If anything we publish contradicts what follows, the piece is wrong and we will amend it.
A review or essay appears only after the writer has finished the game. For work with credits, that means the credits. For open-ended games, it means the point at which the writer can demonstrate — in the play log — that the loop has stopped producing new information, and that point is always stated as an hour count rather than implied.
We do not publish on embargo dates as a matter of course, and we do not treat being early as a virtue. A piece is filed one week after the writer finishes, which means our coverage of a large release frequently arrives a month after everybody else's. That delay is the product, not a shortcoming of it.
Every scored piece carries a condensed play log: hours to completion, difficulty setting, patch or build version, number of runs, and where relevant the count of crashes and technical failures. A score published without that context is an assertion; with it, a reader can decide how much weight the number deserves.
Scores run from one to ten with one decimal place. There is no formula, no category weighting and no separate marks for graphics or sound. The number summarises a judgement that the essay is responsible for justifying.
Price does not enter the score. Value depends on how much time a reader has, which is information we do not possess. Where price changes the recommendation, the essay says so in words.
Unfinished software is described, not scored. Early access builds, open betas and pre-release versions receive written coverage naming the exact build, because a number attached to something openly incomplete misleads readers. Competitive multiplayer games without an ending are covered as essays. Games we abandoned are published as a play log explaining the abandonment, with no score attached.
A score is reopened in three circumstances: a substantial update alters something the original piece criticised; a technical fix addresses performance we described; or we concluded we were wrong. In all cases the original number remains visible beside the new one, with the date of the change and a paragraph explaining it. Two scores have been revised across forty-one issues.
Where a game was provided by its maker, that is stated on the piece. The majority of what we cover is bought at retail. No coverage is arranged in exchange for access, no piece is sponsored, and we carry no advertising, affiliate arrangements or paid placements of any kind. Nothing on this site is for sale.
Writers do not cover games made by anybody they have a working or personal relationship with. Where such an overlap exists, a different writer takes the piece and the relationship is disclosed on it.
We write about endings, because an ending is frequently the most interesting thing a game does and criticism that cannot discuss it is criticism with its hands tied. Where a piece intends to describe a late-game event, it says so before doing it, and we describe the shape of a conclusion rather than its specific content wherever the argument permits.
Factual errors are corrected with a visible note at the point of the error rather than silently overwritten. If a correction materially changes an argument, the note says which paragraph changed and what it previously claimed. Amending the record quietly is, in our view, worse than the original mistake.
This publication collects nothing from the people who read it. There is no account system, no comment section, no newsletter and no form of any kind. There is no analytics platform and no advertising technology on any page. The privacy notice describes this in full.
Quote us freely in discussion and commentary with attribution. Scores may not be reproduced in advertising or on store pages, because a number separated from the essay that justifies it is a misrepresentation of what we published.
