Media and Information Research Commons

A verified research record platform for media and information studies

The Commons is a platform on which research is published as a Verified Research Record: the data, the method, the findings and the documented use of AI, reviewed by named and paid reviewers and made permanently open. A written article can accompany the record but is not required. The Commons is run by the Media and Journalism Research Center and is open to institutions and to individual researchers.

The registration template, the review checklist and the AI protocol are public and may be reused by any journal, department or research group.

Academic publishing is organised around one object, the article. Journals sell it, reviewers assess it, and universities count it. This arrangement made sense when writing the article was the most demanding part of research. That is no longer the case. AI systems now produce a plausible article, references included, in minutes. The article has stopped being evidence that research was carried out.

The publishing system has not adjusted. Publishers charge authors to publish and libraries to read. Reviewers work without payment and assess the text rather than what lies behind it. Paper mills sell authorship, references are fabricated, AI-generated text goes undeclared, and the volume of articles grows because volume is what is measured.

The Commons is based on a different premise: what should be published, reviewed and credited is the research itself, that is, the question, the method, the materials, the findings, and a complete account of how the work was done, including the part done by AI. Written articles and other texts are derived from that record. They remain useful, but they are secondary to what they describe.

MJRC is a research centre, not a publisher. The platform was built for the centre’s own work and is being opened to other institutions and researchers because the problems it addresses are shared across the field.

The Verified Research Record

The Commons publishes Verified Research Records in place of articles. A record is made up of the following elements, listed in the order in which they are reviewed and weighted.

Materials and data reviewed first, weighted mostThe corpus, dataset, codebook, interview guide, sampling frame and coding decisions on which the study rests. Each is published with its own identifier and can be cited independently. A well-documented dataset with no accompanying text is a complete publication on the Commons. Where data cannot be shared for reasons of consent or safety, the documentation is published and the data are described and held under restricted access.
Question and method reviewed before resultsWhat was asked, why, and how it was studied. Registered and reviewed before analysis where possible; for completed work, reviewed against what was actually done. This is where reviewers judge whether the study is sound.
AI and analysis log required in every recordA step-by-step account of how the data were processed and analysed, including every AI-assisted step: tool, model version, inputs, outputs, and how a person checked them. Published in full under the AI protocol.
Findings record the citable resultA structured statement of what was found: each finding on its own line, linked to the materials and analysis steps that support it, with the level of confidence assigned by authors and reviewers. This is the element most readers will cite. It is checked line by line in review and can only be changed through a new version.
Written work optionalA record does not require an article. Authors may attach a full article, a short paper, a policy brief, a methods note or teaching material, and others may later publish work based on the record with attribution. All such texts are listed on the record page and link back to it. Where text is attached, reviewers check only that it is consistent with the findings record and the materials.

Layout of a published record

Every accepted record receives a permanent public page. The example below uses placeholder content.

Commons record 2026-0041 · version 2 · published 14 November 2026 · CC BY 4.0

How local newsrooms in three EU countries use generative AI in daily production

Accountable authors: A. Author (design, interviews, findings) · B. Author (corpus coding, analysis log) · Reviewed by C. Reviewer and D. Reviewer

Materials and data

Interview guide, 42 semi-structured interviews (restricted access, consent summary public) reviewed
Newsroom policy corpus, 118 documents, with codebook reviewed
Sampling frame and recruitment notes

Question and method

Registered 3 March 2026, before fieldwork. Comparative case design, three countries, qualitative coding with AI-assisted first pass and full human recoding of a 25% sample.

Reviewer note: sample skews toward larger local outlets; limitation acknowledged in findings.

AI and analysis log

Transcription: model X v3.1, 42 files, checked by author against audio (12 corrections)
First-pass coding: model Y, prompt file attached, 118 documents
Human recoding: 30 documents, agreement 0.81
Drafting: sections 1 and 5 of the attached article drafted with AI, rewritten by A. Author

Findings record

F1. All 14 newsrooms use AI for transcription; none disclose it to readers. high confidence
F2. Editorial policies on AI exist in 9 of 14 newsrooms; 6 were written after 2025. high confidence
F3. Journalists report that time saved is reallocated to output volume, not verification. moderate confidence

Reviews and responses

Two signed reviews against the Commons checklist, authors’ responses, and the editorial decision. Automated checker report: 96 references checked, 96 found; disclosure complete; materials present.

Written work based on this record

Article, 6,800 words, published in New Ideas in Media and Communication (authors) · Policy brief, 4 pages (authors) · Secondary analysis of the policy corpus, 2027 (E. Other, Commons record 2027-0112). Version history: v1 (14 Oct 2026), v2 (14 Nov 2026, confidence of F3 revised after review).

Each element carries its own citation. A researcher can be credited for a dataset or a findings record without having written any text, and a reviewer’s signed review is itself a published and credited piece of work.

The AI protocol

AI is used throughout research in this field, including at MJRC. The centre uses AI systems extensively and has invested in them, mainly for processing data: transcription, translation, document classification, coding of large corpora, cross-checking of sources and detection of inconsistencies. Over the same period the centre has increased its spending on editorial work and on paid experts, because processing more data well requires more human judgement, not less. That combination is the model the Commons proposes for the field: machines handle volume, people decide what it means and sign for it, and every step is recorded.

Seven rules

AI may be used at any stage of research and writing. It is never the accountable party; a named person is.
Every AI-assisted step is logged: task, tool, model version, inputs, configuration or prompt, outputs, and who checked them.
Any AI output that contributes to a finding is verified by a person on a stated sample, and the result of that verification is reported.
AI-drafted text is disclosed by section. Undisclosed drafting discovered after publication leads to withdrawal of the record.
Reviewers receive the full log and assess whether the AI use was appropriate to the task and adequately checked.
Prompts, configuration files and code are deposited with the materials so that the processing can be repeated.
The protocol is published, versioned, and updated as tools and practice change.

Application to common tasks

Task	Role of AI	Role of the researcher	Logged
Transcription and translation	Produces the first full text	Checks a sample against the source; corrects	Model, sample size, corrections
Corpus coding and classification	First-pass coding of every item	Writes the codebook; recodes a sample; reports agreement	Codebook, prompt or configuration, agreement score
Literature and source search	Identifies candidates	Reads and selects; confirms every reference exists	Tools used; reference check result
Analysis of large datasets	Runs and summarises	Designs the analysis; interprets; checks edge cases	Code or notebook, model version, checks performed
Drafting	Produces draft text where authors choose	Rewrites, decides, signs	Sections drafted with AI
Interpretation and findings	May suggest	Decides and takes responsibility	Author accountable for each finding

The protocol does not restrict the use of AI. It is what makes AI-intensive research publishable, because it allows a reviewer and a reader to see exactly what was done. Participants are expected to contribute to its revision.

From registration to published record

RegistrationAuthors submit the question, method, planned or existing materials, the AI and analysis log, and the accountability statement naming who signs for what. Registration can take place before the research or after it. The review fee is paid at this point.
Automated checksReferences are checked for existence, the disclosure for completeness, and the materials for presence. Authors see the result and can correct problems before review begins.
ReviewTwo named reviewers, matched to topic and method, work through the checklist in the order of the record: materials, method, log, findings, then any attached text. They sign their reviews. Reviews may be critical; they are not anonymous.
DecisionAccepted, returned for changes, or declined with written reasons. Reviewers are paid in all cases.
PublicationThe record page is published with every element, the signed reviews, the checker report and a version number. Authors retain copyright and may publish derived texts anywhere.
After publicationAuthors can add corrections and new versions. Others can attach replications, secondary analyses or comments. Nothing is deleted; every change is dated.

Review fee

Reviewing a record means reading the materials, checking the method, going through the AI log and verifying the findings record against the evidence. For a typical record this takes each reviewer two to three hours. The Commons pays for that time, and the fee is calculated from it.

2 reviewers × 2.5 hours × €50 per hour = €250 for review. Review is 70% of the fee, so the fee is €250 ÷ 0.7 ≈ €360. The hourly rate is derived from €400 per day, the customary rate for an experienced consultant on EU-funded and comparable projects.

€360

Standard fee

Per record, paid at registration, regardless of the decision.

70%

To reviewers

€125 to each of the two named reviewers.

30%

To MJRC

€110 for automated checks, editorial coordination, hosting and long-term preservation.

For records that require substantially more reviewer time, such as those with very large datasets, several methods or multiple data languages, the fee may be higher. It is quoted at registration, before payment, using the same hourly basis.

A reduced fee of €180 applies to researchers without project funding and to institutions in lower-income countries, on request and without documentation. Full waivers are available and are decided by the editorial coordinator, not by reviewers. Participating institutions may pay for their researchers in bulk. MJRC publishes annual accounts showing records received, fees collected and payments made.

There is no charge to read, cite or reuse anything on the Commons.

NIMC and the Commons

New Ideas in Media and Communication publishes work that brings a new idea to the field rather than incremental findings. It has no author fees and is free to read. The journal now operates on the Commons:

Every NIMC article begins as a Verified Research Record and goes through the same registration, automated checks and paid, signed review.
NIMC is the journal layer for accepted records whose main contribution is conceptual. These are published in NIMC as articles, with the record as the verified basis behind them. Records that are sound but do not meet NIMC’s new-idea criterion are published on the Commons without going into the journal.
NIMC publishes a small, fixed number of articles per year.
NIMC’s editorial board provided the first reviewers and the first version of the review checklist.

The Commons is the process and the record. NIMC is one journal built on it; other journals can be built on it in the same way.

NIMC: scope and submission guidelines

Register with the Commons

Individual researchers register directly. Institutions submit a request first; formal participation follows the institution’s own approval process. Both are free.

Individual researchers

Complete the short form below (two minutes).
MJRC confirms by email.
You can then submit records and be nominated as a paid reviewer.

Registered researchers can submit records, be nominated as paid reviewers, and contribute to the annual revision of the AI protocol. No commitment beyond that.

Institutions

A contact person submits a request to participate (no commitment).
MJRC sends a one-page participation note describing what the institution can do on the Commons and what it agrees to, for internal approval.
Once the note is confirmed by an authorised person, the institution is listed and its researchers can be grouped under it.

Institutions can submit records from their staff, nominate reviewers, pay fees in bulk, use the templates internally, or run a journal or review process on the Commons.

Questions

Does the Commons restrict the use of AI in research?

No. The model assumes AI is used and requires that the use be logged, checked and signed for. AI-intensive research is publishable on the Commons because the log makes it reviewable.

Does the model apply to qualitative and theoretical work?

Yes. For interview and ethnographic work the materials are the guide, the codebook and the corpus description, with raw data usually under restricted access. For theoretical work there are no data; the record consists of the argument, its sources and a findings record of the claims made, and reviewers assess whether the argument is new and whether it holds.

Is a Verified Research Record equivalent to a journal article?

It is peer-reviewed, permanent and citable, with an open review record, and its elements can be cited separately. Whether an institution counts it is the institution’s decision. The published review record gives an evaluator more to assess than most journal articles do.

Can text based on a record be published in a journal?

Yes. Authors retain copyright. Most journals accept work that has been reviewed or preprinted elsewhere; the record remains the verified basis.

Who are the reviewers?

Participating researchers, matched to topic and method. They sign their reviews and are paid per review. If two suitable reviewers cannot be found, the fee is returned.

What happens when authors disagree with a review?

The response is published alongside the review. If a review is considered unfair rather than critical, a third person appointed by the editorial coordinator examines it.

The Media and Information Research Commons is run by the Media and Journalism Research Center. Contact: mjrc@journalismresearch.org.

Text on this page, the registration template, the review checklist and the AI protocol are released under CC BY 4.0. Published records are released under the licence chosen by their authors, CC BY by default.