About this prototype
A prototype that publishes the Kouigenji Monogatari TEI/XML without an index server.
What this does
The Kouigenji Monogatari text database (TEI/XML based on Ikeda Kikan, Chuokoron-sha) is published as 55 entries: the 54 volumes plus the whole. The catalogue is read straight from on-chain metadata, and the text itself comes from IPFS. There is no database and no index server behind this page.
Source text and licence
The base text is Ikeda Kikan's Kouigenji Monogatari (Chuokoron-sha). The TEI/XML is published on the upstream site under CC0 1.0. No permission or account is needed to read it.
What lives on chain
Three kinds of contract sit on Sepolia, an Ethereum test network: a data NFT that stands for the role of publishing a volume, a datatoken that covers a single access, and CorpusAnchor, which records the root that folds all 54 volumes into one value. Each contract has its own page explaining what Etherscan is actually showing you.
Where the interface comes from
The interface is ported from Ocean Market (Ocean Protocol) by way of deltaDAO's Pontus-X Portal and Clio-X, under Apache-2.0. The CSS and the DOM structure are carried over as they are, rather than reimplemented from the look of them. Only two things changed: routing (Pages Router to App Router) and where the data comes from (the chain and IPFS directly, instead of the Aquarius index server).
All 54 volumes as a single record
Alongside the 54 per-volume records, the same text is published as one TEI/XML file of 25,065 lines. It is a different level of granularity, so it is kept out of the per-volume listing.
What the upstream site does better
This is not a replacement for the upstream site (the Kouigenji Monogatari text database). Full-text search, the SPARQL endpoint, IIIF alignment with page images, and plain rendering speed are all better there. Only four things are added here.
- A fixed version. Which state of the text a citation points at can no longer be moved after the fact.
- A record of citation. Consulting the text can be recorded on chain, so a footnote can carry the transaction hash.
- Third-party verification. You do not have to trust this page; your own browser can recompute the root.
- Per-volume reference counts. The record of who consulted what, and when, lives outside the publisher's hands.
What is not here
Some features of the upstream portal were left out. There is nothing behind them here, so porting them would only produce empty panels.
- Publishing and editing (there is a single publisher, and nothing is published through the interface)
- Buying and pricing (the text is CC0, so there is nothing to sell)
- Compute-to-Data (there is no compute environment to run jobs in)
- Profiles, bookmarks and onboarding
Caveats
This is a prototype, running on a test network. The root was anchored by the publisher alone, so at this stage it is no more than a record the publisher wrote about themselves. It starts to mean something only once an independent third party runs the same computation and anchors the same root. And putting the text on IPFS does not make it permanent: if nobody keeps holding it, it stops being retrievable.