Mapping cultural infrastructure.
An experiment note. Can open records, published by the people and places that make culture happen, be read back into a map of the data ecology of the cultural and creative industries? And can that map put artists in it from the start, rather than as rows in someone else's database?
Where the idea comes from
In 2018, while working on Mycelia, Mark Simpkins sketched an Institute of Cultural Infrastructure: a project to map, share and explain the data that sits under culture and creative work. Infrastructure meant everything that gets an idea out of an artist and in front of an audience. Teaching and training. Recording studios, photographic studios and rehearsal rooms. Venues, theatres and galleries. The organisations that represent artists and manage their rights.
It drew on two earlier pieces of work: the British Council's Mapping the Creative Industries toolkit (2010) and the Open Data Institute's guide to mapping data ecosystems (2018). It hoped to find which data is, or could be, a shared commons that everyone builds on, and to show where data and value are being held and controlled.
The draft also named the problem. Much of culture is run through event-sales and content-management systems. They are good at dates, headline names and scrolling calendar blocks. They are bad at the things that keep going after the night: collaborations, influences, who helped whom. Websites built on them go stale as soon as the season ends, so the connections that make a scene work are never written down anywhere that lasts.
The idea stalled in 2018 because there was nothing to map with. Each organisation's data lived in its own systems, and there was no shared format artists could publish into themselves. That has changed.
What CultureBlocs already gives it
The records needed for a first map already exist in the format. Each one lives in its publisher's own ATProto account, so the map is read from many places and owned by none of them.
| The map needs | Already in the format |
|---|---|
| Artists and makers as named parties | creative.profile |
| What they made | creative.work |
| Who worked with, showed, released or represents whom | creative.connection |
| Venues, studios, rehearsal rooms, suppliers and other places | place.profile |
| What happened, and who was on | calendar.event + place.lineup |
| What a person or place offers or needs, for a while | offer |
| Audiences meeting the work | bead (visit, watch, listen, read…) |
| The making side: rehearsing, recording, presenting | bead (performance, bloc, note) |
| Which outside registries hold a person or a work | externalIds (ISNI, IPI, ISRC, ISWC, Wikidata…) |
| Which platforms a venue depends on | place.profile links (ticketing, listings) |
Three things in the format matter here more than anywhere else:
- The artist publishes their own node. A creative profile is a self-asserted record in the creative's own account. They are on the map because they put themselves there, not because a venue or a label listed them.
- Connections outlast events. A
creative.connectionsuch as collaborated_on or represented_by stays true after the run ends. Relationship types are advisory, so recorded_at, rehearsed_at or trained_at can be tried without a schema change. - The registries show up on their own. Every external identifier is a trace of a database someone else controls. Counting them shows which registries a scene depends on, and who is missing from them.
The experiment
Map one small, real patch of cultural infrastructure, using only records people chose to publish, then compare that map with one drawn by hand.
1. Pick a patch
Peckham and Nunhead. There is already sample listings data for local venues, a Peckham Digital directory and the Nunhead Art Trail, and the people are close enough to meet. A patch, not a sector: it keeps the numbers small and lets everyone involved see the result.
2. Help people publish themselves
With their agreement, help around ten to twenty practitioners and five to
ten places publish a profile, a few works and the connections they're happy
to state, using the Start tools. Places should
include the less visible ones as well as venues: a rehearsal room, a
print studio, a supplier, a college course, a rights or support body.
place.profile’s spaceTypes says what kind of
place each one is (rehearsal_room, recording_studio,
supplier…), and new kinds need no schema change.
3. Read it back
A read-only page in the style of the Frieze Week report. It asks the public backlink index and each publisher's own repository, stores nothing, and draws two views:
- The ecology. People, works, places and events as one graph, with audiences arriving through beads that point at events and works.
- The dependencies. Which ticketing and listings platforms the places link to, which registries the people and works appear in, and who is described by others but publishes nothing themselves.
4. Draw it by hand too
Run one workshop using the ODI's data-ecosystem mapping method, with the same people, to draw the flows that records don't carry: money, contracts, who holds the mailing list, who sets the price. Put the two maps side by side. The difference is the finding: what a shared commons is missing, and where data and value are held by someone outside the room.
How it fits together
Why a venue or studio would take part
The 2018 draft asked this directly, and it's the right question. Some honest answers, to be tested in the experiment rather than assumed:
- Being findable and pointable-at without paying a listings platform, and without handing over audience data.
- The record survives a change of website, ticketing provider or CMS, because it lives in the venue's own account.
- Collaborations and credits stay attached after the event page is gone.
- A picture of who came and what it connected to, read from beads people chose to keep, that can go into a funding report without the venue collecting personal data.
- A clearer view of its own dependencies, which helps when a platform changes its terms.
What could go wrong
- A map of people is a map of people. Networks of freelancers and informal spaces can be exposed by being drawn. Only what each person published goes on the map; one-way connection claims are shown as claims; anyone can be left out through the site's hidden list; small counts are never shown as audience figures.
- Open data is open to everyone. The large players can read the commons too. The difference is that they can't own the source or switch it off. That trade-off should be stated plainly to participants, not buried.
- The map shows who publishes. It will favour people who are comfortable with the tools. It is a sample, not a census, and the hand-drawn map is there to show who is absent.
- Location. Places publish exact locations; people do not. Practitioners' "based in" stays at whatever precision they choose, and bead locations stay stripped on publish as they are now.
Questions it should answer
- Are the existing records enough to draw a useful map of a small scene, or is something structural missing?
- Does a place need a declared kind (rehearsal room, studio,
rights body) beyond a tag? Answered on 6 November 2026:
venue.profilebecameplace.profile, with aspaceTypeslist. Do its starter kinds cover the places this workshop finds? - Do relationships like recorded_at or trained_at get used, and do they belong in the advisory list?
- Is a record for data flows between organisations needed, or is the hand-drawn map the right place for those? Nothing new goes in the format until the experiment shows it's needed.
- Would a participant find the map useful enough to keep their records up to date?
Related
The Frieze Week report is the reading-back method this would reuse. The format has the full record definitions. The how-to covers publishing a profile, a venue and an event.