce · Interactive
Status: closed by proposal 1 — no separate decision left
Listed here so nothing disappears silently.
★: Confirm — closed by proposal 1.
if you approve: Nothing separate executes.
Ongoing piles like Customer Stories
Before
The live case grew while we walked: FOUR customer-story piles now exist, each stored as a campaign of type customer_story_campaign — and one of them is a second brand (Humans of HOKA), which is the first evidence this pattern generalizes past OpenAI. Your instinct at the gallery round: these are “always-on” campaigns — the industry's word too. The pick: (a) keep piles as campaigns and add always_on to the campaign-type vocabulary · (b) mint a new grouping kind (program/series) · (c) both.
Customer Stories: 30 pieces
Customer Proof: AI in Action: 9 pieces
Humans of HOKA: 8 pieces — a SECOND brand's pile, same shape
Customer Stories: AI in Action: 4 pieces
campaign_type registry: 21 active values; always_on not among them yet
◮: Humans of HOKA — the pile pattern already crossed brands on its own, before any pipeline asked it to.
★: (a) Always-on campaign — one vocabulary row, no new grouping kind, amends nothing ruled.
or: (b) a new program/series grouping kind · (c) both
if you approve: Under (a): one registry row in the pipeline lane; all four piles keep their storage. Under (b): a new grouping kind plus a re-file of the four piles. $0 either way today.
Now
Four customer-story piles are stored as campaigns, and one belongs to a second brand. The pick is whether a pile that never ends is a campaign type or its own kind of container.
Customer Stories: 30 pieces
Customer Proof: AI in Action: 9 pieces
Humans of HOKA: 8 pieces — a second brand's pile, same shape
Customer Stories: AI in Action: 4 pieces
The campaign-type list: 21 active values; always-on is not among them yet
Your own word for these at the gallery round was “always-on”, and it is the industry's word too.
◮: Humans of HOKA — the pile pattern crossed brands on its own, before any pipeline asked it to.
★: (a) Always-on campaign — one vocabulary row, no new container kind.
or: (b) a new program or series container kind · (c) both
if you approve: Under (a) one registry row changes and all four piles keep their storage; under (b) they are re-filed.
folded · Each pile is stored today as a campaign whose type is “customer story campaign” — the storage already exists, so (a) adds a word and nothing else.
Naming the middle level
Before
First, the ground you already ruled, because “work” landed badly as a word: a content item stays EXACTLY the smallest atomic unit — an Instagram post, a YouTube video, a hero image — each fully tagged on its own. The only thing above it is a LINK (one creative, many appearances — your ruling, aspect ratios included). The pick is purely what to CALL the middle when it helps: a 30-second 9:16 French cut is a version of the creative; the YouTube upload is an appearance of that version. Naming the middle helps dedup, metrics and rights later — and changes zero storage now.
Ruled ground: one creative, many appearances (aspect-ratio variants included)
What (b) adds: one word — “version” — between creative and appearance; no table, no column today
★: (b) Name the middle level — naming only.
if you approve: A vocabulary sentence in the model. Nothing about per-item tagging changes; content_item stays atomic under every reading.
Now
A content item stays the smallest unit and is tagged on its own. The only pick is what to call the middle level, which changes no storage.
Ruled already: one creative, many appearances — aspect-ratio cuts included
The example: a nine-by-sixteen cut for France is a version; the upload of it is an appearance
What (b) adds: one word — “version”; no table and no column today
A name for the middle helps us spot duplicates, and helps metrics and rights later.
★: (b) Name the middle level — naming only.
or: (a) keep the two levels as ruled — a creative and its appearances, with nothing named between
if you approve: A sentence of vocabulary in the model; per-item tagging does not change.
folded · “Work” landed badly as a word, so it is gone. An Instagram post, a YouTube video, a hero image each stay atomic and fully tagged, and the only thing above an item is a link between its appearances.
One type per piece, or several
Before
You flagged “secondary type” as sounding wrong, and your example IS the rule in force: a screen-recording of a feature being used is a whole piece whose type is product demo; a product demonstration inside a hero commercial is a MOMENT, never the commercial's type. The only open confirm: may a piece carry MORE than one whole-piece type — genuinely a testimonial AND a product demo end to end — with one marked primary (your August ruling), or exactly one?
Ruled in August: several-with-one-primary; a second type must describe the WHOLE piece
The moments pattern: already multi-shelf with one main shelf — the same shape, proven
★: (a) Confirm several-with-one-primary.
or: (b) amend to one type only — a second-type link stays addable later
if you approve: The moment rule stands untouched either way; (a) matches the moments pattern already in production.
folded · A WHOLE PIECE of the type — “A research preview of Operator…” — 8bb3e130 — end to end, this IS a product demo; product_demo is its type.
folded · A MOMENT inside a piece — The Nike Ada Hegerberg commercial — 0a4d8928 — contains demonstration beats; its type stays commercial, the demo lives at the moment grain.
Now
Your own example states the rule: a demo inside a commercial is a moment, not the commercial's type. The confirm left is whether a piece may carry more than one whole-piece type.
Ruled in August: several types, one marked primary
The bar for a second type: it must describe the whole piece, end to end
Primary: the one type that leads when a piece carries more than one
The moments pattern: already multi-shelf with one main shelf
You flagged “secondary type” as sounding wrong — “primary” is the word that survived.
★: (a) Confirm several types with one primary.
or: (b) one type only — a second-type link stays addable later
if you approve: The moment rule stands either way, and (a) matches the moments pattern already in production.
folded · A whole piece of the type — “A research preview of Operator…” — 8bb3e130 — end to end, this is a product demo.
folded · A moment inside a piece — The Nike Ada Hegerberg commercial — 0a4d8928 — contains demonstration beats, but its type stays commercial.
Content scraped from a product
Before
Your words: “should we probably put Digital Product somewhere… like where do we put that ChatGPT like ios and or web app etc — re content we scrape from it.” The answer: coming from a product does not change what KIND of piece something is. A screen captured from the ChatGPT app keeps its real type. WHICH product it came from is a connection we already store between the piece and the product. WHERE in the product it appeared (an app screen, a web page) is answered by the where-it-ran fields you already ratified. At most one honest new type — product interface screen — joins the Interactive & Digital Experience shelf if raw captures turn out to need a kind of their own.
The product layer, live: 139 products (every row entity-linked) + 394 product-line entities
What a stored family would add: a second copy of the source/touchpoint facts — the duplication the re-cut measured
★: (a) Product-ness is a LINK, never a family.
or: (b) a stored Digital Product family
if you approve: Under (a): zero new storage; at most one new type on your say-so. Under (b): a stored value on every product-scraped piece that must forever agree with the link that already exists.
Now
Coming from a product does not change what kind of piece something is. Which product it came from is a link we already store.
The question: where do we put the ChatGPT app and web content we scrape
The product layer, live: 139 products, each linked to its company, plus 394 product lines
Where it ran: answered by the where-it-ran fields you already ratified
What a stored family would add: a second copy of facts the link already carries
At most one new type — product interface screen — if raw captures need a kind of their own.
★: (a) Product-ness is a link, never a family.
or: (b) a stored Digital Product family
if you approve: Under (a) nothing new is stored; under (b) every product-scraped piece carries a value that must forever agree with the link.
folded · Your words: “should we probably put Digital Product somewhere… like where do we put that ChatGPT like ios and or web app etc — re content we scrape from it.” A screen captured from the ChatGPT app keeps its real type; the product it came from and the place it appeared are both already answered by machinery you have.
The Groupings rename — your glance
Before
Your banked words, in order: “maybe Groupings or Grouping Content idk… i still dont like works bc i never call anything really ‘works’… campaign & groupings isnt that bad.” Of the three names you floated, Groupings was taken (Grouping Content and Campaign & Groupings both failed your own tests — the second privileges one kind). The pillar is now Groupings, every cross-reference re-pointed, the old route genuinely 404s, and the display wording is “one creative, many appearances” — the ruling itself unchanged. Your two edge cases are answered ON the page: an event's pile is usually campaign-shaped (campaigns attach to events — already ruled), and app/web product content is an entity link, not a container.
The pillar today: pillars/groupings — live, renamed, all references re-pointed
★: Confirm the Groupings name.
if you approve: Nothing further executes — the rename is already live on the site; an amend is one config+page pass.
Now
Groupings was taken from the three names you floated, and the pillar is already renamed. A glance is all this one needs.
The pillar today: renamed, every cross-reference re-pointed, the old route 404s
The display wording: one creative, many appearances — the ruling itself unchanged
Silence here reads as confirmed.
★: Confirm the Groupings name.
if you approve: Nothing further executes; an amend is one config and page pass.
folded · Your banked words, in order: “maybe Groupings or Grouping Content idk… i still dont like works bc i never call anything really ‘works’… campaign & groupings isnt that bad.” Grouping Content and Campaign & Groupings both failed your own tests — the second privileges one kind over the others.
folded · Your two edge cases are answered on the page: an event's pile is usually campaign-shaped, because campaigns attach to events, and app or web product content is an entity link rather than a container.
2 · The shelves, one card each
Campaign
Before
The Campaign shelf exactly as your carve places it — every type, with live usage and a real stored piece where one exists. Of its 10 types, 4 are new (they arrive with the migration) and the rest carry 1,312 live pieces between them. Confirmed = these types sit on this shelf, as shown; if one belongs elsewhere, mark Amended and name the move in the note.
Brand Film: 136 in use
Campaign Film: new — no stored value yet
Manifesto / Anthem Film: new — no stored value yet
Launch Film: 1 in use
Teaser: new — no stored value yet
Trailer: 4 in use
Sizzle Reel: 10 in use
Brand Story / Origin Story: new — no stored value yet
Commercial / Spot: 55 in use
Promotional Ad: 1,106 in use
e.g. (advertisement_image · linkedin_ad): “L’IA accompagne le flux de travail de votre équipe…” — 4df5bd1d
e.g. (brand_film · tiktok): “🥹 We’re not crying, you’re crying. Today’s Sora Matinee: Thank you, mom” — 58e75c32
★: Confirm Campaign as carved — membership as listed.
if you approve: Nothing is stored by a mark — $0, reversible. A Confirmed shelf becomes registry re-pointing in the pipeline lane when proposal 1 closes.
Now
Every label your carve puts on this shelf, and how many pieces carry it today. If one belongs somewhere else, mark Amended and name the move.
Brand Film: 136 pieces
Campaign Film: new — arrives with the migration
Manifesto / Anthem Film: new — arrives with the migration
Launch Film: 1 piece
Teaser: new — arrives with the migration
Trailer: 4 pieces
Sizzle Reel: 10 pieces
Brand Story / Origin Story: new — arrives with the migration
Commercial / Spot: 55 pieces
Promotional Ad: 1,106 pieces
For example: “L’IA accompagne le flux de travail de votre équipe…” — 4df5bd1d
6 of the 10 labels are in use today, over 1,312 pieces; the other 4 arrive with the migration.
★: Confirm Campaign — these labels belong on this shelf.
if you approve: Nothing is stored by a mark; a confirmed shelf becomes registry re-pointing in the build lane.
folded · Another real piece on this shelf: “🥹 We’re not crying, you’re crying. Today’s Sora Matinee: Thank you, mom” — 58e75c32.
folded · Some labels here have a card of their own later in the walk, which may rename or retire them: Brand Film (N-04) · Campaign Film (N-03) · Manifesto / Anthem Film (N-02) · Launch Film (N-01) · Brand Story / Origin Story (N-10) · Commercial / Spot (N-08) · Promotional Ad (N-11). Confirming the shelf confirms where a label sits, not what it is called.
Brand & Identity
Before
The Brand & Identity shelf exactly as your carve places it — every type, with live usage and a real stored piece where one exists. ⚑ rows moved here from a different family and want your eye. Of its 6 types, 5 are new (they arrive with the migration) and the rest carry 10 live pieces between them. Confirmed = these types sit on this shelf, as shown; if one belongs elsewhere, mark Amended and name the move in the note.
Key Visual: new — no stored value yet
Logo Animation: new — no stored value yet
Product Shot: 10 in use
Template: new — no stored value yet
Brand Guidelines: new — no stored value yet
Icon System: new — no stored value yet
e.g. (product_shot · landing_page): “Choice AI phone screens in Furusato Choice app” — 7b189f27
★: Confirm Brand & Identity as carved — membership as listed.
if you approve: Nothing is stored by a mark — $0, reversible. A Confirmed shelf becomes registry re-pointing in the pipeline lane when proposal 1 closes.
Now
Every label your carve puts on this shelf, and how many pieces carry it today. If one belongs somewhere else, mark Amended and name the move.
Key Visual: new — arrives with the migration
Logo Animation: new — arrives with the migration
Product Shot: 10 pieces
Template: new — arrives with the migration
Brand Guidelines: new — arrives with the migration
Icon System: new — arrives with the migration
For example: “Choice AI phone screens in Furusato Choice app” — 7b189f27
6 of these labels moved onto this shelf from a different family under your carve — those are the ones to look at hardest.
★: Confirm Brand & Identity — these labels belong on this shelf.
if you approve: Nothing is stored by a mark; a confirmed shelf becomes registry re-pointing in the build lane.
folded · A flagged label moved onto this shelf from a different family under your carve, so it is the one to look at hardest.
Product & Feature
Before
The Product & Feature shelf exactly as your carve places it — every type, with live usage and a real stored piece where one exists. Of its 7 types, 5 are new (they arrive with the migration) and the rest carry 782 live pieces between them. Confirmed = these types sit on this shelf, as shown; if one belongs elsewhere, mark Amended and name the move in the note.
Product Film: new — no stored value yet
Feature Overview: new — no stored value yet
Product Demo: 751 in use
Product Walkthrough: new — no stored value yet
Explainer: 31 in use
Product Comparison: new — no stored value yet
App Store Video: new — no stored value yet
e.g. (product_demo · tiktok): “Our open models are here: gpt-oss reasoning models” — 257782ac
e.g. (explainer · linkedin_ad): “Three new enterprise capabilities for OpenAI deployments” — a3c9e027
★: Confirm Product & Feature as carved — membership as listed.
if you approve: Nothing is stored by a mark — $0, reversible. A Confirmed shelf becomes registry re-pointing in the pipeline lane when proposal 1 closes.
Now
Every label your carve puts on this shelf, and how many pieces carry it today. If one belongs somewhere else, mark Amended and name the move.
Product Film: new — arrives with the migration
Feature Overview: new — arrives with the migration
Product Demo: 751 pieces
Product Walkthrough: new — arrives with the migration
Explainer: 31 pieces
Product Comparison: new — arrives with the migration
App Store Video: new — arrives with the migration
For example: “Our open models are here: gpt-oss reasoning models” — 257782ac
2 of the 7 labels are in use today, over 782 pieces; the other 5 arrive with the migration.
★: Confirm Product & Feature — these labels belong on this shelf.
if you approve: Nothing is stored by a mark; a confirmed shelf becomes registry re-pointing in the build lane.
folded · Another real piece on this shelf: “Three new enterprise capabilities for OpenAI deployments” — a3c9e027.
folded · Some labels here have a card of their own later in the walk, which may rename or retire them: Product Film (N-05). Confirming the shelf confirms where a label sits, not what it is called.
Users & Customers
Before
The Users & Customers shelf exactly as your carve places it — every type, with live usage and a real stored piece where one exists. Of its 6 types, 3 are new (they arrive with the migration) and the rest carry 175 live pieces between them. Confirmed = these types sit on this shelf, as shown; if one belongs elsewhere, mark Amended and name the move in the note.
Testimonial: 16 in use
Customer Story: new — no stored value yet
Case Study: 149 in use
User Story: new — no stored value yet
Expert Endorsement: new — no stored value yet
Review: 10 in use
e.g. (case_study · linkedin_ad): “Singular Bank built Singularity — an internal assistant” — ea8769b9
e.g. (testimonial · linkedin_ad): ““GPT-5.5 with Codex is hands down the most creative assistant…”” — b951adbd
★: Confirm Users & Customers as carved — membership as listed.
if you approve: Nothing is stored by a mark — $0, reversible. A Confirmed shelf becomes registry re-pointing in the pipeline lane when proposal 1 closes.
Now
Every label your carve puts on this shelf, and how many pieces carry it today. If one belongs somewhere else, mark Amended and name the move.
Testimonial: 16 pieces
Customer Story: new — arrives with the migration
Case Study: 149 pieces
User Story: new — arrives with the migration
Expert Endorsement: new — arrives with the migration
Review: 10 pieces
For example: “Singular Bank built Singularity — an internal assistant” — ea8769b9
3 of the 6 labels are in use today, over 175 pieces; the other 3 arrive with the migration.
★: Confirm Users & Customers — these labels belong on this shelf.
if you approve: Nothing is stored by a mark; a confirmed shelf becomes registry re-pointing in the build lane.
folded · Another real piece on this shelf: “GPT-5.5 with Codex is hands down the most creative assistant…” — b951adbd.
folded · Some labels here have a card of their own later in the walk, which may rename or retire them: Customer Story (N-12) · User Story (N-12). Confirming the shelf confirms where a label sits, not what it is called.
Partnership & Creator
Before
The Partnership & Creator shelf exactly as your carve places it — every type, with live usage and a real stored piece where one exists. Of its 3 types, 3 are new (they arrive with the migration) and the rest carry 0 live pieces between them. Confirmed = these types sit on this shelf, as shown; if one belongs elsewhere, mark Amended and name the move in the note.
Creator Story: new — no stored value yet
Partner Story: new — no stored value yet
Sponsored Content: new — no stored value yet
★: Confirm Partnership & Creator as carved — membership as listed.
if you approve: Nothing is stored by a mark — $0, reversible. A Confirmed shelf becomes registry re-pointing in the pipeline lane when proposal 1 closes.
Now
Every label your carve puts on this shelf, and how many pieces carry it today. If one belongs somewhere else, mark Amended and name the move.
Creator Story: new — arrives with the migration
Partner Story: new — arrives with the migration
Sponsored Content: new — arrives with the migration
Every label here is new; this shelf arrives with the migration.
★: Confirm Partnership & Creator — these labels belong on this shelf.
if you approve: Nothing is stored by a mark; a confirmed shelf becomes registry re-pointing in the build lane.
folded · Some labels here have a card of their own later in the walk, which may rename or retire them: Sponsored Content (N-11). Confirming the shelf confirms where a label sits, not what it is called.
Education & Support
Before
The Education & Support shelf exactly as your carve places it — every type, with live usage and a real stored piece where one exists. Of its 8 types, 6 are new (they arrive with the migration) and the rest carry 227 live pieces between them. Confirmed = these types sit on this shelf, as shown; if one belongs elsewhere, mark Amended and name the move in the note.
Tutorial: 177 in use
How-To: new — no stored value yet
Guide: new — no stored value yet
Lesson / Course: new — no stored value yet
Webinar: 50 in use
FAQ: new — no stored value yet
Support Article: new — no stored value yet
Training: new — no stored value yet
e.g. (tutorial · tiktok): ““How I built Forever Land… using Sora + ChatGPT”” — cc7f1d48
e.g. (webinar · youtube): “Social learning in multi-agent RL | OpenAI Scholars” — 7cd66016
★: Confirm Education & Support as carved — membership as listed.
if you approve: Nothing is stored by a mark — $0, reversible. A Confirmed shelf becomes registry re-pointing in the pipeline lane when proposal 1 closes.
Now
Every label your carve puts on this shelf, and how many pieces carry it today. If one belongs somewhere else, mark Amended and name the move.
Tutorial: 177 pieces
How-To: new — arrives with the migration
Guide: new — arrives with the migration
Lesson / Course: new — arrives with the migration
Webinar: 50 pieces
FAQ: new — arrives with the migration
Support Article: new — arrives with the migration
Training: new — arrives with the migration
For example: “How I built Forever Land… using Sora + ChatGPT” — cc7f1d48
2 of the 8 labels are in use today, over 227 pieces; the other 6 arrive with the migration.
★: Confirm Education & Support — these labels belong on this shelf.
if you approve: Nothing is stored by a mark; a confirmed shelf becomes registry re-pointing in the build lane.
folded · Another real piece on this shelf: “Social learning in multi-agent RL | OpenAI Scholars” — 7cd66016.
folded · Some labels here have a card of their own later in the walk, which may rename or retire them: Lesson / Course (N-10). Confirming the shelf confirms where a label sits, not what it is called.
Corporate Communication
Before
The Corporate Communication shelf exactly as your carve places it — every type, with live usage and a real stored piece where one exists. Of its 9 types, 5 are new (they arrive with the migration) and the rest carry 626 live pieces between them. Confirmed = these types sit on this shelf, as shown; if one belongs elsewhere, mark Amended and name the move in the note.
Announcement Article: 176 in use
News Article: new — no stored value yet
Feature Article: 226 in use
Interview / Q&A: 114 in use
Thought Leadership: new — no stored value yet
Opinion: new — no stored value yet
Press Release: 110 in use
Newsletter: new — no stored value yet
Research Report: new — no stored value yet
e.g. (feature_article · newsroom_rss): “An OpenAI model disproved a central conjecture in discrete geometry” — 26b76486
e.g. (announcement · landing_page): “Disney and OpenAI reach landmark character agreement” — 03f0c94b
★: Confirm Corporate Communication as carved — membership as listed.
if you approve: Nothing is stored by a mark — $0, reversible. A Confirmed shelf becomes registry re-pointing in the pipeline lane when proposal 1 closes.
Now
Every label your carve puts on this shelf, and how many pieces carry it today. If one belongs somewhere else, mark Amended and name the move.
Announcement Article: 176 pieces
News Article: new — arrives with the migration
Feature Article: 226 pieces
Interview / Q&A: 114 pieces
Thought Leadership: new — arrives with the migration
Opinion: new — arrives with the migration
Press Release: 110 pieces
Newsletter: new — arrives with the migration
Research Report: new — arrives with the migration
For example: “An OpenAI model disproved a central conjecture in discre…” — 26b76486
4 of the 9 labels are in use today, over 626 pieces; the other 5 arrive with the migration.
★: Confirm Corporate Communication — these labels belong on this shelf.
if you approve: Nothing is stored by a mark; a confirmed shelf becomes registry re-pointing in the build lane.
folded · Another real piece on this shelf: “Disney and OpenAI reach landmark character agreement” — 03f0c94b.
folded · Some labels here have a card of their own later in the walk, which may rename or retire them: Interview / Q&A (N-10). Confirming the shelf confirms where a label sits, not what it is called.
Documentation & Technical
Before
The Documentation & Technical shelf exactly as your carve places it — every type, with live usage and a real stored piece where one exists. Of its 7 types, 5 are new (they arrive with the migration) and the rest carry 131 live pieces between them. Confirmed = these types sit on this shelf, as shown; if one belongs elsewhere, mark Amended and name the move in the note.
Technical Documentation: new — no stored value yet
API Documentation: new — no stored value yet
Developer Guide: new — no stored value yet
Release Notes: new — no stored value yet
White Paper: 126 in use
Specification: 5 in use
Technical Reference: new — no stored value yet
e.g. (white_paper · landing_page): “Forecasting potential misuses of language models…” — a709f1ca
e.g. (spec_release · landing_page): “OpenAI Ad Tools Data Processing Addendum” — 97aa2ed8
★: Confirm Documentation & Technical as carved — membership as listed.
if you approve: Nothing is stored by a mark — $0, reversible. A Confirmed shelf becomes registry re-pointing in the pipeline lane when proposal 1 closes.
Now
Every label your carve puts on this shelf, and how many pieces carry it today. If one belongs somewhere else, mark Amended and name the move.
Technical Documentation: new — arrives with the migration
API Documentation: new — arrives with the migration
Developer Guide: new — arrives with the migration
Release Notes: new — arrives with the migration
White Paper: 126 pieces
Specification: 5 pieces
Technical Reference: new — arrives with the migration
For example: “Forecasting potential misuses of language models…” — a709f1ca
2 of the 7 labels are in use today, over 131 pieces; the other 5 arrive with the migration.
★: Confirm Documentation & Technical — these labels belong on this shelf.
if you approve: Nothing is stored by a mark; a confirmed shelf becomes registry re-pointing in the build lane.
folded · Another real piece on this shelf: “OpenAI Ad Tools Data Processing Addendum” — 97aa2ed8.
Events & Experiential
Before
The Events & Experiential shelf exactly as your carve places it — every type, with live usage and a real stored piece where one exists. Of its 3 types, 2 are new (they arrive with the migration) and the rest carry 61 live pieces between them. Confirmed = these types sit on this shelf, as shown; if one belongs elsewhere, mark Amended and name the move in the note.
Event Film: 61 in use
Keynote: new — no stored value yet
Booth / Installation Piece: new — no stored value yet
e.g. (event_coverage · linkedin_ad): “Our first-ever OpenAI Partner Summit” — 7a758ae5
★: Confirm Events & Experiential as carved — membership as listed.
if you approve: Nothing is stored by a mark — $0, reversible. A Confirmed shelf becomes registry re-pointing in the pipeline lane when proposal 1 closes.
Now
Every label your carve puts on this shelf, and how many pieces carry it today. If one belongs somewhere else, mark Amended and name the move.
Event Film: 61 pieces
Keynote: new — arrives with the migration
Booth / Installation Piece: new — arrives with the migration
For example: “Our first-ever OpenAI Partner Summit” — 7a758ae5
1 of the 3 labels is in use today, over 61 pieces; the other 2 arrive with the migration.
★: Confirm Events & Experiential — these labels belong on this shelf.
if you approve: Nothing is stored by a mark; a confirmed shelf becomes registry re-pointing in the build lane.
folded · Some labels here have a card of their own later in the walk, which may rename or retire them: Event Film (N-06). Confirming the shelf confirms where a label sits, not what it is called.
Retail & Commerce
Before
The Retail & Commerce shelf exactly as your carve places it — every type, with live usage and a real stored piece where one exists. Of its 1 types, 1 are new (they arrive with the migration) and the rest carry 0 live pieces between them. Confirmed = these types sit on this shelf, as shown; if one belongs elsewhere, mark Amended and name the move in the note.
Retail Display Piece: new — no stored value yet
★: Confirm Retail & Commerce as carved — membership as listed.
if you approve: Nothing is stored by a mark — $0, reversible. A Confirmed shelf becomes registry re-pointing in the pipeline lane when proposal 1 closes.
Now
This shelf has one label today, and card R-01 decides whether that label survives at all. Confirm the shelf itself; the membership follows from R-01 and R-06.
Retail Display Piece: new — arrives with the migration
Every label here is new; this shelf arrives with the migration.
★: Confirm Retail & Commerce — this label belongs on this shelf.
if you approve: Nothing is stored by a mark; a confirmed shelf becomes registry re-pointing in the build lane.
folded · Some labels here have a card of their own later in the walk, which may rename or retire them: Retail Display Piece (R-01). Confirming the shelf confirms where a label sits, not what it is called.
Interactive & Digital Experience
Before
The Interactive & Digital Experience shelf exactly as your carve places it — every type, with live usage and a real stored piece where one exists. ⚑ rows moved here from a different family and want your eye. Of its 3 types, 3 are new (they arrive with the migration) and the rest carry 0 live pieces between them. Confirmed = these types sit on this shelf, as shown; if one belongs elsewhere, mark Amended and name the move in the note.
Interactive Demo: new — no stored value yet
Product Configurator: new — no stored value yet
Interactive Experience: new — no stored value yet
★: Confirm Interactive & Digital Experience as carved — membership as listed.
if you approve: Nothing is stored by a mark — $0, reversible. A Confirmed shelf becomes registry re-pointing in the pipeline lane when proposal 1 closes.
Now
Every label your carve puts on this shelf, and how many pieces carry it today. If one belongs somewhere else, mark Amended and name the move.
Interactive Demo: new — arrives with the migration
Product Configurator: new — arrives with the migration
Interactive Experience: new — arrives with the migration
3 of these labels moved onto this shelf from a different family under your carve — those are the ones to look at hardest.
★: Confirm Interactive & Digital Experience — these labels belong on this shelf.
if you approve: Nothing is stored by a mark; a confirmed shelf becomes registry re-pointing in the build lane.
folded · A flagged label moved onto this shelf from a different family under your carve, so it is the one to look at hardest.
Still Open After Your Amendments
Before
The holding shelf. Offer is the one type you have not settled — your words: “what is an offer promotion — wouldn’t that be like a product campaign,” with the deal tagged as marketing intent. The dedicated Offer card (N-09) is where that pick lands; mark this card Confirmed if the holding shelf itself reads right as shown.
Offer / Promotion: new — no stored value yet
★: Confirm the holding shelf as shown; the Offer pick itself is card N-09.
if you approve: Nothing is stored by a mark — $0, reversible. A Confirmed shelf becomes registry re-pointing in the pipeline lane when proposal 1 closes.
Now
The holding shelf, for the one label you have not settled. The pick itself is card N-09, so this card only asks whether the holding shelf reads right.
Offer / Promotion: new — arrives with the migration
1 of these labels moved onto this shelf from a different family under your carve — those are the ones to look at hardest.
★: Confirm the holding shelf as shown; the pick itself is card N-09.
if you approve: Nothing is stored by a mark; a confirmed shelf becomes registry re-pointing in the build lane.
folded · Your words on it: “what is an offer promotion — wouldn’t that be like a product campaign,” with the deal itself tagged as marketing intent.
folded · A flagged label moved onto this shelf from a different family under your carve, so it is the one to look at hardest.
folded · Some labels here have a card of their own later in the walk, which may rename or retire them: Offer / Promotion (N-09). Confirming the shelf confirms where a label sits, not what it is called.
3 · The naming picks
Launch Film
Before
Your worry started this whole section — banked verbatim: “having the word ‘film’ implies film (feels higher end) and or always video what if its a static isntagram post etc.” For Launch Film the kind survives the medium completely — a launch announced in an article and a launch announced in a film are the same kind of piece. And the data makes the case comic: exactly ONE piece in the whole library is stored product_launch today, and it is Nike's “NIKE MIND USER GUIDE” — so the rename costs one row and buys a medium-neutral kind forever. Under the rename, the media-format field WOULD compose “launch film” for display whenever it is a film (part of what this mark ratifies, not yet built).
Stored today: product_launch — exactly 1 piece
That one piece: “NIKE MIND USER GUIDE” — 716c5b3c · youtube
◮: The entire launch population — one Nike video, stored product_launch. Renaming this type is the cheapest decision on the console.
★: Rename to bare Launch; the media-format field composes “launch film” for display.
if you approve: Renames execute only after your picks, as one pass, in the pipeline lane. One row migrates.
Now
A launch announced in an article and a launch announced in a film are the same kind of piece, so the lean is the bare noun.
Stored today: exactly 1 piece carries it
That one piece: “NIKE MIND USER GUIDE” — 716c5b3c, on YouTube
Under the rename: the media-format field composes “launch film” for display
Cost of the rename: one row
This is the cheapest decision on the console.
◮: The entire launch population is one Nike video.
★: (a) Rename to bare Launch.
or: (b) keep Launch Film
if you approve: One row migrates, in the build lane, as part of one renaming pass.
folded · Your worry started this whole section, banked word for word: “having the word ‘film’ implies film (feels higher end) and or always video what if its a static isntagram post etc.”
folded · The display composition — Launch plus the media format reading “launch film” — is part of what this mark ratifies. It is not built yet.
Manifesto / Anthem Film
Before
A manifesto delivered as a static poster is still a manifesto — the kind survives with the medium hidden, so the bare noun wins. This one also carries your no-slashes rule (“we should never have like slashes or things like that in our data ontology”): Manifesto becomes the saved word, and anthem / anthem film stay as aliases — type either and it saves as Manifesto. No piece is stored under it yet, so the pick costs nothing.
Stored today: none — the type is new
If you keep the Film suffix: the stored value becomes manifesto_film (the resolve table states the contingency)
★: Rename to Manifesto (anthem = alias).
if you approve: Zero rows move; the value lands with the migration under whichever name you pick.
Now
A manifesto delivered as a static poster is still a manifesto, so the bare noun wins here too. Nothing is stored under it, so the pick costs nothing.
Stored today: none — the label is new
Under the rename: Manifesto is saved; anthem and anthem film become aliases
If you keep the suffix: the saved word becomes manifesto film instead
Type the old word and it saves as the new one, so no wording is lost.
★: (a) Rename to Manifesto, with anthem as an alias.
or: (b) keep the Film suffix
if you approve: No rows move; the label lands with the migration under whichever name you pick.
folded · This one also carries your no-slashes rule: “we should never have like slashes or things like that in our data ontology”.
Campaign Film — does it survive at all?
Before
You asked where Brand Film, Campaign Film and Brand Story genuinely differ — and the honest answer is that Campaign Film may not survive as its own kind. A film made FOR a campaign is a brand film, product film, or launch piece PLUS a campaign membership — the membership is already a stored fact, so “campaign film” restates the container on the piece, which is the exact defect this whole reconciliation removes. Nothing is stored under it today, so a merge is free.
Stored today: none — the type is new
What already carries campaign-ness: campaign membership + the piece's role in it (hero · supporting · validation)
★: Merge it away — a campaign's film is its real kind + a membership.
if you approve: Under merge: the type never enters the registry. Under keep: its definition must say what it names that membership doesn't.
Now
A film made for a campaign is already a brand film or a product film plus a campaign membership. The lean is that Campaign Film does not survive as its own kind.
Stored today: none — the label is new
What already carries it: the campaign membership, plus the piece's role in it
Cost of merging: none — nothing is stored
Naming it would restate the container on the piece, which is the defect this whole round removes.
★: (a) Merge it away.
or: (b) keep Campaign Film
if you approve: Under (b) its definition has to say what it names that the membership does not.
folded · You asked where Brand Film, Campaign Film and Brand Story genuinely differ, and this is the honest answer to the middle one. Naming it would mean “campaign film” restates the container on the piece.
folded · A piece's role in its campaign is already stored — hero, supporting, validation.
Brand Film
Before
Brand Film passes the keep test: its static sibling is already its own type — a Key Visual or a Brand Story post — so nothing static would ever be typed Brand Film, and the word is honest. One thing the real rows add: today's 136 brand_films include TikTok-native Sora shorts (the Sora Matinée pieces), which tells you the classifier reads “brand film” generously — worth remembering when definitions are written, not a reason to rename.
Stored today: brand_film — 136 pieces
e.g. (brand_film · tiktok): “🥹 We’re not crying, you’re crying. Today’s Sora Matinee: Thank you, mom” — 58e75c32
e.g. (brand_film · tiktok): ““Deflated” — companion piece to “Air Head”; shy kids × Sora” — d3362525
◮: A third of a minute of TikTok Sora footage and a cinematic anthem currently share one type — the definition, not the name, is where that tightens.
★: Keep Brand Film — inherently filmic.
if you approve: No change under keep; the definition pass later decides how tight “film” reads.
Now
Nothing static would ever be typed Brand Film — its static sibling is already its own label. The word is honest, so the lean is keep.
Stored today: brand film — 136 pieces
Its static sibling: a Key Visual or a Brand Story post
For example: “🥹 We’re not crying, you’re crying. Today’s Sora Matinee: Thank you, mom” — 58e75c32
What needs tightening here is the definition, not the name.
◮: A TikTok short and a cinematic anthem currently share this one label.
★: (a) Keep Brand Film — inherently filmic.
or: (b) rename it
if you approve: Nothing changes under (a); how tight “film” reads is settled when definitions are written.
folded · Today's 136 brand films include TikTok-native Sora shorts, which tells you the classifier reads “brand film” generously — worth remembering when definitions are written, not a reason to rename.
folded · Another real one: ““Deflated” — companion piece to “Air Head”; shy kids × Sora” — d3362525.
Product Film
Before
A product piece that arrives static is a Product Shot or a Feature Overview — both already on the list — so Product Film keeps its word honestly, same as Brand Film. Nothing is stored under it yet.
Stored today: none — the type is new
The static siblings: Product Shot (10 live, e.g. “Choice AI phone screens…” — 7b189f27) · Feature Overview (new)
★: Keep Product Film — inherently filmic.
if you approve: No rows move either way.
Now
A product piece that arrives static is a Product Shot or a Feature Overview, both already on the list. So Product Film keeps its word honestly, the same as Brand Film.
Stored today: none — the label is new
Its static siblings: Product Shot (10 pieces) · Feature Overview (new)
For example: “Choice AI phone screens in Furusato Choice app” — 7b189f27
Same test as Brand Film, same answer.
★: (a) Keep Product Film — inherently filmic.
or: (b) rename it
if you approve: No rows move either way.
Event Film
Before
An event captured static is an Event Photo — its own type — so Event Film keeps its word. One echo worth holding beside it: Recap/Highlights — a candidate from your old lists, card X-03 in a later section — overlaps this kind and could become its works-in-any-medium rename later; the two cards ride together.
Stored today: event_coverage — 61 pieces
e.g. (event_coverage · linkedin_ad): “Our first-ever OpenAI Partner Summit” — 7a758ae5
★: Keep Event Film.
if you approve: No change under keep; X-03's Recap pick can revisit the name in one pass if you want.
Now
An event captured static is an Event Photo, which is its own label, so Event Film keeps its word. One later card could still rename it.
Stored today: event coverage — 61 pieces
Its static sibling: Event Photo — already its own label
The echo: Recap / Highlights, card X-03, overlaps this kind
For example: “Our first-ever OpenAI Partner Summit” — 7a758ae5
The two cards ride together — X-03's pick can revisit this name in one pass.
★: (a) Keep Event Film.
or: (b) rename it
if you approve: Nothing changes under (a).
folded · Recap / Highlights is a candidate from your old lists. If you take it there, it could become the works-in-any-medium rename for this kind later.
Product Launch — compose or store
Before
You want a Product Launch on Product & Feature, distinct from Product Film. With the bare-Launch rename (N-01) there is ONE medium-neutral Launch kind, and WHAT launched — a product, a feature, the brand — is already carried by the product and feature links, exactly the proposal-9 pattern. “Product Launch” would then compose for display (Launch + its product link) instead of being a second stored kind. The one stored piece behind launch today is already product-flavored (the Nike Mind piece), so nothing is lost either way. If you want it stored regardless, mark (b) and it joins.
Today's launch storage: product_launch — the 1 Nike Mind piece (N-01)
What composes it: Launch (type) + the product link + the media format
★: (a) One Launch kind; product-ness composes from the links.
if you approve: Under (b) the shelf question (Campaign vs Product & Feature) rides with your pick — name it in the note.
Now
You asked for a Product Launch on Product & Feature, separate from Product Film. The lean is that it composes from what we already store rather than being a second label.
Launch storage today: the one Nike Mind piece — see N-01
What composes it: Launch, plus the product link, plus the media format
What (b) adds: a second stored label alongside Launch
What launched — a product, a feature, the brand — is already carried by the product and feature links.
★: (a) One Launch label; product-ness composes from the links.
or: (b) store Product Launch as its own label
if you approve: Under (b) the shelf question rides with your pick — name it in the note.
folded · This is the same pattern as proposal 9: coming from a product is a link, not a kind. Under (a) “Product Launch” composes for display — Launch plus its product link — instead of being a second stored kind.
folded · The one stored piece behind launch today is already product-flavoured, so nothing is lost either way.
Commercial vs Brand Film
Before
No — it is NOT the channel, and the two real pieces below make the boundary readable in one glance. A Commercial is a kind of piece defined by its editorial contract: short, offer- or product-forward, made to run as an advertisement — and it stays a commercial posted organically on YouTube. A Brand Film is long-form brand storytelling. Where either ran is a separate fact on the distribution axes.
In use: commercial 55 · brand_film 136
★: (a) Keep Commercial with a confusable-with note against Brand Film (length · offer-forwardness · made-to-run-as-an-ad — never where it ran).
if you approve: The confusable-with note lands in the registry when definitions are written; 55 pieces keep their type under (a).
folded · A COMMERCIAL — “Get Ada Her Way. Can anyone stop Ada Hegerberg?” — 0a4d8928 · short, product-forward, built to run as an ad — on YouTube, and still a commercial.
folded · A BRAND FILM — “Thank you, mom” — Sora Matinée — 58e75c32 · brand storytelling with no offer in sight — the editorial contract, not the channel, separates them.
Now
No — the boundary is not the channel. A Commercial is short and made to run as an advertisement; a Brand Film is long-form brand storytelling.
In use: commercial 55 pieces · brand film 136 pieces
What separates them: length, offer-forwardness, made-to-run-as-an-ad
What does not: where it ran — that is a separate fact
A real commercial: “Get Ada Her Way. Can anyone stop Ada Hegerberg?” — 0a4d8928
A real brand film: “Thank you, mom” — the Sora Matinée piece — 58e75c32
A commercial posted organically is still a commercial.
★: (a) Keep Commercial, with a confusable-with note against Brand Film.
or: (b) drop it, and let films plus the paid-placement facts carry it
if you approve: Fifty-five pieces keep their label under (a), and the note lands when definitions are written.
folded · The commercial is short, product-forward and built to run as an ad — on YouTube, and still a commercial. The brand film is brand storytelling with no offer in sight. The editorial contract, not the channel, separates them.
Offer — the one still-open label
Before
Your words: “what is an offer promotion — wouldn’t that be like a product campaign,” with the deal tagged as marketing intent. Remove means: the piece keeps its real kind (promotional ad, key visual, social post…) and the offer becomes an intent tag on it. Keep means: Offer stays a piece kind for pieces whose whole identity is the deal. Nothing is stored under it, so removing costs zero rows — and marking (a) empties the Still-Open shelf entirely.
In use today: no stored value — removing costs nothing
Where the meaning goes: marketing intent (promotion · deal) on the piece's real kind
★: (a) Remove — offer-ness becomes marketing intent (your lean).
if you approve: Under (a) the type never enters the registry and the resolve-table row drops (N-10 already rides this). Under (b) it shelves where you say.
Now
Your lean is to remove it: the piece keeps its real kind and the deal becomes an intent tag. Nothing is stored under it, so removing costs no rows.
In use today: no stored value
Where the meaning goes: marketing intent — promotion, deal — on the real kind
If you keep it: Offer stays a kind for pieces whose whole identity is the deal
Marking (a) empties the Still-Open shelf entirely.
★: (a) Remove — offer-ness becomes marketing intent.
or: (b) keep Offer as a label
if you approve: Under (a) the label never enters the registry and its resolve-table row drops.
folded · Your words: “what is an offer promotion — wouldn’t that be like a product campaign,” with the deal tagged as marketing intent.
folded · Card N-10's resolve table already rides this pick.
The slash names, as one batch
Before
“We should never have slashes… in our data ontology taxonomy.” Every ratified slash pair becomes one clean stored value plus an alias that still resolves on write: Manifesto (anthem) · Commercial (spot, TVC) · Interview (Q&A) · Offer (promotion) · Course (lesson) · Brand Story (origin story) · Installation (booth piece). Nothing is lost — type the old word and it saves as the new one. The Offer row RIDES card N-09: if you remove Offer there, its row simply drops from this table.
The pairs: 7 slash pairs → 7 clean values + aliases
Live impact: only interview (114 pieces) and commercial (55) carry rows today; the rest are new values
★: Confirm the whole resolve table as one batch (the Offer row rides N-09); strike any row in the note.
if you approve: The values land with the migration; aliases keep every historical word findable. $0 today.
Now
Every slash pair becomes one saved word plus an alias that still resolves when typed. Nothing is lost — type the old word and it saves as the new one.
The pairs: 7 pairs → 7 clean words plus aliases
The list: Manifesto · Commercial · Interview · Offer · Course · Brand Story · Installation
Live impact: only interview (114 pieces) and commercial (55) carry rows today
The Offer row: rides card N-09 — if you remove Offer there, this row drops
Strike any row you disagree with in the note; the rest stand as one batch.
★: Confirm the whole resolve table as one batch.
if you approve: The words land with the migration, and aliases keep every historical spelling findable.
folded · Your rule, word for word: “we should never have slashes… in our data ontology taxonomy.”
folded · The pairs in full: Manifesto (anthem) · Commercial (spot, TVC) · Interview (Q&A) · Offer (promotion) · Course (lesson) · Brand Story (origin story) · Installation (booth piece).
Promotional Ad and Sponsored Content
Before
Rule 3 of the evidence page (paid-ness is a distribution state) presses on two surviving types, and honesty says show you. Promotional Ad (1,106 pieces — the second-biggest type in use) and Key Visual can be the same image in two roles; the boundary is made-to-run-as-an-ad versus made-as-hero-art, and that becomes the confusable-with note. Sponsored Content names a deal as much as a kind — with the medium hidden it could be an article, video, or post. Keeping both is defensible (the designed ad unit and advertorial are recognizable kinds); the alternative gives each piece its real type and records the paid-or-partner fact separately.
advertisement_image: 1,106 pieces
e.g. (advertisement_image · linkedin_ad): “L’IA accompagne le flux de travail de votre équipe…” — 4df5bd1d
sponsored content: new — nothing stored
★: Keep both, with confusable-with notes when definitions are written.
if you approve: Striking either sends its meaning to the ownership/placement axes — say which in the note. Keeping costs nothing now; the notes land with definitions.
Now
Two surviving labels name a paid arrangement as much as a kind of piece. Keeping both is defensible, and the lean is to keep them with a note about what they are confused with.
Promotional ad: 1,106 pieces — the second-biggest label in use
Where it blurs: the same image can be hero art or an ad unit
Sponsored content: new — nothing stored
Where it blurs: with the medium hidden it could be an article, video or post
For example: “L’IA accompagne le flux de travail de votre équipe…” — 4df5bd1d
The boundary is made-to-run-as-an-ad versus made-as-hero-art.
★: Keep both, with confusable-with notes when definitions are written.
if you approve: Striking either sends its meaning to the ownership and placement fields — say which in the note.
folded · Rule 3 of the evidence page says paid-ness is a distribution state, not a kind, and it presses on both of these. Honesty says show you rather than quietly keep them.
folded · The defence for keeping: the designed ad unit and the advertorial are both recognizable kinds of piece in their own right.
Customer Story or User Story
Before
Both are ratified types today, both new (nothing stored under either). The honest line, if you want one: a paying customer's story versus any user's story — and the live neighbors show the family resemblance: testimonial (16 pieces, real quotes) and case study (149, e.g. Singular Bank / Boston Children's) already carry the proof weight. If the customer/user line will not hold in practice, merge with an alias and let ownership + audience carry the nuance.
Both on the list: Customer Story · User Story — both new
The live neighbors: testimonial 16 · case_study 149 · product_review 10
★: Your pick — keep both with the customer/user line, or merge with an alias.
if you approve: A merge lands with the migration; keeping both adds the line to their definitions. Zero rows either way.
Now
Both are on the list and nothing is stored under either. The line, if you want one, is a paying customer's story versus any user's story.
Both on the list: Customer Story · User Story — both new
The live neighbours: testimonial 16 · case study 149 · review 10
If the line will not hold: merge them, with an alias
For example: “Singular Bank built Singularity — an internal assistant” — ea8769b9
Ownership and audience can carry the nuance if the label cannot.
★: Your pick — keep both with the customer-or-user line, or merge with an alias.
if you approve: A merge lands with the migration; keeping both adds the line to their definitions.
folded · The live neighbours already carry the proof weight: testimonials are real quotes, and case studies run to 149 pieces including Boston Children's Hospital.
4 · The leftover values
Infographic
Before
62 pieces are stored as infographic today — not on the ratified list, and the real rows below show what the value actually holds. The real rows are research-paper chart images — closer to a figure than to designed marketing infographics. The open question: joins the list as its own type, or maps to Key Visual. Your call — the options are genuinely level; no recommendation is minted where the surface offers none.
Stored today: infographic — 62 pieces
e.g. (infographic · landing_page): “Four line graphs: lifetime-in-tournament across opponents” — 7dc28cf8
★: Your call: joins the list as its own type, or maps to Key Visual.
if you approve: Either way the migration executes in the pipeline lane with the family ruling; nothing moves today.
Now
62 pieces are stored as infographic today, and it is not on your ratified list. The pick is whether it joins the list as its own label, or maps to Key Visual.
Stored today: 62 pieces
For example: “Four line graphs: lifetime-in-tournament across opponent…” — 7dc28cf8
The real rows are research-paper charts rather than designed marketing art — read them before you pick.
★: Make it a label of its own, or map it away — the options are level.
if you approve: Either way the change executes with the family migration; nothing moves today.
folded · The real rows are research-paper chart images — closer to a figure than to the designed marketing infographics the word usually brings to mind.
Podcast Video
Before
56 pieces are stored as podcast_video today — not on the ratified list, and the real rows below show what the value actually holds. The real rows are podcast-style conversations posted as TikToks — the kind is real, the name is the question. The open question: podcast episode is registered but unratified. Your call — the options are genuinely level; no recommendation is minted where the surface offers none.
Stored today: podcast_video — 56 pieces
e.g. (podcast_video · tiktok): “Ben Goodger & Darin Fisher on the browsers they shaped” — 1d41268a
★: Your call: podcast episode is registered but unratified.
if you approve: Either way the migration executes in the pipeline lane with the family ruling; nothing moves today.
Now
56 pieces are stored as podcast video today, and it is not on your ratified list. The pick is whether it becomes a label of its own — podcast episode is registered but was never ratified.
Stored today: 56 pieces
For example: “Ben Goodger & Darin Fisher on the browsers they shaped” — 1d41268a
The kind is real; the name is the question.
★: Make it a label of its own, or map it away — the options are level.
if you approve: Either way the change executes with the family migration; nothing moves today.
folded · The real rows are podcast-style conversations posted as TikToks.
Behind The Scenes
Before
25 pieces are stored as behind_the_scenes today — not on the ratified list, and the real rows below show what the value actually holds. The shy-kids Sora BTS pieces are exactly what the label promises — the cleanest of the open six. The open question: a real editorial kind, or a treatment. Your call — the options are genuinely level; no recommendation is minted where the surface offers none.
Stored today: behind_the_scenes — 25 pieces
e.g. (behind_the_scenes · tiktok): “Behind the scenes with shy kids: merging Sora with live-action” — ebffd41b
★: Your call: a real editorial kind, or a treatment.
if you approve: Either way the migration executes in the pipeline lane with the family ruling; nothing moves today.
Now
25 pieces are stored as behind the scenes today, and it is not on your ratified list. The pick is whether it is a real editorial kind, or only a treatment.
Stored today: 25 pieces
For example: “Behind the scenes with shy kids: merging Sora with live…” — ebffd41b
The pieces here are exactly what the label promises, which makes this the easiest of the six to judge.
★: Make it a label of its own, or map it away — the options are level.
if you approve: Either way the change executes with the family migration; nothing moves today.
folded · The shy-kids Sora behind-the-scenes pieces are exactly what the label promises, which is why this is the cleanest of the six.
System Card
Before
14 pieces are stored as system_card today — not on the ratified list, and the real rows below show what the value actually holds. Real OpenAI system-card addenda — a genuinely technical piece kind with no home until Documentation & Technical. The open question: a real technical piece kind; candidate for Documentation & Technical. Your call — the options are genuinely level; no recommendation is minted where the surface offers none.
Stored today: system_card — 14 pieces
e.g. (system_card · landing_page): “GPT-5.1 Instant and GPT-5.1 Thinking System Card Addendum” — da675eb3
★: Your call: a real technical piece kind; candidate for Documentation & Technical.
if you approve: Either way the migration executes in the pipeline lane with the family ruling; nothing moves today.
Now
14 pieces are stored as system card today, and it is not on your ratified list. The pick is whether it joins Documentation & Technical as a label of its own.
Stored today: 14 pieces
For example: “GPT-5.1 Instant and GPT-5.1 Thinking System Card Addendu…” — da675eb3
A genuinely technical kind of piece with no home until Documentation & Technical.
★: Make it a label of its own, or map it away — the options are level.
if you approve: Either way the change executes with the family migration; nothing moves today.
folded · Real OpenAI system-card addenda — a technical kind of piece that had nowhere to sit until the Documentation & Technical shelf existed.
Vlog
Before
9 pieces are stored as vlog today — not on the ratified list, and the real rows below show what the value actually holds. Nike Playlist episodes — creator-format series content; kind vs treatment is a genuine coin-flip. The open question: a creator-native kind, or a treatment. Your call — the options are genuinely level; no recommendation is minted where the surface offers none.
Stored today: vlog — 9 pieces
e.g. (vlog · youtube): “Chipping with golfer Lucy Li! (Nike Playlist)” — dcf8d38f
★: Your call: a creator-native kind, or a treatment.
if you approve: Either way the migration executes in the pipeline lane with the family ruling; nothing moves today.
Now
9 pieces are stored as vlog today, and it is not on your ratified list. The pick is whether it is a creator-native kind, or only a treatment.
Stored today: 9 pieces
For example: “Chipping with golfer Lucy Li! (Nike Playlist)” — dcf8d38f
Kind or treatment is a real coin-flip here.
★: Make it a label of its own, or map it away — the options are level.
if you approve: Either way the change executes with the family migration; nothing moves today.
folded · Nike Playlist episodes — creator-format series content.
Event Photo
Before
5 pieces are stored as event_photo today — not on the ratified list, and the real rows below show what the value actually holds. One of its two example rows is a Codex AD stored as event_photo — this value is catching strays, which argues for deciding it rather than leaving it open. The open question: its own kind, or existing imagery types. Your call — the options are genuinely level; no recommendation is minted where the surface offers none.
Stored today: event_photo — 5 pieces
e.g. (event_photo · vendor_portfolio): “DevDay 2025 — venue atrium & ASCII flower mural” — 084c3ecc
e.g. (event_photo · linkedin_ad): “⚠ “Take everyday work from idea to finished output with Codex” — a Codex ad stored as event_photo” — 1784c037
★: Your call: its own kind, or existing imagery types.
if you approve: Either way the migration executes in the pipeline lane with the family ruling; nothing moves today.
Now
5 pieces are stored as event photo today, and it is not on your ratified list. The pick is whether it is its own kind, or belongs with the imagery labels we already have.
Stored today: 5 pieces
For example: “DevDay 2025 — venue atrium & ASCII flower mural” — 084c3ecc
For example: “⚠ “Take everyday work from idea to finished output with…” — 1784c037
This label is catching strays, so whichever way you go, deciding it beats leaving it open.
★: Make it a label of its own, or map it away — the options are level.
if you approve: Either way the change executes with the family migration; nothing moves today.
folded · One of its two example pieces is a Codex ad stored as an event photo. A label catching strays is a label worth settling.
The nine settled leftovers
Before
The other nine leftovers have definite proposed fates. The two giants split (social_post 2,518 and social_clip 943. “Social” moves to the where-it-ran fields, and each piece gets its real type — the examples show why: a GPT-5.5 endorsement post is really a testimonial-shaped piece, and a Sora clip is really a slice of brand filmmaking). blog_post refines to real editorial types · feature_launch and livestream move to container/distribution facts · case_study_video and animated_explainer fold (the format field already says video) · documentary becomes genre riding a film type · brand_asset splits to the identity types (its real rows are DevDay patches and motion loops — identity work mis-filed). Confirm as one batch; strike any in the note.
social_post: 2,518 — Split
social_clip: 943 — Split
blog_post: 290 — Refine
feature_launch: 188 — Move
case_study_video: 68 — Fold into case study
documentary: 19 — Becomes creative genre/treatment
brand_asset: 13 — Split
livestream: 10 — Move
animated_explainer: 1 — Fold into explainer
e.g. (social_post · linkedin_ad): ““GPT-5.5 is here! Have been daily driving this model…”” — c334aa5a
e.g. (brand_asset · vendor_portfolio): “DevDay 2025 — embroidered character patches” — 8809b931
e.g. (documentary · youtube): “The Final Year | Ep. 4 | Sue Bird & Diana Taurasi (Nike x TOGETHXR)” — 3b7e6e33
★: Confirm the nine proposed fates as written on the surface.
if you approve: Migration facts, identical under every carve; they execute with the family ruling. 4,050 pieces re-home across the nine (the sum is computed from the same table the surface renders).
Now
The other nine leftovers already have a proposed fate each, and none of them depends on which carve wins. Confirm them as one batch, and strike any you disagree with in the note.
Social Post: 2,518 pieces — split
Social Clip: 943 pieces — split
Blog Post: 290 pieces — refine
Feature Launch: 188 pieces — move
Case Study Video: 68 pieces — fold into case study
Documentary: 19 pieces — becomes creative genre or treatment
Brand Asset: 13 pieces — split
Livestream: 10 pieces — move
Animated Explainer: 1 piece — fold into explainer
For example: “GPT-5.5 is here! Have been daily driving this model…” — c334aa5a
And one more: “DevDay 2025 — embroidered character patches” — 8809b931
4,050 pieces re-home across the nine, and the sum is computed from the same table the surface renders.
★: Confirm the nine proposed fates as written.
if you approve: They execute with the family migration, identically under every carve.
folded · The two giants split: “social” moves to the where-it-ran fields and each piece gets its real label. The examples show why — a GPT-5.5 endorsement post is really a testimonial-shaped piece, and a Sora clip is really a slice of brand filmmaking.
folded · The rest: blog post refines to real editorial labels · feature launch and livestream move to container and distribution facts · case study video and animated explainer fold away, because the format field already says video · documentary becomes a genre riding a film label · brand asset splits to the identity labels.
folded · Brand asset's real rows are DevDay embroidered patches and motion loops — identity work mis-filed, e.g. “DevDay 2025 — embroidered character patches” — 8809b931.
folded · And one documentary, for the record: “The Final Year | Ep. 4 | Sue Bird & Diana Taurasi (Nike x TOGETHXR)” — 3b7e6e33.
5 · Retail & Commerce
The display pick
Before
The ratified list has Retail Display Piece as the generic in-store kind; the table names Endcap, Point-of-Purchase, Window and Aisle. Keep the specifics and retire the generic (a), or keep the generic and carry location as a detail (b)? Shown honestly because it cuts against (a): the retail-media industry's own standard carves the store into placement ZONES (exterior · entrance · checkout · in-aisle · connected) and keeps the piece kinds smaller — shelf talker, hanging sign, floor graphic, screen. In that carve an endcap is WHERE a piece sits, and the standard's shape is closer to (b), with the store zone recorded among the where-it-ran fields. The kind-test logic still leans (a): a merchandiser recognizes an endcap the way an editor recognizes a press release. Nothing is stored under any of these yet — the pick is pure vocabulary.
Ratified today: Retail Display Piece — new, no stored value
The standard's carve: 5 placement zones ⊥ named piece formats (shelf talker · hanging sign · floor graphic · screen…)
Live retail storage: none yet — the shelf fills when retail collection exists
★: (a) Keep the specifics — the kind-test read.
if you approve: Under (a): four types join, the generic retires. Under (b): the store-zone words join the where-it-ran build instead. $0 either way; the standard's format list stays available as future seeds under both.
Now
Keep the four specific display kinds and retire the generic one, or keep the generic and record the location separately? The lean is the specifics.
Ratified today: Retail Display Piece — new, nothing stored
The four specifics: Endcap · Point-of-Purchase · Window · Aisle
The industry standard: five store zones, with smaller piece kinds
Live retail storage: none yet — the shelf fills when retail collection exists
This is the one pick where the industry standard leans against the recommendation, and shelf card S-10 rides it.
◮: Under the standard's carve an endcap is WHERE a piece sits, not what it is.
★: (a) Keep the specifics — the kind-test read.
or: (b) keep the generic, and record the store zone among the where-it-ran fields
if you approve: Under (a) four labels join and the generic retires; under (b) the zone words join the where-it-ran build instead.
folded · Shown honestly because it cuts against the lean: the retail-media industry's own standard carves the store into placement zones — exterior, entrance, checkout, in-aisle, connected — and keeps the piece kinds smaller: shelf talker, hanging sign, floor graphic, screen.
folded · The kind test still leans the other way: a merchandiser recognizes an endcap the way an editor recognizes a press release.
folded · Nothing is stored under any of these yet, so the pick is pure vocabulary — and the standard's format list stays available as future seeds under both options.
folded · Shelf card S-10 shows Retail & Commerce holding one label, Retail Display Piece. That is the label option (a) retires, so mark this card and S-10 consistently.
Aisle Display or Endcap
Before
Both are in-aisle display units; Aisle Display could fold into Endcap or stand alone. The surface deliberately offers no recommendation — a merchandiser might keep them apart (an endcap is premium placement), a taxonomist might fold them. Nothing is stored under either.
Status: both kind ✓ on the table · both new · overlap flagged
★: Your pick — the table deliberately leaves this open.
if you approve: One row's fate on the expansion list; free either way.
Now
Both are in-aisle display units, and Aisle Display could fold into Endcap or stand on its own. Nothing is stored under either, so this is free.
Status: both pass the kind test · both new · overlap flagged
A merchandiser might keep them apart, since an endcap is premium placement; a taxonomist might fold them.
★: Fold Aisle Display into Endcap, or keep both — the table deliberately leaves this open.
if you approve: One row's fate on the expansion list, free either way.
Interactive Kiosk — which shelf
Before
Mall kiosks and information booths pass the kind test; the open question is only which shelf browses it — Retail & Commerce (a physical commerce surface) or Interactive & Digital Experience (an interactive one). Shelves are browse choices, not schema, so this is a taste call — the surface offers no lean.
Status: kind ✓ · new · shelf question flagged
★: Your pick — the table deliberately leaves the shelf open.
if you approve: A registry pointer; movable later at zero cost.
Now
Mall kiosks and information booths pass the kind test. The only open question is which shelf browses them.
Status: passes the kind test · new
The two shelves: Retail & Commerce, or Interactive & Digital Experience
Shelves are a browse choice, not schema, so this one is taste.
★: Retail & Commerce, or Interactive & Digital Experience — the table leaves the shelf open.
if you approve: A registry pointer, movable later at no cost.
Product Detail Page
Before
Your seed listed PDPs; the pushback is built from your own ruling. A PDP is a page SURFACE — the atomic-unit ruling says the pieces ON it keep their real types, and the biggest commerce platform's own vocabulary agrees: its enhanced content (A+ Content / Brand Story modules) is named content composed ONTO the product page. So “product detail page” becomes one of the where-it-ran answers, and a hero image captured from a PDP is a key visual that ran there. Wanting whole-page captures as a KIND would amend the atomic-unit ruling — a deliberate mark, not a drive-by.
Your ruled ground: content item = the smallest atomic unit, fully tagged on its own
The industry read: A+ Content = pieces composed onto the PDP; the page is the surface
★: (a) A product detail page is one of the where-it-ran answers; the pieces on it keep their kinds.
if you approve: Under (b) say what a whole-page capture's type would be — it reopens ruled ground, and every composed-page kind since (marketplace listings, app-store pages) follows the same amendment.
Now
You seeded product detail pages as a kind, and the pushback is built from your own ruling: a page is a surface, and the pieces on it keep their real kinds.
Your ruled ground: a content item is the smallest unit, tagged on its own
The industry read: enhanced content is composed onto the page; the page is the surface
What it becomes: one of the where-it-ran answers
A hero image from a page: still a key visual, that ran there
Option (b) reopens ruled ground, so it is a deliberate mark rather than a drive-by.
★: (a) A product detail page is one of the where-it-ran answers.
or: (b) amend the atomic-unit ruling so a whole-page capture is a kind
if you approve: Under (b), say what a whole-page capture's kind would be — every composed-page kind after it follows the same amendment.
folded · Under (a), “product detail page” becomes one of the where-it-ran answers rather than a kind of piece.
folded · The biggest commerce platform's own vocabulary agrees with (a): its enhanced content — the A+ Content and Brand Story modules — is named content composed onto the product page, not the page itself.
folded · Marketplace listings and app-store pages are the same shape, so the amendment would not stop at this one.
Corporate Gift
Before
Branded gift items read as a Merchandise sibling — the same physical-goods kind with a different recipient. Fold unless you want them distinct (a gifting program might one day want its own view, but that is a lens, not a type).
Status: fold proposed on the table · nothing stored under either
★: Fold into Merchandise.
if you approve: One row's fate; free either way.
Now
Branded gift items read as a Merchandise sibling — the same physical goods with a different recipient. The lean is to fold them in.
Status: fold proposed · nothing stored under either
A gifting programme might one day want its own view, but that is a lens, not a kind.
★: (a) Fold into Merchandise.
or: (b) keep Corporate Gift distinct
if you approve: One row's fate, free either way.
The clean retail kinds
Before
The rows that pass the kind test cleanly, every one traced to a source of yours: Endcap Display · Point-of-Purchase Display · Window Display · Retail Signage · Packaging · Unboxing (named medium-neutral on purpose — the format field would compose “unboxing video”) · Catalog · Brochure · Poster · Merchandise · Branded Vehicle. All new — the library holds no retail pieces yet because retail collection has never run; the shelf fills when it does (the design-for-scale principle: capture-worthiness, not current fill). NOTE: the three specific displays (Endcap, Point-of-Purchase, Window) RIDE card R-01 — under its option (b) they fold into the generic instead of joining. Confirm as one batch; strike any in the note.
Status: 11 kind ✓ rows · all new (no stored values — retail collection unbuilt)
Sources: your seed lines · your 2026-07-08 touchpoints sheet · the guideline books' environment sections
★: Confirm the batch — they join Retail & Commerce on your mark (the three specific displays riding R-01).
if you approve: They enter the registry with the family migration; the starved shelf gets its vocabulary before its pipeline — deliberately.
Now
Eleven rows pass the kind test cleanly, every one traced to a source of yours. Confirm them as one batch and strike any you disagree with in the note.
Endcap Display: new · rides card R-01
Point-of-Purchase Display: new · rides card R-01
Window Display: new · rides card R-01
Retail Signage: new
Packaging: new
Unboxing: new · named medium-neutral on purpose
Catalog: new
Brochure: new
Poster: new
Merchandise: new
Branded Vehicle: new
All eleven are new, because retail collection has never run — the shelf gets its vocabulary before its pipeline, deliberately.
★: Confirm the batch — they join Retail & Commerce on your mark.
if you approve: They enter the registry with the family migration.
folded · The three specific displays ride card R-01: under its option (b) they fold into the generic instead of joining.
folded · Unboxing is named without the medium on purpose — the format field would compose “unboxing video” for display.
folded · Sources: your seed lines · your 2026-07-08 touchpoints sheet · the guideline books' environment sections.
folded · This is the design-for-scale principle in practice: capture-worthiness, not current fill.
The three not-kinds
Before
Three ideas that live elsewhere, each with its reason: In-Store Digital Signage folds into whichever display kind survives R-01 (a display whose media format is video needs no type of its own) · Shoppable Content is a placement feature (where it runs, not what it is) · a Sales/Promo Banner is ad format + marketing intent — the piece keeps its real kind. Confirm the redirections as one batch.
Status: 3 rows, each redirected to its real axis on the table
★: Confirm the redirections.
if you approve: Nothing enters the type list; each idea lands on the question it really answers (where it ran · what the marketing intent was) when those fields are built.
Now
Three ideas from the table are real, but each answers a different question than “what kind of piece is this?”. Confirm the redirections as one batch.
In-Store Digital Signage: folds into whichever display kind survives R-01
Shoppable Content: a placement feature — where it runs, not what it is
Sales or Promo Banner: ad format plus marketing intent; the piece keeps its kind
A display whose media format is video needs no kind of its own.
★: Confirm the redirections.
if you approve: Nothing enters the label list; each idea lands on the field that really answers it.
6 · Seeds beyond retail
Earnings Report
Before
The quarterly results piece — investor communication with a recognizable shape. Nothing is stored under it (financial-comms collection has never run); it would shelve under Corporate Communication beside Press Release (110 live) and Research Report.
Status: kind ✓ · new
The shelf neighbors: press_release 110 · white_paper 126 (as documentation)
★: Add it to Corporate Communication.
if you approve: Joins the list on your mark; fills when investor-comms collection exists.
Now
The quarterly results piece is investor communication with a recognizable shape. The lean is to add it to Corporate Communication.
Status: passes the kind test · new
Why nothing is stored: financial-comms collection has never run
Its shelf neighbours: Press Release (110 pieces) · Research Report (new)
The documentation neighbour: White Paper — 126 pieces, on its own shelf
It fills when investor-communications collection exists.
★: Add it to Corporate Communication.
if you approve: It joins the list on your mark.
Annual or Investor Report
Before
The designed annual report — a recognizable piece with real craft in it (your touchpoints sheet lists it too). The table names no shelf; Corporate Communication is the natural read, yours to amend. The slash resolves per your rule — say the single name in the note if you have a preference (Annual Report is the common one).
Status: kind ✓ · new · no shelf named on the table
★: Add it; Corporate Communication as the shelf unless you say otherwise.
if you approve: Joins on your mark; the alias carries the other name.
Now
The designed annual report is a recognizable piece with real craft in it, and your touchpoints sheet lists it too. The table names no shelf, so Corporate Communication is the natural read.
Status: passes the kind test · new
The shelf: not named on the table — Corporate Communication is the inference
The name: Annual Report is the common one; the other becomes an alias
Say the single name in the note if you have a preference.
★: Add it, with Corporate Communication as the shelf unless you say otherwise.
if you approve: It joins on your mark, and the alias carries the other name.
Recap or Highlights
Before
Event and launch recaps are recognizable — and this overlaps Event Film (61 live pieces, card N-06): a recap could even be Event Film's medium-neutral rename. Decide this with N-06 in view so the two don't split one kind.
Status: kind ✓ with a lean · overlap flagged
The neighbor: event_coverage — 61 pieces (N-06)
★: Add Recap as a kind; settle its relationship to Event Film with card N-06 in view.
if you approve: Joins on your mark; if it becomes Event Film's rename, that lands in the same naming pass.
Now
Event and launch recaps are recognizable kinds. This one overlaps Event Film, so decide it with card N-06 in view.
Status: passes the kind test, with a lean · overlap flagged
The neighbour: Event Film — 61 pieces, card N-06
The possibility: Recap could be Event Film's medium-neutral rename
Deciding the two together stops one kind from splitting in half.
★: Add Recap as a kind, and settle its relationship to Event Film with card N-06 in view.
if you approve: If it becomes Event Film's rename, that lands in the same naming pass.
The four folds and parks
Before
Trend Report folds into Research Report (an alias, not a second kind) · Listicle is a treatment of an article, not a kind · Reaction Video parks until the social-native question (P-03) settles · Corporate Film folds into Brand Film (the “vision and strategy” film is a brand film with a corporate subject — alias at most). Everything else on your old lists was a duplicate, a campaign kind, or a placement fact — deliberately not proposed, listed on the surface so nothing was curated away silently.
Status: 4 rows, each with its redirection
★: Confirm the folds and parks.
if you approve: Nothing enters the list; aliases land with the migration.
Now
Four more rows from your old lists are real ideas that do not become labels of their own. Confirm the redirections as one batch.
Trend Report: folds into Research Report, as an alias
Listicle: a treatment of an article, not a kind
Reaction Video: parks until the social-native question settles — card P-03
Corporate Film: folds into Brand Film — an alias at most
Everything else on your old lists was a duplicate, a campaign kind, or a placement fact.
★: Confirm the folds and parks.
if you approve: Nothing enters the list; the aliases land with the migration.
folded · The “vision and strategy” film is a brand film with a corporate subject, which is why Corporate Film folds rather than joining.
folded · The rows that were deliberately not proposed are still listed on the review surface, so nothing was curated away silently.
7 · Brand & Identity depth
The identity aliases
Before
The extraction over your named sources (your 133-row spreadsheet · the GitHub 2025 book · the OpenAI/Area-17 pages · the motion framework) proposes NO new piece type — the six ratified types hold. What it found instead is names that are the same kind under a different word, proposed as aliases (type the old word, it saves as the real one): Brand Book, Style Guide and Digital Guidelines → Brand Guidelines · Key Art → Key Visual (the GitHub book literally calls its nine hero illustrations “Key Art”).
The shelf's types: six ratified — Key Visual · Logo Animation · Product Shot · Template · Brand Guidelines · Icon System
Evidence in the corpus: the DevDay identity work currently mis-filed as brand_asset (“motion identity — loop 04” — 0f522a25) is exactly what these types will hold
★: Confirm the four aliases.
if you approve: Alias rows in the registry, pipeline lane; the brand_asset split (card L-07) re-homes its 13 pieces onto these types.
Now
The extraction over your own sources proposes no new kind of piece — the six ratified ones hold. What it found instead is four names for kinds we already have.
Brand Book, Style Guide, Digital Guidelines: all become Brand Guidelines
Key Art: becomes Key Visual
The shelf's labels: six ratified, from Key Visual to Icon System
Evidence in the library: “DevDay 2025 motion identity — loop 04” — 0f522a25
Type the old word and it saves as the real one, so nothing is lost.
★: Confirm the four aliases.
if you approve: Alias rows in the registry, in the build lane.
folded · The sources: your 133-row spreadsheet · the GitHub 2025 book · the OpenAI and Area-17 pages · the motion framework. The book literally calls its nine hero illustrations “Key Art”.
folded · The identity work currently mis-filed as brand asset is exactly what these labels will hold — card L-07's split re-homes those 13 pieces onto them.
folded · The six ratified labels in full: Key Visual · Logo Animation · Product Shot · Template · Brand Guidelines · Icon System.
Identity elements — the visual group
Before
What a piece can SHOW (never each a kind of piece — that split is the whole trick): logo marks and lockups, palettes, typography, photography and illustration styles, iconography, patterns, layout — plus the four the guideline books forced, flagged new: topic color themes (GitHub keeps blue for security, purple for AI), gradients (the design-token standard makes them a first-class type — real standards support), shape systems (GitHub's Git-Grid), and mascots (Mona, Copilot, Duckie — absent from your own spreadsheet, present in the wild).
Rows: 22 visual elements on the review page, each with its source cited
The new four: topic color themes · gradients · shape systems · mascots
★: Confirm the Visual group including the four new-flagged elements.
if you approve: Element vocabulary only; storage design (where element tags live, at what grain) is stated in the pipeline lane before anything is written.
Now
These are things a piece can SHOW, never kinds of piece — that split is the whole trick. Four of the twenty-two are new, forced by the guideline books.
Rows: 22 visual elements, each with its source cited
Topic colour themes: new — one book keeps blue for security, purple for AI
Gradients: new — the design-token standard makes them first-class
Shape systems: new — one book's grid is a named system
Mascots: new — absent from your spreadsheet, present in the wild
The long-standing rows are the familiar ones: logo marks, palettes, typography, photography, iconography, patterns, layout.
★: Confirm the visual group, including the four new ones.
if you approve: Element vocabulary only; where element tags live is settled in the build lane before anything is written.
Identity elements — verbal and strategic
Before
Verbal: voice, tone, key messages, taglines, calls to action, naming, terminology, mission/values statement — industry-standard, first-class beside visual. Strategic is deliberately flagged as a DIFFERENT GRAIN: purpose, values, promise, architecture, value proposition, differentiation, competition view, audience definitions classify the BRAND, not the piece — kept because your spreadsheet files them here, honest about what they describe. The audience rows connect to your audience-as-real-object ruling.
The grain flag: strategic rows read as the brand's strategy record, not piece tags
★: Confirm both groups with the strategic-grain flag as written.
if you approve: The audience list itself stays owed by the Concepts chapter — one vocabulary for both grains.
Now
The verbal group sits first-class beside the visual one. The strategic group is flagged as a different grain: it describes the brand, not the piece.
Verbal: voice, tone, key messages, taglines, calls to action, naming
Strategic: purpose, values, promise, architecture, value proposition
The grain flag: strategic rows read as the brand's strategy record
Audience definitions: sit in the strategic group too
Why they stay: your spreadsheet files them here
The audience rows connect to your ruling that audience is a real object.
★: Confirm both groups, with the strategic-grain flag as written.
if you approve: The audience list itself stays owed by the Concepts chapter — one vocabulary for both grains.
Identity elements — audio and motion
Before
Audio: sonic logo, brand music, voice talent, interface sounds. Motion: the five-layer cascade from your framework (north star → principles → behaviors → gestures → system). The honest note now on the page: the big platforms do NOT list motion as a top-level identity element — one files “video and motion standards” under digital and product usage. Keeping Motion top-level is a deliberate extension built on your framework — and the corpus already argues for it: the DevDay motion-identity loops exist as real pieces. Logo animation stays a piece TYPE; the cascade is what a piece EXPRESSES.
Industry default: motion subordinated under digital/product usage
Our evidence: the motion framework's 5-layer cascade + live motion-identity pieces (0f522a25)
★: Keep Motion as a top-level group (the framework justifies it) — with the provenance note standing.
if you approve: A grouping choice in the element vocabulary; movable later at zero cost.
Now
Keeping motion as a top-level group is our own extension — the big platforms file it underneath digital. Your framework and the live pieces justify it.
Audio: sonic logo, brand music, voice talent, interface sounds
Motion: the five-layer cascade from your framework
Industry default: motion subordinated under digital and product usage
Our evidence: real motion-identity pieces in the library — 0f522a25
Logo animation stays a kind of piece; the cascade is what a piece expresses.
★: (a) Keep motion as a top-level group.
or: (b) subordinate it under digital
if you approve: A grouping choice in the element vocabulary, movable later at no cost.
folded · The cascade in full: north star → principles → behaviors → gestures → system.
folded · The honest note stays on the page: one major platform files “video and motion standards” under digital and product usage rather than as an identity element of its own.
Identity elements — environment and digital
Before
Environment: environmental graphics, retail design (this row connects straight to the Retail & Commerce expansion — a store's design system is identity; the endcap built from it is a piece), exhibition, workplace. Digital: web and app design systems, responsive rules, icon suites. All from your spreadsheet.
Rows: 8 elements across the two groups
★: Confirm both groups.
if you approve: Element vocabulary, pipeline lane.
Now
Eight more elements across two groups, all from your spreadsheet. The environment group is the bridge to the Retail work.
Environment: environmental graphics, retail design, exhibition, workplace
Digital: web and app design systems, responsive rules, icon suites
A store's design system is identity; the endcap built from it is a piece.
★: Confirm both groups.
if you approve: Element vocabulary, in the build lane.
The seven candidates that ask for a mark
Before
— this card did not exist
Now
Seven real kinds of piece were found in your sources but not proposed for the identity shelf. The review page says to mark them if you want any of them first-class.
eBook: the documentation and editorial world
One-Pager: the documentation and editorial world
Speaker Card: a downloadable asset template
Email Header: a downloadable asset template
Blog Header Image: a downloadable asset template
Social Post Graphic: a downloadable asset template
Wallpaper: a downloadable asset template
Most of these are the media-format or page-surface reading of pieces the list already covers.
★: Take none of them, or name the ones you want on the list.
if you approve: Nothing changes unless you name one; anything you name joins as a proposed label.
folded · The review page's words: “Found in the sources but NOT identity-shelf types — real piece kinds that belong to other shelves, listed so they are not lost… If you want any of them first-class, mark it.”
folded · The five templates were all found as downloadable asset templates in one guideline book, pages 61 to 63. None was proposed as a new ratified kind in that round.
folded · This card exists because the review page asked for a mark and the console had no card for it — found by the outside audit of this walk.
8 · The rest of the agenda
The eight open calls on the schema page
Before
Eight numbered calls live on the schema page: the media_type name collision (OC1) · four subject-word values like tennis_challenge (OC2) · the empty Effects group (OC3) · dissolving the Shot Types group whose words duplicate the Framing axis (OC4 — the biggest structural one) · the stranded pricing_cta value (OC5) · three same-idea pairs (OC6) · three details duplicating Framing values (OC7) · words shared across axes (OC8). Too much for one card — walk that page and mark there; mark THIS card with how you want to handle it.
Where: the classification-schema page, calls OC1–OC8, each with its probed evidence in place
★: Walk the page this round, or defer it explicitly to the next sitting.
if you approve: Nothing executes until the page itself carries your picks.
Now
Eight numbered calls live on the schema page, each with its evidence in place. This card only routes you there — mark it with how you want to handle them.
Where they live: the classification-schema page, calls 1 to 8
The biggest one: dissolving the shot-types group, whose words duplicate the framing axis
The rest: a name collision · four subject words · an empty group · a stranded value
Too much for one card, so the marks land on that page.
★: Walk the page this round, or defer it explicitly to the next sitting.
if you approve: Nothing executes until that page carries your picks.
folded · The eight in full: the media-type name collision · four subject-word values such as tennis_challenge · the empty Effects group · dissolving the Shot Types group · the stranded pricing_cta value · three same-idea pairs · three details duplicating framing values · words shared across axes.
The moment-type promotion round
Before
64 moment-detail values clear the five-use floor (a coincidence — these are moment details, unrelated to the 64 content types), covering three quarters of everything written at that level; 76 spellings fold into them; 2 are really shelf names. The page carries a drafted definition per value and the rule is strike-to-reject: an unstruck line executes in the pipeline lane.
Where: the promotion-list page — every value with uses, shelf, and drafted definition
Scale: 577 distinct values · 379 used exactly once (the tail nobody reads)
★: Walk the page this round, or defer explicitly.
if you approve: Unstruck lines execute as written, pipeline lane.
Now
Sixty-four moment-detail values clear the five-use floor and 76 spellings fold into them. The page carries a drafted definition per value, and an unstruck line executes.
Where it lives: the promotion-list page
Scale: 577 distinct values · 379 used exactly once
The rule: strike to reject — anything left standing executes
Not the same 64: these are moment details, unrelated to the 64 content labels
The 379 single-use values are the tail nobody reads.
★: Walk the page this round, or defer explicitly.
if you approve: Unstruck lines execute as written, in the build lane.
The eight further schema proposals
Before
Beyond the numbered calls, the schema page carries eight standing proposals, each glossed here so you know what you are letting stand: rename two family rows in the word list to the names you already picked (it still says proof_advocacy where you chose Social proof) · build the what-is-it-about field the way your earlier ruling split it · un-nest the one shelf that sits inside another shelf · draw in the three typography sub-kinds you already ruled (full-screen titles · pricing end cards · opening titles) · replace the three-way shelf tier with one browse-or-catalogue label · retire two detail words that just repeat their shelf's name · retire the fifth kind-of-thing value nothing uses · build storage for occasion (ruled first-class, nowhere to live yet). Strike-or-let-stand on that page.
Where: the schema page's PROPOSE rows, each marked in place
★: Walk with agenda 2, or defer explicitly.
if you approve: Let-stand executes as written.
Now
Beyond the numbered calls, the same page carries eight standing proposals. They work by strike-or-let-stand, so this card tells you what you would be letting stand.
Two family rows: renamed to the names you already picked
The what-is-it-about field: split the way your earlier ruling said
One shelf inside another: pulled out to sit beside it
Three typography sub-kinds: drawn in — full-screen titles, pricing end cards, opening titles
The three-way shelf tier: replaced by one browse-or-catalogue label
Two detail words: retired — they only repeat their shelf's name
One kind-of-thing value: retired — nothing uses it
Occasion: gets storage at last; you ruled it first-class and it has nowhere to live
One of them matters more than it looks: the word list still says proof advocacy where you chose Social proof.
★: Walk it with agenda 2, or defer explicitly.
if you approve: Anything left standing executes as written.
folded · They work by strike-or-let-stand on the schema page: anything you do not strike executes as written.
folded · The stored word is proof_advocacy — kept here in its stored spelling so you can find the row it names.
The two overlapping product axes
Before
Every classified moment can record how VISIBLE the product is and what PART it plays — and the visibility column is being written with the role axis's words. The live distribution below shows it: contextual (122), active_use (73) and focal (43) are role-axis words sitting in the visibility column — 238 rows of cross-axis drift, against only 438 rows of the visibility axis's own distinctive words (device · in_hand). Keep one axis (what part the product plays) and fold the genuinely-different visibility words in as values, or keep two with a defining sentence each.
absent: 1,064 rows
ui_shown: 965 rows
device: 351 rows
implied: 311 rows
contextual: 122 rows
in_hand: 87 rows
active_use: 73 rows
focal: 43 rows
◮: Three of the eight words in the visibility column belong to the other axis — the merge proposal is just reading what the data already did.
★: (a) Merge to one axis — what part the product plays — folding device / in_hand / ui_shown in as values.
if you approve: A registry + classifier-prompt change, pipeline lane; existing rows re-map deterministically.
Now
Every classified moment can record how visible the product is and what part it plays, and the visibility column is being written with the other axis's words. The lean is to merge them.
Absent: 1,064 rows · also a role word
Shown on screen: 965 rows
On a device: 351 rows
Implied: 311 rows · also a role word
Contextual: 122 rows · also a role word
In hand: 87 rows
In active use: 73 rows · also a role word
Focal: 43 rows · also a role word
The three words that are genuinely visibility's own carry 1,403 rows between them.
◮: Five of the eight words in the visibility column belong to the other axis, and they carry 1,613 rows.
★: (a) Merge to one axis — what part the product plays — keeping the three genuinely visual words as values.
or: (b) keep two axes, each with a defining sentence
if you approve: A registry and classifier-prompt change in the build lane; existing rows re-map deterministically.
folded · Re-probed 24 August: five of the eight values written into the visibility column are also values of the role axis — absent, implied, contextual, in active use, focal. The three that are visibility's own are shown-on-screen, on-a-device and in-hand.
folded · An earlier version of this card said three of the eight and compared 238 rows against 438. Both numbers were built by leaving shown-on-screen out, which the recommendation itself keeps. The corrected reading is five of eight, 1,613 rows against 1,403 — a stronger case for the merge, honestly stated.
The three ungoverned moment axes
Before
Three things are written on every classified moment with ZERO registry rows governing them — nothing defines them, nothing validates them. The 17 interest words and the 5 motion states below are real vocabulary that formed by itself (note kinetic_type appearing as an interest word — it is ALSO a motion-state word, the exact cross-axis drift a registry prevents). visual_theme is 550 free-text phrases — the sample shows they cluster naturally, which suits the loop you already ruled: let free words come in, then promote the ones that keep recurring. Register the first two, run the third through that loop — or stop writing them.
interest_type — all 17, verbatim: brand_lore · category_signal · craft_detail · data_proof · data_viz · ecosystem · identity_update · interaction_detail · kinetic_type · lifestyle · lifestyle_detail · micro_detail · new_release · none · product_behavior · ui_detail · visual_motif
motion_state: 5 distinct values
visual_theme — top 8 of 550: technical/UI · clean UI · modern office · casual office · minimal clean · minimal UI · office interview · minimal studio
★: (a) Register interest type + motion state; visual theme stays open and its recurring words get promoted on review (the loop you already ruled for the detail level).
if you approve: Registry writes + a prompt change, pipeline lane; the 3,355 classification rows keep their reads either way.
Now
Three things are written on every classified moment with no registry rows governing them — nothing defines them and nothing validates them. The lean is to register two and run the third through the harvest loop you already ruled.
Interest type: 17 words, formed by themselves
Motion state: 5 words
Visual theme: 550 free-text phrases
The tell: one word appears on two of these axes at once
The theme phrases cluster naturally, which is exactly what the harvest-then-promote loop is for.
★: (a) Register interest type and motion state; let visual theme stay open and promote its recurring words on review.
or: (b) stop writing them
if you approve: Registry writes plus a prompt change in the build lane; the 3,355 classification rows keep their reads either way.
folded · The 17 interest words, verbatim: brand_lore · category_signal · craft_detail · data_proof · data_viz · ecosystem · identity_update · interaction_detail · kinetic_type · lifestyle · lifestyle_detail · micro_detail · new_release · none · product_behavior · ui_detail · visual_motif.
folded · kinetic_type appears there AND as a motion-state word — the exact cross-axis drift a registry prevents.
folded · The eight most common theme phrases: technical/UI · clean UI · modern office · casual office · minimal clean · minimal UI · office interview · minimal studio.
Where a product lives
Before
The ruling says a product is an entity that can carry its own brand identity, and the live system holds products in two CONNECTED homes: 394 entities labeled product lines inside their companies, and a 139-row products table — every row entity-linked, able to carry its own identity, parent, launch date. Nothing is lost or duplicated; the natural reading is that the products table is the detail layer under the product-line entities. Confirm that reading, and say which home a browsing surface should lead with.
Product-line entities: 394
Products table: 139 rows — all entity-linked
★: Confirm the two-home shape with the products table as the detail layer.
if you approve: A model-page statement; no storage change.
Now
A product has two connected homes in the live system, and nothing is lost or duplicated between them. The natural reading is that one is the detail layer under the other.
Product-line entities: 394, sitting inside their companies
The products table: 139 rows, every one entity-linked
What a row can carry: its own identity, parent and launch date
This is a confirmation, not a defect.
★: Confirm the two-home shape, with the products table as the detail layer.
if you approve: A statement on the model page; no storage changes.
folded · Say in the note which home a browsing surface should lead with — that is the part the ruling does not settle.
The middle level of the services tree
Before
The tree's outer levels are ruled: top = Service (what a client buys — creative production, brand strategy), bottom = craft — the level where every entry carries a full written rulebook (when to tag it, when never to; all 95 crafts have one — the only finished vocabulary we hold). Your own July frame already names the middle level DISCIPLINE, with Specialization and Execution below — under creative production sit animation, copywriting, design, photography, post-production… This card presents your own word back for confirm rather than assuming it; group and family are the alternatives.
The stored tree: Service → [middle] → craft — three levels verified live
Your July words: Service → Discipline → Specialization → Execution-as-tag (turn 77, correction pass)
★: Confirm Discipline.
if you approve: Renaming anything stored is pipeline-lane work; the reference surfaces update in the same pass.
Now
The tree's outer levels are already ruled, and your own July frame names the middle one Discipline. This card puts your own word back in front of you rather than assuming it.
The stored tree: three levels, verified live
The top: Service — what a client buys
The bottom: craft — 95 of 95 carry a full written rulebook
Your July words: Service → Discipline → Specialization → Execution
The alternatives: group, or family
The craft level is the only finished vocabulary we hold, which is why it is the standard.
★: Confirm Discipline.
if you approve: Renaming anything stored is build-lane work; the reference surfaces update in the same pass.
folded · Under creative production sit animation, copywriting, design, photography, post-production and the rest — those are the middle level this card names.
folded · A craft rulebook says when to tag it and when never to.
The per-shot crew roles
Before
The craft ruling settled multiple parents; the separate question — a vocabulary of crew roles at the per-shot level (who operated, who lit, who graded THIS shot) — was in the options and never ruled. Rule it now, or park it deliberately (parked ≠ lost — it stays on the ledger, dated; the credit_role vocabulary at the piece grain, 65 roles, is unaffected either way). The agenda offers no lean; your pick.
Status: open since the August walk · nothing stored at shot grain
The piece-grain neighbor: credit_role — 65 curated roles, unaffected
★: Your pick — rule it or park it; the agenda deliberately carries no lean.
if you approve: Parking keeps it on the ledger, dated, revisitable when the moments work needs it.
Now
The craft ruling settled multiple parents but left one question open: whether we keep a vocabulary of crew roles at the per-shot level. Rule it now, or park it deliberately.
Status: open since the August walk · nothing stored at shot level
What it would name: who operated, who lit, who graded this shot
The neighbour: credit roles at the piece level — 65 roles, unaffected
Parked is not lost — it stays on the ledger, dated, and comes back when the moments work needs it.
★: Rule it now, or park it deliberately — the agenda carries no lean.
if you approve: Parking keeps it on the ledger, dated and revisitable.
The definitions queue
Before
Most registry values have no written definition — 370 of 973 defined at the last measure, and 21 values hold an engineer's working note where a definition belongs (all six moment_family values among them — your “definitions still a little off,” counted). The crafts are the finished exception: 95 of 95 complete rulebooks, the standard everything else is measured against. A batch of drafted definitions sits on the reference, every one marked proposed; strike or amend any — the rest becomes a standing queue worked in the pipeline lane, batch by batch.
Defined: 370 of 973 live values (38%)
Working-notes-as-definitions: 21 values — incl. all six moment families
The finished standard: crafts: 95/95 complete rulebooks
★: Let the drafted definitions stand as the queue's seed; amend any that read wrong.
if you approve: Definitions land in the registry as approved; the boundary terms from the outside review (Documentary · Interview · Testimonial …) become confusable-with notes in the same pass.
Now
Most registry values still have no written definition, and this is the reference's main finding. A batch of drafted ones is on the page for you to strike or amend.
Defined: 370 of 973 live values
Holding a working note instead: 21 values, including all six moment families
The finished standard: crafts — 95 of 95 complete rulebooks
The moment families are your “definitions still a little off”, now counted.
★: Let the drafted definitions stand as the queue's seed, and amend any that read wrong.
if you approve: Definitions land in the registry as approved, batch by batch.
folded · The boundary terms the outside review raised — Documentary, Interview, Testimonial and the rest — become confusable-with notes in the same pass.
A last-reviewed date on every page
Before
The strongest documentation teams fight staleness with a per-page last-reviewed date plus an alarm when a page goes untouched too long (Google's freshness dates are the cited original). Our checks catch bad EDITS; nothing catches a page nobody edited — that class was exactly what this window's audits kept finding by hand. Proposal: every model and decision page carries a last-reviewed date, and the review loop flags pages older than a set window.
Today: per-commit gates + the re-read-every-round loop rule; no per-page stamp
What found the gap: the boot audit + final pass caught six by-hand — a stamp automates the sweep's targeting
★: Adopt the stamp — it joins the loop rules and the writing standard.
if you approve: A frontmatter field + one check; no content changes.
Now
Our checks catch bad edits, but nothing catches a page nobody edited. The proposal is a last-reviewed date per page, with a flag when one goes untouched too long.
Today: per-commit gates and the re-read-every-round rule
What is missing: anything that notices an untouched page
What found the gap: this window's audits, six pages caught by hand
The cited original: the freshness dates the big documentation teams use
A stamp automates the targeting that has been done by hand all round.
★: Adopt the stamp — it joins the loop rules and the writing standard.
if you approve: One frontmatter field and one check; no content changes.
Renaming the site
Before
“Our atlas which we should really just rename as our brandtrackers docs… done whenever it makes sense.” It touches the site title, the front door's language, and the hosting name. You said not urgent — so the mark here is timing, not whether the thread exists: with the next promote, at the final confirm, or keep Atlas after all.
Today: the site titles itself Atlas; served on the atlas host
★: Your pick on timing — with the next promote, at the final confirm, or keep Atlas after all.
if you approve: A rename pass over title, front door and host wiring — sized and run when you say.
Now
You raised renaming Atlas to the BrandTrackers docs and said it was not urgent. So the mark here is when, not whether.
Today: the site calls itself Atlas, on the atlas host
What it touches: the site title, the front door's language, the hosting name
Your words: “our atlas which we should really just rename as our brandtrackers docs… done whenever it makes sense.”
★: With the next promote, at the final confirm, or keep Atlas after all.
if you approve: A rename pass over title, front door and host wiring — sized and run when you say.