About Cold Open
A small independent publication about games, run by five people, funded by advertising and readers, and accountable for everything it puts its name to.
What we are for
Cold Open exists because most games coverage is built around the release calendar, and the release calendar is a marketing artefact. A game is most interesting to write about somewhere between the second week and the second year, once the launch noise has cleared and it is possible to tell the difference between a novel idea and a novelty.
So we are slow on purpose. We publish about six reviews and sixteen news pieces a month, every one of them reported or played to completion by a named person. We do not run previews, we do not cover leaks, and we have never published a review of a game that a member of staff had not finished. That is a smaller output than a site of this kind would normally have, and it is the entire point.
We started in March 2021 with two people and a shared spreadsheet of games nobody had written about properly. There are five of us now. The spreadsheet still exists.
Review methodology
Every review is scored out of 100 using the same six criteria. What changes between genres is the weighting, which is chosen before anybody starts playing and is printed on the review itself. Frame rate stability is worth more in a twitch platformer than in a turn-based strategy game, and pretending otherwise would make the number meaningless.
The six criteria
| Criterion | Default weight | What it measures |
|---|---|---|
| Systems design | 25% | Whether the rules are coherent, legible and worth learning. |
| Pacing and structure | 20% | How the game spends your time across its full length. |
| Interface clarity | 15% | Whether the game communicates its own state without a wiki. |
| Writing and tone | 15% | Dialogue, framing, and whether the game means what it says. |
| Technical performance | 15% | Frame stability, load behaviour and defect rate as shipped. |
| Accessibility | 10% | Options present by default, not buried or added post-launch. |
What the bands mean
90–100 Exceptional. We expect to still be recommending it in five years.
80–89 Strong. Clear recommendation with named reservations.
70–79 Good, with at least one structural problem we could not talk ourselves out of.
60–69 Worthwhile for the right person, frustrating for everyone else.
Below 60 Not recommended at any price.
The rules we hold ourselves to
- Every game is bought at retail price with our own money. We do not accept review copies, hardware samples, paid travel or event hospitality.
- The reviewer finishes the game, or plays far enough past the point of diminishing returns that finishing could not change the argument. Hours played is printed on every review.
- Two people who did not write the review argue with the score before it publishes. They cannot change the text; they can force the writer to defend the number.
- Scores are never revised quietly. If a game changes materially we re-review it, date the revision, and leave the original visible.
- No advertiser has ever seen a review before publication, and no advertiser can buy placement next to specific editorial.
Who writes here
Five people, all of whom are credited by name on everything they write. If you want to reach any of them directly, the newsroom address on our contact page reaches all five.
Rhea Vance
Editor-in-chief
Co-founded the site in 2021. Writes about studio economics and how the work gets organised. Previously spent nine years in production at two mid-size studios.
Tomás Ibarra
Reviews editor
Owns the rubric and the argument about the rubric. Drawn to systems games, logistics and anything with a spreadsheet underneath it.
Nadia Kohl
News editor
Reports on accessibility standards and the business of publishing. Responsible for the measurement work behind our subtitle and captioning coverage.
Sam Oyelaran
Guides editor
Writes and re-checks every guide on the site. Insists that a guide is not finished until someone who has never done the task can follow it unaided.
Priya Raman
Hardware correspondent
Covers supply chains, repairability and input hardware. Built and published the latency measurement rig we use for hardware testing.
The portraits above are abstract marks rather than photographs. We would rather identify our writers by name and by the work than by a headshot.
Ownership and funding
Cold Open is published by Cold Open Media Ltd, a company registered in England and Wales. The company is owned entirely by its two founders; no publisher, platform holder, hardware manufacturer, games company or investment fund holds any stake in it, and none has ever been offered one.
The site is funded from two sources. The first is display advertising, sold programmatically and served by third-party ad technology including Google. Advertising is bought against the site as a whole and against broad sections; it cannot be bought against a named game, a named company or a specific article, and the sales side has no visibility of unpublished editorial. The second is direct reader contributions, which currently cover a little under half of what we spend.
We do not run sponsored posts, native advertising, affiliate links, or paid review coverage of any kind. If that ever changes, it will be disclosed here first and labelled on every affected page. Our full advertising disclosure sits in the privacy and cookie policy, alongside the details of what advertising cookies do and how to opt out of them.
We take no money from any company whose products we review, in any form, including conference sponsorship and speaking fees.
Corrections log
Every correction we have issued, oldest last. Corrections are also attached to the top of the article they concern.
- — Our piece on studio closures described a team as having forty-two staff at the time of closure. The correct figure was thirty-eight. The wider argument was unaffected.
- — A guide to frame pacing recommended a setting that had been removed from the relevant menu the previous November. The step was rewritten and the guide re-dated.
- — A review stated a game had no offline mode. It has one, behind a menu we did not find. The score was recalculated and rose by two points.
- — We misspelled a developer name throughout an interview. Corrected, and we apologised directly.
If you think something here is wrong, please tell us. We would rather be corrected by a reader than be wrong in public indefinitely.