Established 2021 · Sheaf Street, Leeds
Most reviews are written by whoever happened to be free that week, and the reader never learns whether anyone else at the desk saw it the same way. We do the opposite. Every computer game that comes through Split Screen is played by two of us, on the same build, in the same fortnight, on machines a normal household might actually own.
Then we write separately, publish both columns untouched, and add a short block naming the exact point where we stopped agreeing. Some weeks the two halves nearly match. Some weeks they read like they belong to different games. Both outcomes are useful, and neither gets flattened into a single number.
This fortnight
A tactical squad game with a generous tutorial and a brutal second act. Two of us finished it. Neither recognised the other’s description of the difficulty curve.
The second act is where this one finally trusts you. Everything before it is scaffolding: safe maps, forgiving reinforcement timers, enemies that telegraph their intent a full second early. Once the campaign moves north, that scaffolding comes away in one piece and you are left holding a squad you have quietly overbuilt in one direction.
I lost three missions in a row and enjoyed every one of them, because each loss pointed at a specific decision I had made eleven hours earlier. That is rare. Most games of this shape punish you for something invisible.
My reservation is the writing between missions, which keeps introducing characters it has no room to develop. I started skipping it by hour twenty and lost nothing.
I do not think the second act is a step up. I think it is a different game wearing the first one’s clothes. The forgiving early hours teach you to plan carefully and rewards patience; then the north campaign starts scoring you on reaction speed and punishing the exact habits it spent eleven hours building.
Nadia read her three failures as feedback. I read mine as the game changing its mind about what it wanted. Neither of us is wrong about what happened on screen — we simply disagree about whether it was designed or accidental.
On the between-mission writing we agree completely, which is the only thing we agreed on all fortnight.
Where we split
Nadia treats the difficulty shift as the point of the campaign; Callum treats it as a design fault that undoes the first act’s teaching. We replayed the same two missions together and still came away with different readings, so we are publishing both and calling it unresolved.
Recent write-ups
Each entry below links through to the full pair of columns. The chip on the right tells you whether the desk landed together, apart, or nowhere at all.
Played entirely in the same room, which is the only way we could work out why this one falls apart in public lobbies. The design assumes a partner who will interrupt you — and that assumption is load-bearing.
A logic game with a three-tier nudge system. One of us used it twice in nine hours. The other refused it entirely and spent nineteen. Same game, wildly different shape by the end.
Where we split
Whether an optional aid counts as part of the design or as an admission that the design failed. We are not going to settle that one this year.
A deliberately narrow role-playing game that keeps its world at village scale for the entire run. We both went the distance and both came out slightly changed about what scale is for.
Forty hours of squad tables against twelve hours of playing the matches out. Two entirely legitimate ways to hold this game, and the write-ups reflect that split without resolving it.
The three outcomes
A number out of ten hides the interesting part. These three labels do the opposite — they tell you how much confidence to place in the write-up before you read a word of it.
| Label | What it means at our desk | How to read it |
|---|---|---|
| Consensus | Both write-ups arrived at the same conclusion about who the game is for, independently, before either of us read the other. | Reasonably safe ground. If the description sounds like your kind of thing, it probably is. |
| Split | We agree on what the game does and disagree about whether that is a virtue. The disagreement is named explicitly in the write-up. | Read the block marked “where we split” first. Whichever side sounds like you is your answer. |
| Stalemate | We could not find common ground on what the game even is, usually because we played it in incompatible ways. | Treat both columns as separate accounts of separate experiences. Useful, but not a recommendation. |
Standing rules
Five conditions. If any one of them fails, the write-up waits or gets dropped. They are dull on purpose.
A game with only one player at the desk does not become an article. It goes into the queue until a second person has time for it, even if that means missing a release window by two months.
Both players run the same patch version and finish within fourteen days of each other. Comparing a launch-week experience against a year-two experience produces a false disagreement, and we have made that mistake before.
Neither column is shared until both are finished. We used to discuss games over lunch while drafting and the write-ups quietly converged. Now the drafts go into a locked folder until both are in.
“We had different feelings about it” is not a disagreement. The block has to point at a specific mechanic, act, or design decision, or the verdict gets downgraded to a stalemate.
Every write-up says which machine it was played on, at the foot of the page. Both our desks are mid-range and several years old, which is the point.
Reader questions
Because a single voice can only tell you what one person values, and you have no way of knowing how idiosyncratic that is. Two voices from the same desk, playing the same build, give you a rough error bar. When both land in the same place, that means something. When they do not, you learn which axis the game divides people along, which is often more useful than a recommendation.
No. We copy-edit for clarity, spelling and length, and we will ask a writer to add a sentence if a claim is unclear. We never ask anyone to soften a position because the other column contradicts it. If a factual error turns up after publication, the correction is added at the foot of the page with the date, and the original sentence stays visible.
Between three and seven weeks, depending on length. Two people finding forty hours each for the same game is the bottleneck, not the writing. This is why we publish two or three pairs a month rather than a review a day, and why you will find games here that came out well before we got to them.
Yes, and reader suggestions account for roughly a third of what we cover. Use the form on the talk back page and say why you think the game would divide a pair of players. Suggestions that come with a specific argument get read first, because they tell us what to look for.
