Glossary
The vocabulary, in one sentence each.
A record page uses a handful of words in narrower senses than usual. None of them is meant to keep anybody out, and none of them is required to file a request or read a plain-language summary. Terms are linked from the pages they appear on, so this page is somewhere to arrive rather than somewhere to read first.
- asserted_by
- Which named party is answerable for a statement. Nothing in a record is unattributed, including the plain-language parts.
- attestation
- A person putting their name to a statement that no machine can settle, such as whether formal code says the same thing as a paper.
- backlog
- An activity-sorted view of active records. Older recorded activity means only that less has been recorded recently, not that a claim is weaker.
- basis
- What kind of thing a statement in a record is: a quote from a source, something an author attested, an editor's inference, a machine's output, or a restatement of material already in the record.
- claim
- The specific mathematical statement a record is about, quoted from wherever its author actually stated it.
- coverage
- How much of a claim the evidence actually reaches. Formal code covering one lemma of a paper is real coverage of that lemma and no more.
- depends_on
- An attributed link saying that a record cites, builds on, or uses a result from another record or an external source.
- digestion
- A restatement of a claim for a particular audience, signed by whoever wrote it and kept separate from the claim itself.
- evidence entry
- One piece of material a record holds about its claim — a formal artifact, a rebuild, a review — each with its own attribution and its own limits.
- execution receipt
- The record of a program actually having been run: which tool, which version, on what, when, and the fingerprint of the log it produced.
- formalization handoff
- A short, attributed map from an informal claim to the definitions, tools, existing code, and next open task someone would need to formalize it.
- freshness
- Whether the record has been checked against the version of the source that is currently published. It is computed, not chosen.
- lifecycle
- The state the record's author has put it in — active, superseded, or withdrawn — which describes the record and never the standing of the mathematics.
- manuscript hash
- A fingerprint of the exact file a record was built from, so that a later reader can tell whether they are looking at the same document.
- manuscript version
- The publisher's own name for a revision, such as an arXiv v3, which is compared before hashes because a regenerated file can differ byte for byte without anything having changed.
- needs checking
- A specific paper passage a reader has suggested examining, with a reason and a version of the source attached.
- open invitation
- A bounded task nobody has done yet, named in the record so that somebody else can pick it up.
- record
- One page about one mathematical claim, holding what was claimed, who said it, what evidence exists, and what nobody has done yet.
- review map
- The short, attributed guide at the top of a record that says where to begin, what looks delicate, and what background a reader needs.
- reviewer page
- A static page listing one identity's attributed entries across records. It is a census, not a rating.
- stale
- The source has moved on since this record was last checked against it. It says nothing about whether the mathematics changed.
- stratum
- Which layer a statement comes from — the author, an editor, or the wider community. Layers are always labelled and never merged into one voice.
- taken
- A reviewer has named themselves as working on a bounded invitation. It remains open work until a recorded entry points to what they did.
- withdrawn
- The record has been taken out of use by whoever maintains it. The page and its history stay where they are.
What these definitions are
- They describe how this project uses each word, which is sometimes narrower than the field's ordinary usage.
- None of them defines a word for whether a claim is right, because the project has no such word.