Steam Labs it is a virtual place where Valve conducts "experiments" aimed at making Steam easier and more immediate to use. One of the most interesting experiments we have seen so far is the one called Interactive Recommender, which uses machine learning instead of conventional tags to recommend potential titles to the player.
The most recent experiment, Play Next, performs a similar function and is actually based on the same technology that powers the Interactive Recommender. But instead of scanning the store to find games that we might want to buy, analyze the library we already have to find games that we may have purchased but have never started.
The Play Next page lists three games from our library that according to the system we should start, each with a microtrailer, tags and a short list of similar games to give us an idea of why we may be interested. A "next choices" button will show a list of three different games. It is not clear if there is a limit to the recommendations or if the number of suggestions will vary on aspects such as the size of the library or something else.
Valve said the system is still in the making, so there may be major changes: “The algorithm we use to choose comparable games is a new system that we are experimenting with and is a work-in-progress. So don't be surprised if he sometimes shows you wrong or unexpected results. "