If you’ve had your Steam account for any length of time, you know how it goes – through some combination of uncautious sale purchases, big bundles, and free giveaways, you accumulate a massive pile of the best PC games that will haunt you until the day you die. But Valve’s relieving a little of the backlog pressure with a new Steam Labs experiment that’ll tell you what game to play next.
When you’re debating what Steam game to play, Valve says “our machine learning system helps you to choose, by suggesting the games it thinks you’ll enjoy most, among games you already own.” It’s based on the same tech as the Interactive Recommender, and appears to largely be based on your playtime in similar titles.
The Play Next system will return a list of recommended games, three at a time, and you can page through a looping series of recommendations. You can hover over the game for a quick version of the trailer, or click a button to see the game directly in your library.
It’s a neat, randomised way to get some options for what to play next, though it’s got some limitations. You can’t select games to ignore and reroll, for example. So if Steam’s recommending a bunch of games you’ve already played on other platforms – like it’s doing for me – you can’t get a fresh selection of titles.
Introducing Steam Labs Experiment 008: Play Next! 🔬
Got a bunch of unplayed games in your Steam library, unsure what to play next? 🤔
Let the Interactive Recommender system help you choose by predicting which of your games you'll enjoy the most!https://t.co/VrD1ZJgrYd
— Steam (@Steam) February 13, 2020
As with everything in the Steam Experiment line, it’s a work-in-progress that could get new features before it actually hits the library – or dies completely, depending on how the community responds.