Today, Valve announced a new subset of features for Steam called Steam Labs. Like a public test realm but for Steam, Steam Labs is a location for players to try out some of the Steam store's work-in-progress features. To accompany its launch, Valve has released three such features for users to test and give feedback on: Micro Trailers, Interactive Recommender, and Automatic Show.
"Experiment 001" is Micro Trailers. Micro Trailers are a collection of six-second game trailers organized by category, like RPG, Roguelike, Popular New Games, etc. You can choose to have these trailers presented to you in a variety of ways, such as one at a time, entire rows at a time, automatically with scrolling, etc. They are meant as a way to quickly digest gameplay for numerous games at a time.
Experiment 002 is the Interactive Recommender, a machine-learning neural network that recommends games based on your the play time of games in your library. The network "disregards most of the usual data about a game, like genre or price point" and instead uses information gathered from across Steam to make suggestions based on what other people with similar gaming habits to you have played. Users can refine those results by including or excluding certain tags, searching for titles that released within a specified time frame, and adjusting a niche vs mainstream slider. Developers can also see how much traffic is driven to their store page from the Interactive Recommender.
Experiment 003 is the Automatic Show. The Automatic Show is an algorithm that stitches together trailers for Steam's current top games into a roughly half-hour showcase. The showcase is divided into categories like top 20 new and popular, top 20 indie, top 10 card games, etc., and appears to grab both new and older titles for display. If successful, Valve intends to turn the experiment into a daily or weekly show.
There seems to be a common thread among each of the Steam Labs experiments and that's discoverability. A common complaint among users is that Steam's massive audience and gigantic reach also comes with a daily tidal wave of new titles to sift through, making games that appeal to an individual user increasingly difficult to find. These experimental features all attempt to address this issue in a different way. It is important to note though that these are all considered work-in-progress features and there is no guarantee on when—or even if—any of them will see full implementation in the store. To smooth the feedback process along, however, Valve invites users to discuss and make suggestions on these new features in the new Steam Labs group page.