In the ongoing, never-ending effort to make user reviews useful, rather than merely a conduit for rage and silliness, Valve has added a new feature to Steam that it hopes will make reviews more helpful for people who actually want to know more about the games they're thinking about buying.
Under the old regime, Steam users could vote on whether or not a review was helpful, and Steam would sort them accordingly, prioritizing those with high helpfulness ratings. The problem, as usual, is the Boaty McBoatface syndrome: When you let people on the internet vote on things, you're going to end up with shenanigans.
«We've seen that many players use reviews for sharing jokes, memes, ASCII art and other content that might not be the most helpful for a potential purchaser,» Valve wrote, explaining the new helpfulness system, which is literally called the Helpfulness System. «That content is usually fine, and often a lot of fun for existing customers of a game, but it doesn't always help new players in making informed purchasing decisions.»
The Helpfulness System will deprioritize user reviews identified as unhelpful, including «one-word reviews, reviews comprised of ASCII art, or reviews that are primarily playful memes and in-jokes.» Those reviews will still be viewable, but they'll now be sorted behind reviews that are deemed to be actually informative. The new system will not impact the generation of review scores—just the order in which user reviews appear on store pages.
The obvious question here, as the Insane Clown Posse once mused, is how does it work? As usual, it's all a bit opaque: Valve said helpfulness categorization «is a mix of techniques, including user reports, the Steam moderation team looking closely at a lot of reviews, and some machine learning algorithms to help scale the human judgment calls,» made necessary by the fact that «well over 140 million user reviews» have been posted on Steam. Interestingly, marking a review as helpful or not is still «taken
Read more on pcgamer.com