Tags
Six axes, none of them the category axis. Categories say what area an entry serves; these say what it works on, how it works, what can go wrong with it, and who it is for.
The White Lily (variegated-leaved), from The Temple of Flora, Robert John Thornton (ed.); painted by Philip Reinagle and others, 1799–1807. Amgueddfa Cymru — National Museum Wales (image 132041). Public domain. Credits
29 tags across 6 axes · 0 of 29 in use · 0 of 86 entries tagged
Every tag is reserved
The vocabulary is registered; none of it is in use. Each tag has a permanent address so it can be linked and cited now, and each will carry entries once the vocabulary goes into use. A tag turns active only when it can truthfully assert membership — the build fails if an active tag has no members, or a reserved one has any.
The axes are also served as JSON at /api/tags.json, with the
emptiness stated field by field.
Artifact — what an entry works on
The material an entry reads or writes. Factual, and orthogonal to the category axis — one entry may work on two materials at once.
- Venture record 0 of 86 Entries that read or write the shared record of company facts — the single substrate of assumptions, decisions and evidence that other tools consume instead of copying. artifact:venture-record
- Market evidence 0 of 86 The material is collected market evidence: competitor pages, industry reports, pricing sheets, signal feeds. The evidence set is what gets processed, never the conclusion drawn from it. artifact:market-evidence
- Financial model 0 of 86 Entries that read or write unit economics, runway projections or budget spreadsheets — structured numbers whose assumptions must stay visible beside their outputs. artifact:financial-model
- Customer interview 0 of 86 Discovery material captured as speech — interviews, call notes, support threads — where the unit of analysis is an utterance rather than a row. artifact:customer-interview
- Pitch narrative 0 of 86 Entries that read or write the venture story: deck prose, plan documents, investor updates. A single narrative under active revision rather than a published record. artifact:pitch-narrative
- CRM records 0 of 86 Pipeline rows are the material: contacts, stages, follow-ups, where every write has to be reversible and every deletion is suspect. artifact:crm-records
- Web property 0 of 86 Entries that read or write the venture’s public surfaces: site pages, metadata, launch checklists, rollback plans. artifact:web-property
- Media asset 0 of 86 Demo and promo material — captures, cuts, audio, captions — where rights and provenance travel with every file. artifact:media-asset
Method — how an entry works
The machinery an entry uses to do its job. Factual: it says how, and never how well.
- Structured search 0 of 86 Queries run against a defined source with recorded parameters, so the same search can be run again and compared. Coverage is bounded and the bound is declared. method:structured-search
- Field extraction 0 of 86 Named fields are pulled out of unstructured or semi-structured material into a schema, leaving a trace from every extracted value back to where it came from. method:field-extraction
- Cross-reference check 0 of 86 The entry compares a claim against the source attached to it and reports agreement, disagreement, or an inability to check. A resolution failure is itself a result. method:cross-reference-check
- Schema validation 0 of 86 Material is tested against a declared shape, and a violation fails loudly instead of being coerced into the shape and passed on quietly. method:schema-validation
- Comparison and diff 0 of 86 The entry puts two versions of the same object side by side and reports what changed, including changes that are easy to miss by reading. method:comparison-diff
- Monte Carlo simulation 0 of 86 Outcomes arrive as a distribution rather than a single confident number, with every sampled assumption recorded so the run can be reproduced. method:monte-carlo-simulation
- Drafting with citation 0 of 86 The entry writes prose in which every substantive claim carries a source that resolves. Drafting without that constraint is a different method and does not belong here. method:drafting-with-citation
Risk — what can go wrong
The failure mode to watch for when running an entry. An editorial axis, and the one we consider most load-bearing — none of it is in use yet, so no entry on this site has been inspected for risk.
- Fabricated citation 0 of 86 A source can be named that does not exist, or a real source attached to a claim it does not make. The check is resolution: every reference must resolve to a retrievable record. risk:fabricated-citation
- Silent omission 0 of 86 Material can be dropped without a word — a truncated result set, a skipped record, a filtered case — leaving output that looks complete and is not. risk:silent-omission
- Claim overreach 0 of 86 The entry can state a conclusion stronger than its evidence supports, most often by dropping the qualifier that made the original claim defensible. risk:claim-overreach
- Stale source 0 of 86 Material that has since been retracted, superseded, corrected or moved can be presented as current, because recency was never checked. risk:stale-source
- Irreproducible run 0 of 86 The entry can produce a result that a second run does not reproduce, because a parameter, a seed, a source version, or an ordering was never recorded. risk:irreproducible-run
- Confidential material 0 of 86 Point one of these at material under a confidentiality agreement, a licence restriction or an ethics protocol, and the output can leak it. Consent and confidentiality travel with the material. risk:confidential-material
- Unauthorized outreach 0 of 86 The entry can contact real people or write to live systems with no explicit human gate — the failure mode of any sales or launch automation left unattended. risk:unauthorized-outreach
Audience — who it is for
Who an entry is aimed at. Editorial, and deliberately coarse — an entry may carry several.
- Solo founder 0 of 86 One person running the whole venture, where every hour spent on a tool is an hour not spent on customers, and a step that hides its working costs two. audience:solo-founder
- Founding team 0 of 86 A small team splitting the work, where the output has to be legible to a co-founder who did not run it. audience:founding-team
- Early operator 0 of 86 Aimed at the first hires who take over a founder’s workflow mid-flight, and have to find out what it skips before a customer does. audience:early-operator
- Technical co-founder 0 of 86 Written for the person who will read it as a dependency — pinning versions, checking determinism, deciding whether it can sit in a build that ships. audience:technical-cofounder
- Student founder 0 of 86 Aimed at someone building inside a course or program, where the entry doubles as teaching material and its failure mode must be safe to demonstrate. audience:student-founder
Curated — the curators’ own marks
Hand-maintained editorial assertions by the curators. There is deliberately no sync script for any tag in this namespace.
- Reference entry 0 of 86 The curators point at this one when explaining what the review standard asks for: it is the clearest worked example they have. curated:reference-kit
- Teaching example 0 of 86 Used by the curators when teaching, because its failure mode is easy to demonstrate and easy for a student to reason about. curated:teaching-example
Batch — publication cohorts
Publication cohorts — the freshness axis. Batch ids follow batch:yyyy-w##.
This axis is registered and has no tags yet. The first is minted when an entry first carries one.
The axes that are not tags
Two axes are stored fields on the record rather than tags, because each entry has exactly one value for them. They are linked here for one-stop discovery.
- Category — the area an entry serves, with its denominated counts.
- Status — live, archived, unavailable. An archived repository is still upstream; an unavailable one returned 404 when we retrieved, and its entry stays on the map with that date. See the standard.