Editorial Policy

Arcanum reviews AI RPG platforms and also sells placement and design services to companies in the same market. This page sets out how those two activities are kept separate, what we disclose, how the site is written, and how we handle mistakes.

It covers editorial conduct rather than scoring. For the rubric itself — what each score measures, how it is calculated, and what makes one change — see how we test and rate.

Independence

Ratings are never for sale. Arcanum offers paid placement and paid design services, and all of it is walled off from editorial judgment:

• Reviews, ratings, and directory descriptions are never for sale and never influenced by a commercial relationship.
• Anything paid is visibly labeled — "Featured" on placements, "Sponsored" on announcements.
• A paid placement never re-ranks the directory and never implies a score. An unrated platform stays unrated inside a slot it paid for.
• Comped access, free subscriptions, and paid design work are disclosed in any coverage they touch. Free access gets a platform tested sooner; it never gets it a better score.

The full commercial terms are on the developers page.

Disclosure and conflicts of interest

Disclosures appear on the individual piece, not only on this page.

Comped access is a material connection even when no money changes hands. Any review, first look, or interview involving a free account, an extended trial, or a comped subscription says so on the page, in the same place every time.

We are also a creator on some of the products we cover. Arcanum builds and publishes its own games, and some of them run on third-party platforms that appear in our directory — which makes their author a customer and a reviewer at the same time. Where that is true, the piece says it outright rather than leaving it for someone to discover. The reverse case is covered by policy instead of disclosure: our own games are never scored and never ranked against third-party products, because a score we award ourselves on our own scale is not something a reader should have to take on faith.

If you spot coverage that should carry a disclosure and doesn't, that is a correction and we want to hear about it.

How this site is written

Arcanum covers AI products, so we state directly how our own content is produced.

The direction and the substance of a piece are the author’s. What an article argues, the reasoning that supports each claim, the first-hand findings it is built on, and the information that decides the angle it is written from all come from the person who tested the product. That is the editorial work, and it is not delegated.

AI assists with drafting and research inside that direction — turning notes, arguments and findings into finished prose, and helping check claims against sources. Every published page is reviewed before it goes up, and factual claims are verified rather than carried over from a draft; see what we verify below.

The play, the testing and the verdict are never generated. A score exists only after a person has actually played the product and formed a view of it. No rating on this site was produced by asking a model what it thought of something, and no review describes a session that did not happen.

In short: the argument and the findings are the author’s. AI helps write them up.

What we verify before publishing

Errors in this category are usually factual rather than editorial: outdated pricing, a feature that shipped differently, a model name that changed without notice. We apply the following standards before publishing.

• Prices and tiers are stated only when we can verify them on the vendor's own site or documentation. If pricing is not public, we say it is not public rather than estimating it.
• Features are described from use, not from a marketing page. Where we are reporting a claim we could not test, the page says it is a claim.
• Model rosters are dated, because they change without announcement.
• Every rated entry carries a last-tested date. A score with no date is a score with no meaning.

When a published fact turns out to be wrong, see corrections below.

Corrections

We get things wrong. When we do, we fix the page and say what changed rather than editing quietly — a factual error, a misread feature, a price that was wrong. If you've spotted one, email [email protected] and we'll check it.

When a correction affects a claim that appears on more than one page, we correct every instance, not only the page that was reported.

Disagreeing with a verdict isn't a correction, and we won't pretend otherwise. But if the reasoning behind a score doesn't survive contact with your experience of the same product, we want to hear it.

Dated pieces stay dated

Interviews, first looks, and monthly notes are records of a moment. When the field moves past one, we add a dated note rather than revising the original. The figures, verdicts and framing stay as published, so the piece remains an accurate record of what we said at the time.

Reviews are the exception, because a review is a live verdict rather than a record. When a re-test moves a score, the review is rewritten and carries a dated note saying what it used to score and why the number moved.

Who writes and tests

All of it is Rukka — four decades across tabletop RPGs, CRPGs, MMOs, gamebooks, and AI-driven roleplay systems, and the designer behind the Arcanum Originals.

A single tester is a stated limitation of this publication. It constrains how much we can cover and how quickly, and it means every verdict reflects one person’s judgment. The rubric exists partly to offset that, by applying the same probes in the same order to every product, and where the method is weakest names the class of product it handles least well.

Version

Editorial policy v1.0 — 5 September 2026

When these commitments change, the version changes with them and the change is noted here. The scoring framework is versioned separately on the methodology page.