How we test every game before it goes live
Our publishing checklist has fourteen items. Three of them catch almost every bug we ship, and none of them is automated.
Every game on this site goes through the same fourteen-point checklist before it appears to a visitor. Automated testing catches syntax errors; it does not catch a game that is technically working and simply not fun, which is the failure mode that matters.
The first check is the ten-second test. Load the game on a mid-range phone, start it, and see whether a first-time player can make a meaningful action within ten seconds without instruction. If they cannot, the game gets a simpler opening or an on-screen hint, or it does not ship.
The second check is the failure test. Deliberately lose. Is the failure state readable? Does the player understand what they did wrong? A game where losing feels arbitrary gets reworked, because a player who does not know why they lost will not try again.
The third check is the second-run test. Play once, stop, and consider whether you want to start again immediately. This is the only test that cannot be faked, and it is the reason several finished games never appeared on the site.
We do not publish a game we would not play ourselves. It is a low bar in theory and a surprisingly high one in practice, and it is the whole editorial standard in one line.