Guide · how the desk works
The crosscut method: how we cut through arcade lanes
A crosscut saw goes straight across the grain, so you can see what is inside. We do the same with browser builds. Here are the six steps we follow, written on 10 November 2026 so you can repeat them yourself.
We get asked how a small desk with no store data can rank anything. The honest answer is that we do not rank by numbers; we rank by what we saw, in the same order, every time. That sameness is the whole point of the crosscut: it keeps our opinions comparable even when the builds are completely different.
You do not need special tools to follow along. A phone, a keyboard and a notebook are enough, and the six steps below are written so that you can repeat them on any browser build you like.
-
1
Step one: pick a lane before judging arcade lanes
Every build belongs to a lane, meaning a single repeatable verb: hold and release, lean and hop, steer and draft. Before we play anything seriously, we write the lane down in one line. If we cannot say it in one line, the build is either brilliant or muddled, and the next steps tell us which.
Naming the lane first also protects us from a common bias. A glossy menu or a clever title can make a weak loop feel strong. The sentence we write beforehand is the thing we hold the build to afterwards.
-
2
Step two: the first cut across the arcade lanes
The first cut is the opening ten seconds, played cold. We do not read the controls beforehand. We watch for three things: whether the verb is understood without help, whether the first mistake teaches something, and whether the screen is readable on a small phone.
A build that passes this cut earns trust quickly. A build that fails it is not necessarily bad, but it is asking for patience, and we say so in the review. This cut is why a calmer build can still rank above a flashier one.
-
3
Step three: the second cut through the arcade lanes
The second cut is the long middle of a run. Here we look at how difficulty grows. Does speed climb gently? Do new hazards arrive one at a time, or all at once? Do the short events that interrupt a run feel like new ideas, or just louder versions of the same one?
We also test both input styles in this cut. Each build is played with touch and with a keyboard or mouse, and we write down which one felt more precise. Builds that only work well on one kind of device lose a little here, and we say that plainly in the review.
-
4
Step four: the third cut at the end of arcade lanes
The third cut is the ending. A good failure is readable: you know exactly what killed the run, and your first thought is what you would do differently. A bad failure feels like a coin flip or a camera error. We replay the ending a few times before deciding which one we are looking at.
This is the cut that decided the order of our current ranking. Glow Seine and Barrow Up both fail in ways that are easy to read. Herd Weave sometimes does not, which is why it sits third even though we enjoy it.
-
5
Step five: note what fails in these arcade lanes
After the three cuts, we write down what annoyed us. Not to be harsh, but because a review that lists only strengths is not much use. We try to name at least one limit per build, and we keep the wording specific: “gold is scarce at the start” helps a reader, “needs polish” does not.
We also record what we could not check. If a build has no public store rating, we say so. We never fill the gap with a made-up figure, and we never quote an install count we cannot source.
-
6
Step six: date the arcade lanes check and file it
The last step is to write the checked date next to every fact and file the check in the archive. Browser builds can change without notice, so a fact without a date is a rumour. Our current checks are all dated 10 November 2026.
When a build changes, we re-run the three cuts and add a new entry instead of editing the old one. You can read how the criteria fit together on the methodology page, or see the method applied in our three reviews.
Why this method suits arcade lanes
Short browser builds are easy to love for ten seconds and easy to forget after a minute. A crosscut forces us to look at all three parts: the opening, the middle and the ending. A build that shines only in one of them tells us something useful, and that is exactly what we write down.
The method also keeps us honest about gaps. When we cannot check something, we say so, rather than guess. It is slower than a quick star rating, but it leaves readers with facts they can verify for themselves, and every one of them carries a date.
We host browser builds for editorial review. We are not affiliated with unrelated stores or developers. Facts on this page are dated 10 November 2026.