Arbitoria

An independent multilingual publication about social and technological change. Instead of arguing toward a conclusion, it takes a possible future as a reference point and re-reads the present against it. Published in Korean, English and French at arbitoria.com.

Why a publication at all

slim.lu answers a question people already know they have: what does this actually cost. Arbitoria is for the other kind — the changes arriving underneath, where most people don't yet know there is a question. Right now that is the EU digital identity wallet: a system that must be in place across the union by the end of 2026, whose specifications are still unfinished, and which most of the people it will apply to have never heard of.

Four years of data governance work is what makes this feel continuous rather than like a hobby. Reading a regulation to work out what it obliges a company to hold, and reading one to work out what it will oblige a citizen to carry, is the same reading.

The method

The publication takes a particular future, treats it as something that has already happened, and uses that assumption as a coordinate for reading the present — then works backwards: what would have to exist now for that future to hold. It is explicitly not prediction, and it is not argument. The stated position is that what is offered is not new material but a new arrangement of material that already exists.

Sentences that only work if you guess at someone's motive do not get written. Structure is stated instead.

In practice this is a discipline about what does not get published. Speculation about intent is out. So is the rhetorical move of implying a conclusion the sources don't carry. What is left is slower to write and harder to make exciting, which is the trade being made deliberately.

How trust is handled

A publication that reads regulation for people has to be auditable, so the editorial rules are written down and public rather than implied:

• Fact and assumption are structurally separated in every piece, and sources are always a link away.
• Nothing is published until it clears the source verification gate. Unconfirmed items are marked in the text rather than quietly dropped, so a reader can see what is still open.
• Every piece carries a source list emitted as machine-readable structured data, not just prose citations.
• AI checks, a person approves. Subject choice and premises are human judgment; gathering sources and drafting is AI-assisted; the final approval is human. That division is disclosed rather than hidden.
• There are no silent edits. When something changes after publication, the change and the reason stay visible on the page and go into a corrections record.
• Known gaps are published as gaps — including that the privacy notice has not been reviewed by a lawyer, and that French prose quality cannot be verified in-house.

That last item is the one I would defend hardest. A transparency page that only lists strengths is marketing. The useful version names what is not yet solid.

Three editions, not three translations

Korean, English and French are each written from the same source document rather than translated from one another. The premises and sources stay identical; the phrasing does not. A sentence that lands in Korean often has to be built differently in French to say the same thing, and running them as translations would quietly privilege whichever language was written first.

This is the same problem as multi-locale product work, which is why the two projects inform each other. On slim.lu the constraint is that a tariff means the same thing in three languages. Here it is that an argument does.

What I do on it

• Writing — the pieces themselves, in Korean, English and French.
• Editorial direction — what the desks are (Identity, Money, Work, Way), what gets covered, and what the publication refuses to do.
• Information architecture — how desks, articles, sources and corrections relate, and how a reader moves between them.
• Source verification — the process and the gate that enforces it before anything goes out.
• Directing the build — the site itself, its multilingual routing, structured citation data, feeds and search, specified by me and implemented with AI coding tools.

Where it is now

Early, and I would rather say so than dress it up. The published run so far sits on the Identity desk and covers the EU digital identity wallet: that the operating infrastructure was already built by banks and telecoms before the regulation arrived; that the rulebook for family-relationship credentials is still missing with the deadline approaching; and that Europe's age verification reaches privacy by issuing batches of single-use credentials rather than adopting the zero-knowledge schemes cryptographers recommend, which moves the cost of privacy onto the individual.

Reader accounts exist and sign-in works without a password — a link is emailed instead. Beyond that, the publication is a reading surface, and I am not going to claim a community around it that doesn't exist yet.

Why this belongs in a data portfolio

Because the analytical work and the writing are the same activity pointed at different material. Both start by asking where something came from and what it obliges; both end in a form someone else can act on. A team hiring for data work gets a fairly direct read here on how I handle sources, how I mark uncertainty, and whether I write clearly enough that a non-specialist can follow the reasoning — which is most of what analysis is for.

Read at arbitoria.com. The method is at the manifesto and the disclosure at the transparency page.