How We Decide Which Games Get Published

There are tens of thousands of HTML5 games licensed for web publishing, and the licence terms make it trivially easy to publish all of them. Sign up with a distributor, pull the feed, generate a page per title, and you have a site with a hundred thousand words on it by the end of the afternoon. Almost every browser game portal you have ever landed on was built exactly that way, which is why they all feel the same.

We decided not to do that, and this post explains what we do instead.

The catalogue is capped on purpose

Our published list sits at roughly a hundred and sixty games. That number is not a technical limit and it is not a stage we are passing through on the way to something larger. It is a cap, and it exists because somebody here has to actually play a game before it gets a page, and there are only so many hours.

The obvious objection is that a bigger library is more useful. We think the opposite is true past a certain point. A shelf of a thousand games is not a library, it is a search box, and a search box is only useful if you already know the name of what you want. If you do not — if you have ten minutes and a vague preference for something with numbers in it — a hundred and sixty games that someone has sorted is far more use than a thousand nobody has looked at.

What earns a game a page

Four things, roughly in this order.

It has to work. Not “the frame loads” — actually work. We open it, play it for as long as it takes to understand the loop, and check that the thing it claims to do happens. A surprising share of catalogue games fail here, usually with a menu button that does nothing or a level two that never arrives.

It has to be honest about its controls. Some of the best games on the site are keyboard-only, and that is completely fine. What is not fine is a game that presents a touch interface and then requires a key press. Where a game does not work on a phone, our write-up says so in the controls section, so nobody discovers it four minutes into a bus journey.

It has to bring something the shelf does not already have. We publish two liquid-sorting puzzles, Water Sort and Ball Sort, because they are genuinely different games — one has no undo and no warning when a board dies, the other hands you an undo button and an extra bottle and asks you to be elegant instead. We do not publish a third, because a third would just be the same afternoon again.

It has to have a moment. Something we could describe to a person who has not played it. Bloxorz has the parity problem where you keep arriving at the hole lying flat. Grand Mahjong Connect has the holding bar that turns a calm tile game into inventory management. Arcade GP has the moment where turn-in goes vague and the tyres tell you to pit. If we play a game for twenty minutes and cannot name one thing about it, that is the answer.

What gets a game cut

Broken games go, obviously, and we remove them rather than leaving a dead page behind. Beyond that, the two common reasons are advertising and duplication.

Some catalogue games carry so much of their own advertising that the game is the smaller part of the experience. We cannot change what happens inside a third-party frame, so the only editorial control we have is not publishing it. And where two games are clearly the same product with different art, we keep whichever one is better and drop the other, even when both work fine.

Every page is written here

This is the part that costs the most and matters the most. Distributors supply a description with each game, and that description is shared word for word with every other site carrying the title. Publishing it means your page is a duplicate of several hundred others, and it also means the text is useless — it is marketing copy written before anyone played the finished thing.

So each published game carries a write-up we produced ourselves: what it actually is, how the first few minutes go, which controls you need, and two or three things we learned that make the next attempt better. The tips are the bit we care about. Anyone can restate a genre; noticing that pushing a crate flush against a wall in Sokoban locks it into one dimension takes a session of losing.

The checks that run before anything ships

Editorial intentions are easy to state and easy to quietly abandon, so some of ours are enforced by the build. A game with no write-up cannot be exported to the site at all — the build fails rather than falling back to the distributor’s text. Every guide has a minimum and a maximum length, so nobody can pad one out or leave a stub. And a similarity check refuses to ship if the same long run of words appears across three or more guides, because mechanically reworded text is exactly the thing that makes a site look mass-produced.

None of that guarantees good writing. It only guarantees that when we get lazy, the site stops building instead of quietly getting worse.

Tell us when we are wrong

If a game here is broken, or you think something on the shelf does not deserve its place, we would rather hear it. There is a note about how we work on the about page, and the contact page has the address. Suggestions for games we have missed are welcome too, though the honest answer is that most of them will not make it — that is rather the point.