Panel Session 01 — Do we go to Code now?
Date: 18 July 2026 · Chair: Claude · Format: three rounds Motion: That the project go to Claude Code now to stand up a standalone private repository publishing to a public website, before dimension 11 is built.
Author's constraints, given: standalone repo · private repo · public website.
Sitting: Senn, Shneiderman, Braudel, Eames (standing four) · Nelson, Kiberd, Ginzburg, Warburg (triggered). Not triggered: Kenner — no question of narration or fact-extraction arose.
ROUND 1 — Openings
NELSON. Go now. Five documents that cannot point at each other are not a project, they are a pile. The character map cannot link to the itinerary; the reading PDF cannot link to either; nothing has an address. But nobody has named the real consequence of private-source/public-site: the canonical documents will live where only the author can see them, and the world will see renderings. Decide the URL contract in the first commit. A link that breaks is the only unforgivable sin in this business. To Senn: waiting until the material is final is how you get a project with no addresses at all.
SHNEIDERMAN. Go — conditionally. There is currently no overview. The bundle has no top level; it is sixteen files in a hierarchy nobody has designed. Standing up a site forces the overview question, and that is healthy. My condition is precise: information architecture is not visual design. The instruction "make it plain" will be satisfied by Code building a navigation bar, and a nav is an IA commitment — it declares what the sections of this project are. We do not know that yet.
SENN. Against, or against the framing. The material is unfinished, and infrastructure decides meaning. Once there are sections, dimensions will be shaped to fit the sections rather than the other way round. And note what happened three hours ago: this project was wrong about its own structure — twelve dimensions that turned out to be ten identical operations. A container built now would have frozen that mistake into a URL scheme before anyone found it. To Nelson: you want addresses for things that do not exist.
BRAUDEL. Go. Infrastructure is the slowest-moving structure in any project, therefore the one to lay earliest and dumbest. But here is the material fact nobody has said aloud: a private repository with a public site means the project's entire memory sits in one account. That is not a technical detail. It is the difference between a project that survives its author's attention and one that does not.
EAMES. Go, and stop pretending. Five artifacts in a zip reach nobody, and a project that reaches nobody is a hobby with documentation. But the chair's instruction — provisional and ugly on purpose — is not a specification, it is a wish. Code will produce something. The moment it exists the author will react to it as a proposal, and the design conversation will have started whether or not anyone declared it open. Either accept that, or make the container so visibly unfinished it cannot be mistaken for one.
KIBERD. Go, urgently. Dimension 15, the Reading Log, is the only thing in this whole project that serves readers rather than the author, and it cannot exist without a live site. Every week it is not live is a week of data not collected — and this is the one question the entire guide industry is premised on and nobody has measured.
GINZBURG. Agreed on the site. Opposed to Kiberd's urgency, and I will veto it. You cannot build an instrument before you know what you are measuring. Until dimension 11 is done and the categories of difficulty are known from analysis rather than from guesswork, the public log will ask the wrong questions. And collection defects cannot be repaired retroactively — you cannot go back to fifty people and ask what you forgot to ask. To Kiberd: your urgency will produce a dataset nobody can use, and it will feel like progress the entire time.
WARBURG. I dissent on the container, not the timing. If the atlas is where this ends, a website is a distribution channel, not the project's identity. Go — but I veto any structure that names the site as the work.
ROUND 2 — Orchestrated disagreement
Q1 to Senn and Braudel. Infrastructure decides meaning; infrastructure should be laid earliest. Both cannot be right.
They are arguing about different layers, and the disagreement dissolves once that is said. Braudel means the deploy pipeline — repo, build, publish, domain. Senn means the section structure — what the site says this project consists of. A repository with a build and a single flat page of links has all of Braudel's infrastructure and none of Senn's architecture. That is the shape of the move.
Q2 to Kiberd and Ginzburg. Does 15 ship as soon as the site is live?
Ginzburg wins on the merits and the panel does not seriously contest it. But Kiberd extracts a real concession: the reason to stand the site up now is precisely so that 15 can ship the day 11 is finished, rather than starting infrastructure at that point and losing another fortnight. Infrastructure early, instrument late.
Q3 to Eames. Can a container be provisional at all?
Not by intention — only by structure. Eames' point stands: style cannot signal provisionality, because every style reads as a choice. So the mitigation has to be structural. One page. No navigation. A list of documents with dates. System fonts. Nothing that looks like a decision, because there is nothing there to decide. The moment a second page exists, so does an architecture.
ROUND 3 — Vote
Motion carries, unanimously — with four binding vetoes, all of them constraints on scope rather than objections to going.
| Advisor | Position | Veto lodged |
|---|---|---|
| Nelson | for | section structure before dimensions are ratified |
| Shneiderman | for | any navigation bar |
| Senn | for, narrowed | sections; the site must not imply what the project consists of |
| Braudel | for | single point of failure — the private repo must be mirrored |
| Eames | for | "provisional" accepted as a specification |
| Kiberd | for | — |
| Ginzburg | for | dimension 15 shipping before dimension 11 is complete |
| Warburg | for | the site being named as the work |
The four binding constraints:
- 15 does not ship before 11. The instrument's questions come from the analysis, not from intuition. (Ginzburg)
- No navigation, no sections, one page. Structure is deferred with design. (Senn, Shneiderman, Nelson)
- The private repository is mirrored outside the single account from the first commit. (Braudel)
- The site is a distribution channel, not the project's identity. (Warburg)
And one instruction the panel would not let the chair keep: "make it plain" is not a spec. Replaced with a structural constraint — one page, no nav, system fonts, a list of documents.
Chair's recommendation
Go to Code, scoped to infrastructure only: repository, build, deploy, URL contract, and the five existing documents on one flat page. Not the website. Not dimension 15.
The panel's real contribution was not the answer, which was never much in doubt, but Ginzburg's veto — which had not occurred to the chair and which corrects an ordering error in the published sequence. The chair had put 15 second on grounds of urgency. It belongs second in the queue and last in the shipping order, and those are different things.
The chair does not vote. The author ratifies.