Notes toward an exhibition that does not exist. Three rooms, no venue, no budget, nothing borrowed. The objects and the facts are real and linked. The show is imagined; it exists only here.
Room one. The connector is a colour.
The Kyoto-based designer Kenta Yoshizawa has put out a collection called Red Craft — Akai Kōgei, and it is built out of four trades that have no business sharing a floor.
There is a leaning shelf made with bamboo-fencing technique, its profile lifted from the n-shaped silhouette of a sodegaki screen, the bamboo heated over a gas burner and bent by hand. There is a stool whose braided cords run through the frame and are pulled taut at their ends by natural lava stones, standing in for the tama weights that hold tension during kumihimo braiding. There is a lamp of slender metal rods carrying handmade washi, which sways when the air moves and takes its light from below. And there is a rabbit: a washi box, hardened and made lustrous under urushi lacquer.
Bamboo fencing, cord braiding, papermaking, lacquer. Four bodies of knowledge, four sets of tools, four apprenticeships, four guild histories. What puts them in one collection is a red.
The caption on designboom states it without embarrassment: a vibrant red palette connects the four pieces across the collection. I would print that sentence and hang it at the door. It is the most honest wall label I have read this month, because it names the connector instead of implying that the connector was found in the objects rather than applied to them.
Room two. The connector is a wallpaper.
In 1914 Yanagi Sōetsu was shown Joseon-dynasty ceramics by Asakawa Noritaka and Takumi, and something in the forms took hold of him. He opened a museum of Korean folk art in Seoul in 1924. The following year, with the potters Kawai Kanjirō and Hamada Shōji — on a trip to look at Mokujiki’s carved wooden Buddhas — he coined the word mingei. Then he went collecting: Tōhoku, Kyūshū, Okinawa, Ainu work, work from Taiwan. In 1936 the Japan Folk Crafts Museum opened at Komaba, with Yanagi as its first president. (The museum’s own history; the centenary of the word was marked in Kyoto last autumn.)
The detail I cannot get past is architectural. Yanagi designed the main hall himself, and involved himself in the fittings. The showcases were specially ordered and lined with kudzu-cloth, and the same cloth went onto the walls, so that the objects would appear at their most beautiful.
Read that as a curator rather than an admirer and it stops being charming. A man proposes that anonymous, useful, cheaply made things from places with no contact with one another share a quality. Then he builds the room in which they share it, and chooses the cloth that goes behind them, and decides which pot stands next to which.
This is not a complaint. It worked, and that is the whole of my point. A hundred years on, the coherence reads as a property of the pots.
I once helped rehang a small ceramics collection where we changed the wall colour twice. After the second change, about a third of the loans looked like errors of judgement — provincial, muddy, wrongly chosen. Nothing on the shelves had moved. The category had.
Room three. The connector is a licence.
Common Corpus runs to roughly two trillion tokens, all of it, in the authors’ words, “either uncopyrighted or under open licenses.” It is a collection assembled on a criterion entirely external to its contents. Not subject, not language, not period, not quality. Permission.
I admire it for saying so in the title. Most training corpora are not that legible. The operative criteria in this field are crawlability, litigation exposure, what a scraper could reach in a particular year, what a rights-holder did not notice in time. Those criteria are kudzu-cloth. They sit behind everything and they photograph as nothing.
My guess — and it is a guess — is that a good deal of the sameness people report in machine-written prose is not a property of the models at all. It is the coherence of a selection principle nobody printed on the label, mistaken for a quality of language, the way Yanagi’s cloth is mistaken for a quality of the pots. What I would watch for is not larger corpora or cleaner ones. It is corpora that name their red.
The room I cannot install.
Frankfurt RheinMain holds the World Design Capital title this year under the theme “Design for Democracy. Atmospheres for a better life” — more than 450 participatory projects and some 2,000 events, an entire region’s unrelated production gathered under one phrase. I would like to hang the phrase beside Yoshizawa’s lacquered rabbit and let a visitor work out which is doing more of the unifying. But a motto cannot be crated, and this is where an imaginary exhibition stops being useful.
So the last thing in the folder is not an object. It is a label, and it goes at the entrance, before anything is seen: who chose these, on what grounds, and what the grounds excluded. Every exhibition I have ever worked on had that information sitting in a spreadsheet somewhere. Almost none of them printed it.
Sources
- designboom, “red craft collection reworks four japanese traditional techniques into sculptural domestic objects” (8 August 2026)
- The Japan Folk Crafts Museum, “History of the Japan Folk Crafts Museum and Founder Soetsu Yanagi”
- Kyoto City KYOCERA Museum of Art, “The 100th Anniversary of Mingei: Kyoto’s Legacy of Everyday Life” (13 September – 7 December 2025)
- “Common Corpus: The Largest Collection of Ethical Data for LLM Pre-Training,” arXiv:2506.01732
- World Design Organization, “WDC Frankfurt RheinMain 2026”