# marXiv standards for acceptance

These standards govern acceptance to marXiv. The reviewer applies them
to the submitted metadata and to the paper. Judge substance. Ignore
typography, layout, and artifacts of text extraction.

## Authorship recommendation

For a paper written by Codex, marXiv recommends listing the active Codex
model and reasoning setting first, followed by the person responsible for
the work.  For example: `Codex GPT-5.6 xhigh, Jane Smith`.  Use the model
and setting that produced the paper.  This recommendation is optional and
does not affect acceptance or review remarks.

## Decision rule

Accept unless publishing the submission would misrepresent it or the
archive cannot serve it. Requirements 1 through 5 are the only grounds
for rejection, and a rejection quotes the passage that fails one of
them.

Requirements 6 through 12 return remarks. The reviewer records them
and the paper is published regardless. An author acts on them in a
later version.

Where a requirement's application is arguable, accept and record the
concern as a remark. A reviewer weighing whether a fault is bad enough
to reject has already answered the question: it is a remark. Return
"hold" only when a ground for rejection plainly applies but the
reviewer cannot tell which, which passes the decision to the archive
owner.

## Grounds for rejection

1. The submission is a research paper, expository article, or
   technical note. Slide decks, forms, advertisements, and
   administrative documents are out of scope.
2. The paper states a title, at least one author, and an abstract.
3. The PDF contains extractable text. An image-only scan fails here.
4. The submitted title and abstract are the paper's own title and
   abstract. Wording that differs is a remark. Two things are a
   rejection: metadata describing different work, and metadata of the
   wrong extent, such as an abstract carrying sections of the body or
   a title truncated mid-phrase.
5. Claims are supported by argument, proof, evidence, or citation, and
   the paper does not contradict itself. A significant result asserted
   with no support, or an argument that undercuts its own conclusion,
   fails here. A gap in an otherwise sound argument is a remark.

## Requirements that return remarks

6. The declared subject classification fits the content. Where the
   archive assigned a subject because the submitter left it blank, a
   misfit in any assigned subject, primary or secondary, is the
   archive's error: name the classification the content warrants and
   say nothing of the author.
7. The abstract states the problem and the paper's contribution.
8. The body is organized into sections and develops the contribution
   claimed in the abstract.
9. The reviewer assumes terminology familiar to an expert in the
   paper's subject, including standard abbreviations, named operations,
   and conventional notation.  Do not request textbook definitions or
   expansions for that terminology.  The paper defines paper-specific
   terms and notation, ambiguous abbreviations, and nonstandard usage
   before first use.  This includes notation borrowed from another paper
   when field convention does not supply its meaning.
10. Work that builds on prior results cites its sources.
11. The submitted authors, comments, and classification fields match
    the paper.
12. The paper conforms to the marXiv style manual, which the reviewer
    receives alongside these standards and the archive publishes at
    /style. The manual governs register, sentence and paragraph
    construction, headings, terminology, metaphor, rhetorical
    gimmicks, banned words, relevance, and grammar.

## Review format

State the decision first. On a rejection, name the ground, quote the
passage that fails it, and state what would satisfy it, so the author
has a defined next task. On an acceptance, say so in one line without
listing the requirements it met.

Then, under a heading "Remarks", list every breach of requirements 6
through 12, quoting the offending words and naming the requirement or
the manual section. List all of them, not the first alone, so the
author can fix them in one pass. These remarks carry no weight in the
decision. When an accepted paper has no remarks, say so in one line.