Just Play Already uses a recommendation technology: it suggests a game based on information about your preferences. This page explains how it works.
1. What the technology does
It suggests one game from your own library, among the games whose status is “Backlog”. A game’s card also shows similar games: some from your library, and some from the catalog that you do not have. They are picked by how alike they are to that game; your data only leaves things out: the catalog list skips games you already have and games for devices outside your profile (an empty profile allows any).
2. Processes and methods
- Collecting. Before a pick you answer a few questions: how much time you have, whether you want to think hard, what vibe you want, who you play with, whether you had a tough day, which platform. Questions whose answer is already clear from your library may be skipped.
- Filtering. First, games that fail a hard condition are set aside:
- platform: you do not have the game on the chosen platform, or on an older one whose games it also runs (a PS4 copy counts for a PS5);
- “Multiplayer”: neither the game’s modes from IGDB nor a manual check by the service say it can be played with others;
- “Single-player”: the game’s modes from IGDB say it can only be played with others;
- “With a friend on one screen”: the game fails the “Multiplayer” condition, or it is not known that two can play it on one screen;
- “Just for tonight”: it is not known that the game is short (about four hours in all) or easy to play in short sessions.
- Scoring. Each remaining game gets a score by fixed rules: how well its mood and pace match your answers, how much it resembles games you liked, whether you have recently had enough of similar games, whether its length fits, and how long it has been in your library.
- Choosing. Usually the service takes the three best-scoring games and picks one of them at random: the higher the score, the better the chance. In about one pick in ten, so you do not see the same thing again and again, it picks at random instead, favouring games it has offered less often before. If the best options are too close to recent ones, it may ask what you are in the mood for. In the “I don’t care, just pick” mode the game is picked at random among your “Backlog” games — from those you have not turned down with “Another option” in the last 7 days, or from all of them if there are none.
- Learning. The reactions “Launched!” and “Another option” in a regular pick are saved together with how well the game matched your answers. From them the service recalculates the weights your preferences get; the recalculation runs separately, not at the moment of a pick. Answering “Already played” with a rating changes the game’s status and rating in your library, and ratings affect which games count as similar to ones you liked.
- No neural network takes part at the moment of the pick: the rules are fixed, and the explanation of why a game was suggested is built from the same rules. The descriptions of the games themselves (genres, tags, mood and pace ratings) are prepared in advance, partly with the help of a neural network — it rates the games, not you.
3. What information about your preferences is used
- your answers before a pick;
- your library: games, devices, statuses and ratings;
- your reactions to suggested games and your recommendation history;
- your device profile, if you filled it in;
- preference weights computed from your reactions.
4. Where this information comes from
- from you: answers, reactions, the games you added, ratings and statuses;
- from Steam or PlayStation: the list of your games, if you started an import yourself;
- from the IGDB and Steam catalogs: information about games (not about you).
5. How to influence the recommendations
Answer the questions differently, react to suggested games, change the statuses and ratings of games in your library, and your device profile. Deleting your account deletes all the information recommendations are built on.
6. Owner of the service
S. A. Vilgodskii. Email for legally significant messages: justplayalreadyapp@gmail.com. How we handle data is described in the Privacy Policy.