The Steam Interactive Recommender debuted last year as the second work-in-progress feature conceived in Steam Labs, a kind of digital skunkworks where Valve can develop and test experimental new features for Steam. The idea was to provide users with better, more personalized game recommendations through a "neural network model" based on play history and other data, rather than relying on Steam's tagging system.

Although it seemed fairly comprehensive at the time, Valve chose to separate it from the existing recommendation model in order to better train and test the system and eliminate any bugs that might appear. Today, the system is fully live on Steam: to access it, click on the "Your Store" menu button on the Steam front page, then click on the link at the top of the drop-down menu.

The release version of the interactive recommender has been adjusted somewhat from the first-look experiment. Recommendations are still weighted by popularity and can be filtered by age, but tag filters and exclusions are now searchable, there is an option to exclude games from your wish list, and you can save your recommendation settings.

With the default settings, the recommender will recommend playing The Fall Part 2: Unbound, Lake Ridden, The Signal From Tolva, Conarium, and Beautiful Desolation, in that order. Only the last game on the list is one I've been interested in for some time (although I have a vague recollection of being curious about "The Signal From Tolva"), and while I'm a bit suspicious of the whole thing, I'm also curious to see if the machine might be right. Because that's what this process is about: because that's the point of this process.

"The Interactive Recommender is not meant to replace our existing content discovery system, but to add a variety of ways for Steam to recommend games to players. It's a powerful tool, but there are some things it can't do," Valve explains. For example, we can't recommend new releases that no one has played yet.

"That said, we are starting to use the technology underlying Interactive Recommender for other Steam features like Steam Labs Experiment 008: Play Next recommends games that have been purchased but not yet played for whatever reason. games that have been purchased but not yet played for one reason or another. The result is a Steam experience that more effectively introduces customers to games they will like in a variety of scenarios.

Valve's Play Next effort, revealed in February, is similar to Interactive Recommender, but it works its magic on the existing Steam library, helping to do something about the ridiculous backlog it has ignored for years.