• Valve has rolled out a new update changing the way user reviews on the platform work.
• The company is using a mixture of human moderators and machine learning to move informative reviews to the top and “funny” reviews to the bottom.
• However, if you like meme reviews you can revert the system back to the original.

A lot has been made of Steam user reviews in the last several years, with headline-grabbing stories popping up when popular games fall below the “Positive” rating and into “Mixed.” The way Steam handles its user reviews in games has long been a point of contention for players, and the owners of the online storefront have seemingly been listening.

In an official news post on Steam, Valve says that it is changing the ways user reviews work on the platform as part of a wider update. The popular system has garnered over 140 million reviews across the deep game catalogue available on the platform.

“We are ready for public testing of a new system that changes the way Steam sorts user reviews on store pages with the goal of prioritizing reviews that can best help players make a purchase decision about the game,” the company explains.

“This new helpfulness system is now enabled by default, and can easily be toggled within the user review settings for each game.”

The new review system will now prioritise reviews that are considered the most informative and helpful, and downplay reviews that are “identified as being unhelpful for potential customers, such as one-word reviews, reviews comprised of ASCII art, or reviews that are primarily playful memes and in-jokes.”

However, if users enjoy seeing meme reviews and in-jokes there is an option on the store page to revert the helpfulness system to the original one. “This change doesn’t impact how review scores are generated for each game; it is simply changing the order that reviews appear on each store page,” says Valve.

That then begs the question of how Steam decides which reviews or more informative compared to others. It says the new system will use a mix of techniques. This includes user report, a review moderation team at Valve looking closely at reviews and some “machine learning algorithms to help scale the human judgement calls.”

“Our team has found that a lot of the unhelpful reviews were easy to spot, so we’re targeting those first. This is a work in process, and will likely take quite a while for our team to evaluate the existing reviews and newly posted reviews,” Valve adds.

Users on the Steam subreddit have responded positively to the new updated method.

“What a blessing,” one user said, adding that the new system removed some of the award farming and over-use of memes on some big games like Grand Theft Auto V and Helldivers 2.

“Steam has been a joy to browse since this update,” said another. Some players are now calling for Valve to apply a similar method to Community Guides on Steam.