A few weeks ago, we posted a blog post describing some of our thinking about user reviews, some tools we added, and some of the next areas we were planning on addressing. If you haven't yet, it's worth a read to understand our thinking about reviews and the goals we think user reviews should be addressing.
Today we're making a few more improvements to user reviews to address some concerns and feedback that we've been hearing, and to ensure that the system remains a useful tool for you to identify games that you will enjoy. In this update, we're focusing on how we determine the helpful reviews you see on each game's page and the order in which they are shown.
What is a helpful review
Reviews should help paint a picture of what it's like to play the game and how well the game has been enjoyed by the people who have already played it. A good review typically describes some of the factors that directly impact the experience of playing the game, which can include a wide variety of things like how well the matchmaking works, how buggy the game is, or whether the game represents a good value for the price.
While everyone that has played a game probably has opinions on how much they enjoyed the game or not, some people are better than others at writing thoughtful descriptions that are useful for other potential players. We think it's fine for players to be able to indicate whether they liked a game or not, but not every review is useful to show in greater detail. A helpful review then is one that includes enough information to aid in your understanding of whether you are likely to find the game fun, or whether you should avoid the game and explore more alternatives.
As of this writing, there are over 36,579,839 reviews posted by players across all of Steam. Some games have a handful of user reviews while others have hundreds of thousands. Regardless of the number of reviews, we want to make sure that the most helpful and relevant reviews are the ones that you see first when you are looking at a game's page in the Steam store.
Sorting helpful reviews
Below each review you've probably seen the question of "did you find this review helpful?" along with a set of buttons to indicate yes or no. You can use these controls whether you have already purchased the game or not, because it is usually prior to purchase that you are looking at user reviews and considering which of them are helpful in making your purchase decision.
These inputs are then used to determine which reviews people generally find most helpful. Up to now, our system simply looked at how many people had rated each review as 'helpful' and how many people had rated it as 'not helpful' and then highlighted the ten reviews that the most percentage of people found helpful. Since games can change (sometimes dramatically) over time, we introduced a change a while ago that prioritizes showing recently-posted helpful reviews, as they are more likely to reflect the current state of the game. Of course these are just the defaults that are shown on a store page, and you can still easily browse all of the reviews for a particular game if you want to investigate further.
In a perfect world, people would truthfully mark a few reviews that were helpful for deciding to purchase or not purchase the game and we could use that data to directly determine the ten most helpful reviews. Alas, it turns out that not everyone is as helpful as we would like. Instead, we are seeing more and more feedback from players that the helpful reviews shown on store pages aren't representative of how well people are actually enjoying the game.
Taking a closer look
So we took a closer look at the patterns and behaviors of people that are rating reviews. Of the 11 million people that have used the helpful buttons, most follow a reasonable pattern of usage: Typical players rate a few reviews as helpful or unhelpful while deciding whether to make a purchase. However, we found a small set of users on the far extreme that are clearly trying to accomplish something quite different from normal players, and are rating more than 10,000 reviews as helpful or unhelpful on a single game. This behavior is not only humanly impossible, but definitely not a thoughtful indication of how 'helpful' each of those reviews were. These users also tend to rate up just the negative reviews while rating down the positive reviews (or vice-versa) in an attempt to distort which reviews are shown by default.
Because of how many reviews these users are rating, they each have a disproportionate amount of influence over the display of helpful reviews and cause certain reviews to appear more prominently than they should be. This can result in a confusing appearance where the default set of reviews shown are negative, even when most players have posted positive reviews and clearly enjoy the game.
What we are changing
Up to this point, we've tried to maintain just showing the simple math behind how we calculate whether a review is helpful or not--the percentage and number of people who indicated a review as helpful. We like systems that are transparent and easy to understand, as they are also easier to believe and trust. Unfortunately, this has resulted in a system that allows a small group to manipulate reviews to a degree that is clearly decreasing the value of Steam for many other players. So we're making two main changes.
Firstly, our system will use a new method of calculating the helpfulness of each review, taking into account the users that are trying to manipulate the system. One way we're doing that is by counting the helpful ratings on reviews differently for users that are far outside the norm. Ratings from users that follow normal patterns of rating will continue to be counted the same way that they have, whereas accounts that rate an excessive number of reviews on an individual game will see the weight of each individual rating count for less and less.
Secondly, store pages will now show the default helpful positive and negative reviews in a similar proportion to that of the overall review score for the game. For example, if the game is reviewed positively by 80% of reviewers, then the ten reviews shown by default on the store page will be 80% positive, showing eight positive and two negative. This should keep the reviews shown on a game's page from being so easily manipulated by a few determined players and should more accurately represent the overall sentiment of the people playing the game.
We're rolling out both of these changes as a beta today. You can turn the new method on and off to see how it impacts the default display of reviews on any given store page. Note that these changes only impact the default listing of reviews (Called "Summary") and the "Most Helpful" display option.
Still much more to do
There are still a number of further changes we're considering for the user review system. One thing we're looking at, is how review scores on games change over time as games develop or languish, and ways to better indicate how players are enjoying the game right now. We also want to better indicate when players are reviewing issues in a game that only pertain to a particular region (such as server locations), or particular language (such as poor translations). We're also exploring ways to calculate a personalized review score for each player and thinking about how that would look.
We know the review system is important for players and developers and we're going to continue making improvements to ensure that user reviews remain useful and trustworthy. Please let us know your thoughts on these latest changes.
For the past few months we've been busy working on significant improvements and additions to the Steam Curator system. There's still some work to be done before we can roll these out, but we wanted to share a bit about why we see Steam Curators as a crucial component to exploring Steam, and what changes we're making.
Why Steam Curators?
We've heard from many of you that you want to have a more curated experience when shopping Steam; where the titles that are surfaced and recommended and highlighted are picked by humans that you know and trust. But, we also know that players have different tastes in games, so it's unlikely that any single person or group could cater to the specific interests of every player in the world. This is why we believe that Valve can't be the only form of curation in Steam - we would be under serving the tastes and viewpoints of many players.
So, we're focusing on how to support the streamers, journalists, critics, content creators, writers, enthusiasts, and friends that you already know and trust to be able to help you find your next favorite game. By following a few Curators on Steam, you'll not only start to see their recommendations appear prominently when browsing the Steam Store, but you can also explore each of their customized spaces within Steam and see all the titles they have reviewed.
Using the Steam Curator features on Steam is an opt-in thing. If you’re not interested in the opinions of human beings helping you find games that are worth your attention, then we also have some powerful features coming just for you. We’re hard at work on significant improvements to the core recommendation engine which algorithmically suggests games for all Steam users. We’re anxious to talk in depth about that technology too, and will do so in a future blog post.
What changes are coming?
Over the three years since introduction of Steam Curators, we've gathered a lot of feedback from all kinds of perspectives. We've heard from players, from curators, from streamers, from game developers, and from all kinds of other tastemakers and content creators. The feedback is clear that the system needs to do a bunch of things better in order to work well for the three primary sets of people it's trying to serve: players, curators, and game developers.
This system really only works if players find value from following some Curators. So we're adding to the kinds of content that Curators are able to create, and increasing the places within Steam where that content can be seen.
Recommendations provided by Steam Curators can already appear in the main featured spot on your Steam Home page as well as in a dedicated space on your home page. We're building on this so that recommendations by Curators you follow will also show up at the top of tag and genre pages. This means as you explore, say the Free To Play page, you'll see recommendations from your Curators for Free to Play games. If you are browsing RPG games, you'll see RPG games featured from Curators you follow. And so forth.
Many Curators create videos to accompany their reviews, so we'll now start embedding those videos in a few places alongside the curation. This means that when you click through a recommendation, or when you browse a Curator's page on Steam, you'll be able to watch their videos in-line.
We also know that some Curators will review games within certain themes, genres, or franchises. So, we're adding a new feature for Curators to create lists of games they've reviewed that go together. These can be used to create lists such as "best couch co-op games", "games with amazing Workshop support", "games by my favorite designer", "10 games to play while waiting for Witcher 4", or any other set of interesting ways to organize groups of games.
And if you are looking to find new new Curators that share your tastes, or offer unique information about particular kinds of games, you can explore the 'Recommended Curators' or 'Top Curators' lists. We're fine-tuning the 'Recommended Curators' section to more accurately suggest Curators who recommend games like those you've been playing.
One of the pieces of feedback we received from Curators was that they felt it needed to be more rewarding and meaningful for a Curator to spend the time it takes to build and maintain their curation. So there are a few new things we're building to tackle this.
As we mentioned above, Curators that produce videos as part of their reviews will be see those videos embedded right next to their review in Steam. If you're a Curator who's already doing work to create content elsewhere, we want you to be able to use that work in your Steam curation. This means a few of the most popular video formats such as YouTube, nicovideo.jp, youku.com, and bilibili.com will appear right in Steam where players can easily watch them.
Curators will be able to customize and brand their home on Steam by selecting games, lists, and tags to feature and by uploading a personalized background.
We all know that graphs solve everything, so yes, we're adding more of them. In particular, Curators will be able to see how their reviews impacted their follower's behavior in the Steam store.
We are helping connect developers with Curators that are most likely to have relevant audience of followers for the developers' game. More on this below.
We've heard from many developers that they need a way of getting their game in front of Curators that have the right audience for that game, and to be able to do it in a way that is easy and secure. We've also heard from Curators that it can be a challenge to reach out to developers, who are often swamped with requests that they can't easily filter through. So we've built a whole new system that we are calling Curator Connect.
With Curator Connect, developers can search for appropriate Curators, and then send a copy of their game directly through Steam. We've added a number of tools for finding relevant Curators and for identifying the forms of social impact that Curator may have. To start with, developers will be able to search the listings of Steam Curators, narrowing results by name, OS, language, or tags that the Curator indicates they focus on. In the results, developers will be able to see a snapshot of each Curator, including follower counts and any linked social media accounts such as YouTube, Facebook, Twitter, and Twitch, which can help verify that the Curator is truly who they claim to be. The developer can then build a list of the Curators they wish to send their game to, include a message describing their game, and hit 'send'.
Curators can then browse a list of games that have been sent to them and can choose to accept or decline as they wish. Accepted games are added to that Curators Steam library to play and review. No need to mess with keys or e-mail.
Today we're starting a closed beta with a few dozen Steam Curators of different sizes, niches, and languages. This gives us an opportunity to gather feedback and suggestions from Curators and gives those Curators an opportunity to use the new tools to prepare and personalize their store pages ahead of full release. The Steam Curators that are invited to participate in the beta are free to share their thoughts publicly, so you may see some screenshots or write-ups from these Curators as they explore the new features and discuss them with the community.
We're aiming to run the beta for at least a couple weeks with just the Curators before releasing the update to everyone. Hopefully this blog post helps you understand what we're trying to do, and why, which we believe will help everyone to have a fruitful conversation.
As always, if you have any feedback or suggestions, please let us know.
We've been planning some changes to User Reviews for a while now, and we thought it'd be useful to give you some insight into our thinking on review bombing, which is one of the problems we're tackling. Review bombing is where players post a large number of reviews in a very compressed time frame, aimed at lowering the Review Score of a game. At the same time, they upvote each other's reviews and downvote all the other reviews. To understand why we think this is a problem at all, we first need to talk a bit about the goals of User Reviews, and the associated Review Score.
For User Reviews, the goal is fairly obvious - allow people who've played a game to tell potential purchasers whether or not they should buy the game, and why. When we first implemented reviews, we hoped their existence would be enough. But it became clear very quickly that many potential purchasers wanted a summary of some kind, so we created the Review Score. It shows you the ratio of positive to negative reviews for the people who've purchased the game, and from that, it has the goal of trying to provide a potential purchaser with an estimate of how likely it is that they'll be happy with that purchase.
So why is review bombing a problem? On the one hand, the players doing the bombing are fulfilling the goal of User Reviews - they're voicing their opinion as to why other people shouldn't buy the game. But one thing we've noticed is that the issue players are concerned about can often be outside the game itself. It might be that they're unhappy with something the developer has said online, or about choices the developer has made in the Steam version of their game relative to other platforms, or simply that they don't like the developer's political convictions. Many of these out-of-game issues aren't very relevant when it comes to the value of the game itself, but some of them are real reasons why a player may be unhappy with their purchase.
When it comes to the Review Score itself, however, it's even less clear that these out-of-game reasons are relevant. When we look at what happens with the Review Score after a review bomb, we see that it generally recovers, in some cases fully back to where it was beforehand. We took some time to examine the data more closely, measuring the weekly positive-to-negative ratio of new reviews in the time periods around the review bomb, it was even clearer - the review bomb ends up being a temporary distortion of the Review Score.
This implies that, while the review bombers were unhappy with a decision the developer made, the purchasers of the product afterwards were often as happy with the game as the players before them. In the cases where the Review Score didn't return fully to its prior level, we believe the issue behind the review bomb genuinely did affect the happiness of future purchasers of the game, and ended up being accurately reflected in the regular ongoing reviews submitted by new purchasers. In some review bomb cases, the developers made changes in response to the community dissatisfaction, and in others they didn't - but there didn't seem to be much correlation between whether they did and what happened to their Review Score afterwards.
In short, review bombs make it harder for the Review Score to achieve its goal of accurately representing the likelihood that you'd be happy with your purchase if you bought a game. We thought it would be good to fix that, if we could do it in a way that didn't stop players from being able to voice their opinions.
So what solutions did we explore? An obvious one would be to simply remove the Review Score. Then, as a potential customer you'd be forced to read the User Reviews themselves to see if the product sounded interesting. If you saw reviews talking about something outside the scope of the game, you could decide for yourself if it was an issue that would affect your happiness with the purchase. Unfortunately, we're pretty certain that this isn't really an option - scores were added in response to player demand in the past, and that demand for a summary of some kind is likely to still be there, even if players know it isn't always accurate.
Another idea was a temporary lock on reviewing, similar to how stock markets prevent trading on specific stocks when abnormal behavior is detected. Based on the theory that review bombs are temporary distortions, we could prevent reviews for short periods of time whenever we detect massive distortions in submissions. In the cases where the cause of the community's dissatisfaction truly affects the game's potential happiness to new customers, we're confident it would still result in the Review Score moving down after the lock period ended.
But if we lock reviews on a product for a short period of time, what does that mean exactly? Are players no longer able to post reviews at all during that time? Or should they be able to post them, but we ignore them for the purpose of calculating the Review Score? In the end, we didn't like the way this ultimately meant restricting the ability for players to voice their opinions. We don't want to stop the community having a discussion about the issue they're unhappy about, even though there are probably better places to have that conversation than in Steam User Reviews.
We could change the way the Review Score is calculated, focusing on much more recent data. One of the reasons a review bomb can distort a game's Review Score for a significant period of time is because the score is based on reviews over a period of 30 days for the Recent value, and all time for the Overall value. But doing this would likely result in more fluctuation and potentially less accuracy for all games, not just review bombed ones.
In the end, we decided not to change the ways that players can review games, and instead focused on how potential purchasers can explore the review data. Starting today, each game page now contains a histogram of the positive to negative ratio of reviews over the entire lifetime of the game, and by clicking on any part of the histogram you're able to read a sample of the reviews from that time period. As a potential purchaser, it's easy to spot temporary distortions in the reviews, to investigate why that distortion occurred, and decide for yourself whether it's something you care about. This approach has the advantage of never preventing anyone from submitting a review, but does require slightly more effort on the part of potential purchasers.
It also has the benefit of allowing you to see how a game's reviews have evolved over time, which is great for games that are operating as services. One subtlety that's not obvious at first is that most games slowly trend downwards over time, even if they haven't changed in any way. We think this makes sense when you realize that, generally speaking, earlier purchasers of a game are more likely to enjoy it than later purchasers. In the pool of players who are interested in a game, the ones who are more confident that they'll like the game will buy it first, so as time goes on the potential purchasers left are less and less certain that they'll like the game. So if you see a game's reviews trending up over time, it may be an even more powerful statement about the quality of work its developers are doing.
Hopefully this post has been useful. It's quite possible that we'll need to revisit this when we move to personalized review scores, where our prediction of your happiness with a purchase is based upon the games you've enjoyed in the past. In the meantime, we'll keep a close eye on the community conversation around reviews.
When we first created the Steam Community groups, the community itself was a smaller and simpler place. We wanted to give players on Steam an easy way to find other people with the same interests, coordinate their activities, and share their creative output. To that end, we made the tools and systems for joining and managing groups as simple as we could. When there were few great tools for discovering user groups, sending direct invites offered a way to spread awareness and build communities. But as the Steam Community has grown this approach to invites has not kept up with the needs of players. There are now many more ways to discover new and interesting groups and the invite system has been abused by more and more organized spammers.
So today we are changing the invite-and-join process for private groups, and taking steps to address spam.
Joining a group
In the past, you could create a group that is either publicly joinable by anyone, or private and accessible only by invitation. Private groups are used for a variety of purposes, some of which make sense to restrict only to invites. For example, if you just want a group for your close friends to hang out and chat, then an invite-only group makes sense. But there are many more scenarios where private groups have some form of stated criteria to join, such as “Steam Level 1,000 Club” or “Mod Makers Club”. Invites directly into these kinds of groups still make sense, but getting an invitation for qualified people is more complicated and messy than it needed to be.
So, we’re adding a new group type that will allow users to request to join the group. Requests are then put into a list for one of the group moderators to approve or deny. There are now three types of community groups:
Public group – Users can join on their own or via an invite from a friend.
Restricted group – Users can join by requesting membership. They can also receive an invite from a user that has permission to add members.
Closed group – This is the most restricted group. Users can only join via invites by members who have permission to send invites. Users cannot request to join these groups.
To join a restricted group, you can now request entry into the group. This will put you into a list for the group admins to review and decide whether to approve or deny your request.
Group admin tools
Any group that was previously a “Private group” will be converted to a “Restricted group”, which will allow users to request to join your group. This can be changed any time by the group owner. If your group is currently set to “Restricted” and you change the type to “Closed”, we’ll go ahead and decline any pending requests to join your group. Subsequently, if you change your group from "Restricted" to "Public" we'll approve all pending requests.
By default, only the group owner is able to approve/deny requests to join a restricted group. Group owners can allow officers or moderators to invite and approve/deny membership requests to a group via a new section under “Group Permissions”.
To manage pending group requests, group owners (or officers/moderators if given permission) can find the new Manage Group Requests page from their admin tools section on the right-hand side of the group page:
If you are invited to a restricted or closed group, you will skip the membership queue and become a member once you accept the invite. If you have been denied entry into a group, the only way for you to join in the future is to get an invite from a member who has invite permissions.
Limits to group invites
Over time, we have seen the use of group invites shift from helpful to spammy - to the point where they can become annoying and disruptive to players. As we looked close at what these invites are being used for, we found some interesting and troubling trends.
First off, we see that more and more organized spammers are using bots to create groups on a huge scale. At some times, the number of new groups created explicitly for spamming outweigh the legitimate groups. Once a spammer has created a bunch of new groups, they then use bots to invite random players into the group. Even if only a small percentage of players that were invited end up joining any one of these groups, the spammer still can end up with a significant audience. The spammers then use these groups to advertise various websites or offers by posting frequent announcements to the members.
Players are understandably annoyed by these spam invites from random users.
To address this, you can now only be invited into a group by your friends. This still maintains the common uses of forming a closed group to hang out with your friends, or creating a restricted group for people of similar interests or achievements.
These new features and improvements should make it easier to join groups that you are interested in while eliminating the spam group invites that show up in your Steam inbox. Over the next few weeks, we'll also be working on identifying and banning networks of groups that have been mass-created and exist for the sole purpose of spamming.
We're also working on improvements to the way that friend invites work to make it easier to connect with the people you know and give you more control over invites.
As we announced in our blog post last week, Steam Direct is now available! This new submission path is designed to provide a streamlined, transparent, and accessible route for new game developers from anywhere in the world to bring games to Steam.
The rest of this post will discuss some details regarding the closure of Steam Greenlight and provide some information for developers seeking distribution via Steam Direct.
Steam Direct information for developers
With the release of Steam Direct today, we have a couple of new pieces to the process. It's worth reading the full Steam Direct Overview page to see what the process looks like and what to expect. (Note: If you have non-game software, please see Distributing Non-Game Software for information on supported types of software)
In general, the process for developers of games and VR experiences involves a few key steps:
Digital paperwork. We need to know about the person and/or company that we will be doing business with. So the digital paperwork includes all the expected information such as company name, address, and contact information. There is also a brief tax and identity verification process that a developer will need to go through once to get set up.
The app fee. There is now a $100 recoupable app fee for each application to release on Steam. Steamworks developers will pay this fee once as part of the initial paperwork, which will unlock the first appID. Once all the paperwork has been completed, and the developer is set up in Steamworks, additional appIDs may be purchased for $100 each. This fee for each appID is returned in the payment period after that game has at least $1,000 in Steam store or in-app purchases.
Review processes. Building a release pipeline to support thousands of developers and millions of customers is a delicate balance. We specifically don't want an onerous and detailed certification process that makes it difficult for developers to release games, but we also want some level of confidence that games are configured correctly and aren't going to do unexpected things to customers' computers. So we have a couple of brief review periods where our team plays each game to check that it is configured correctly, matches the description provided on the store page, and doesn't contain malicious content. These processes shouldn't take more than a day or two unless we find something configured incorrectly or problematic.
Along with today's release of this new path, we're also rolling out an entirely overhauled documentation system to detail the Steamworks APIs, tools, features, and best practices. In this new documentation system, developers will find a much better organized layout, including an expandable table of contents, updated search, as well as an entirely new section on the Steamworks APIs themselves.
Additionally, we've been piloting a new help tool for Steamworks developers that have specific questions or that need help with the configuration of a game. This has been helping us provide better information and faster support to Steamworks developers.
The end of Greenlight
Last week, we disabled new submissions and voting via Greenlight and have been reviewing the remaining submissions. As of today, we've greenlit many of the remaining 3,400 titles that were remaining in Greenlight. There are a number of titles that we could not Greenlight, due to insufficient voter data or concerns about the game reported by voters. Titles that are not ultimately Greenlit may still be brought to Steam via Steam Direct, provided they meet our basic criteria of legality and appropriateness.
If you are a developer with a submission in Greenlight that was not Greenlit, you may be able to request a refund if you meet the following criteria:
If your current submission in Greenlight is your only submission, and it has not been Greenlit, or
If you have more than one submission and none of them have been Greenlit.
If one of these situations applies to you, please visit the Steam help site and select the "purchases" category to find your Greenlight submission fee purchase and request a refund.
Next Steps For Steam
With this transition to Steam Direct, we'll be keeping an eye on new submissions and making adjustments as necessary. We aren't quite sure whether there will be a lot more new submissions, just a bit more, or even fewer. It's most likely that there will be an initial surge of new submissions and then a new rate somewhat higher than what was coming through Greenlight.
Our analysis suggests that quite a bit of the previous volume of submissions to Greenlight was motivated by trading card abuse, which we detailed in our blog post Changes to Trading Cards. With the changes detailed in that blog post, we expect there is a category of game-shaped objects that are unlikely to be worth someone paying even $100 to bring to Steam. So that will likely lower the rate of incoming new titles somewhat. But, Steam Direct also intentionally provides a more transparent and predictable path for new game developers, which is something we heard held back many developers, especially in non-Western countries.
After Steam Direct has been up and running for a while, we'll share some analysis of what (if any) changes in volume of submissions or behavior of developers. We also appreciate the scrutiny and feedback from developers and players (such as Lars Doucet and Sergey Galyonkin) that keep holding us accountable, making smart suggestions, and digging into our changes because this whole wonderful platform exists to serve you.
We also have work well under way for more improvements and new features to expand and improve the Steam store.
We're in the progress of completing some major updates to the Steam Curator system as detailed in our previous post here. We're also quite a ways into rewriting the core of our recommendation engine to better predict which games any given user might find most exciting. And we're also in the process of updating various sections of the Steam store that haven't received as much recent attention as the home page.
Future blog posts in this space will detail our progress or announcements on these upcoming features and improvements, and we look forward to sharing more news soon.
Steam Greenlight launched on August 30, 2012, at a time when we realized that we weren't able to predict which titles players were really interested in. Up until that point, a small team here at Valve had been hand-picking games to invite on to the Steam platform, and almost every day we would hear from players wondering why awesome new game X wasn't available on Steam. The more this happened, the less confident we became that our own tastes were accurately representing the tastes of everyone using Steam. Greenlight was introduced as a way to help our team figure out which games players most wanted, by having those Steam users vote. Almost right away, we saw an incredible variety of games being submitted and voted on, which made it clear to us that there are far more distinct tastes and interests among Steam players than we had realized.
Right from the early days and throughout the life of Greenlight, we have been continually surprised by the hits coming through. In just the first year we saw titles such as War For The Overworld, Evoland, Rogue Legacy, and Verdun move through Greenlight and go on to become hugely successful. We found it was easy to explain afterwards why some titles turned out to be big hits, but when we forced ourselves to predict beforehand, we weren't nearly as accurate as we thought we were going to be. Those early years also saw huge growth in some categories of games that we had previously considered extremely niche, like visual novels. Whether you love or hate visual novels (In which case you can customize your preferences here!), they have gone on to form a huge following on Steam. Even today, we still see surprising smash hits come through Greenlight, such as the recent releases of Dead Cells and Blackwake.
Now, five years since Greenlight started, we've seen over 90 Million votes cast on submissions in Greenlight. Nearly 10 Million players have participated in voting in Steam Greenlight, but over 63 million gamers have played a game that came to Steam via Greenlight. These players have logged a combined 3.5 Billion hours of game time in Greenlight titles. Some of those titles, like The Forest, 7 Days to Die, and Stardew Valley, are in the list of top 100 selling games ever released on Steam.
With these kinds of successes, the thousands of niche titles, and everything in between, we realized that a direct and predictable submission process will best serve the diverse interests of players moving forward. So thanks to all of you who voted and played games in Greenlight, as we begin the transition to Steam Direct.
The information below on Greenlight and Steam Direct is going to be most relevant for game developers, as it discusses the nuts-and-bolts details of the transition.
As of now, we are no longer accepting new game or software submissions via Steam Greenlight and voting has been disabled. One week from today, on June 13th, we'll be turning on Steam Direct.
Over the next week, a team here at Valve will be reviewing the list of titles that have not yet been Greenlit and will be selecting the final batch of titles to pass through the Greenlight process. Our goal is to Greenlight as many of the remaining games as we have confidence in. There are some titles that will not be Greenlit, due to insufficient voter data or concerns about the game reported by voters. Titles that are not ultimately Greenlit may still be brought to Steam via Steam Direct, provided they meet our basic criteria of legality and appropriateness.
If you are a game developer with a game in Greenlight that hasn't been Greenlit yet, please be patient as we review the 3,400+ pending submissions. If you bought the Greenlight Submission fee, but haven't had a chance to post a submission, or if your submission has not been Greenlit by the end of this process, you can use the Steam support site to request a refund of your submission fee.
Steam Direct details
The goal with Steam Direct is to provide an understandable and predictable path for developers from anywhere in the world to bring their games to Steam. With that in mind, we're making the process as easy and streamlined as possible. A new developer will simply need to fill out some digital paperwork, including entering bank and tax information and going through a quick identity verification process. After completing the paperwork, the developer will be asked to pay a $100 recoupable fee for each game they wish to release on Steam. This fee is returned in the payment period after the game has sold $1,000.
As we have been doing for the past year, there is a short process prior to release where our review team installs each game to check that it is configured correctly, matches the description provided on the store page, and doesn't contain malicious content.
Additionally, brand-new developers that we haven't worked with before will need to wait 30 days from the time they pay the app fee until they can release their first game on Steam. This gives us time to review the developer's information and confirm that we know who we're doing business with. Developers will also need to put up a 'coming soon' page for a couple of weeks prior to release, which helps get more eyes on upcoming releases and gives players a chance to point out discrepancies that our team may not be able to catch.
Steam Direct will launch in one week, on June 13th.
In this post we want to talk about Steam Direct's publishing fee, and some additional changes we're making to improve the way the Store works. If you haven't read our first post, where we talked about the Steam Store algorithm, it's probably worth your time to nip back and read that first. That should help you better understand the reasoning behind the changes we're going to talk about in this post.
As you read, it's worth keeping in mind that there has been a subtle, but important shift in the way the Steam Store is designed. In the past, the challenge was to figure out what products should be on the Store. Now, we think the challenge is to figure out what products a specific player wants to see. There are many different kinds of players, with many different interests, so flexibility in how they view the Store seems like a requirement.
Determining the Steam Direct Fee
Back when we announced Steam Direct in February, we hadn't decided how much developers would need to pay to publish their games. We knew that we wanted it to be as small as possible to ensure it wasn't a barrier to beginning game developers, while also not being so small as to invite easy abuse by people looking to exploit our systems. We thought it would be great if the game community at large had a conversation about it, including both players and developers, which was why we chose not to highlight a specific amount in that original post.
Since then, we've seen a bunch of great conversations discussing the various pros and cons of whether there should be an amount, what that amount should be, ways that recouping could work, which developers would be helped or hurt, predictions for how the store would be affected, and many other facets to the decision. There were rational & convincing arguments made for both ends of the $100-$5000 spectrum we mentioned. Our internal thinking beforehand had us hovering around the $500 mark, but the community conversation really challenged us to justify why the fee wasn't as low as possible, and to think about what we could do to make a low fee work.
So in the end, we've decided we're going to aim for the lowest barrier to developers as possible, with a $100 recoupable publishing fee per game, while at the same time work on features designed to help the Store algorithm become better at helping you sift through games. We're going to look for specific places where human eyes can be injected into the Store algorithm, to ensure that it is working as intended, and to ensure it doesn't miss something interesting. We're also going to closely monitor the kinds of game submissions we're receiving, so that we're ready to implement more features like the the Trading Card changes we covered in the last blog post, which aim to reduce the financial incentives for bad actors to game the store algorithm.
We believe that if we inject human thinking into the Store algorithm, while at the same time increasing the transparency of its output, we'll have created a public process that will incrementally drive the Store to better serve everyone using it.
Upcoming Updates to Steam Curators
We know that some players really want to see a human involved in the selection of products they see in the Steam Store. Prior to the Discovery update, that human was someone at Valve. As Steam grew, so did the types of players using it, and the range of games they wanted to see. Over time, we became less and less confident that we represented the interests of all those different players.
The Steam Curators features was designed to be a solution to this problem, allowing anyone to contribute to the process of highlighting games on Steam, and then allowing players to find someone who they felt represented their gaming interests. Some players feared that we'd give too much power to curators to control the Steam store, and that their own particular interests wouldn't be represented. So, as with the other filtering features in the Store, we chose to keep Curators strictly an opt-in feature, where the input from each curator only affected the Store algorithm for players who explicitly stated they wished to follow that curator.
Unfortunately, while we shipped the Curator feature in the first Discovery update, it hasn't received the attention it needs to be a good solution. So recently we've spent some time talking and listening to members of both sides - the curators using the system to provide commentary on the Store, and the players using the system to inform their decisions. In both areas we've identified a set of work that we believe will make it more useful.
We're expanding the kinds of content that Curators can create, allowing them to provide more information to players who are thinking about buying a game, and improving the tools to allow them to easily manage all their recommendations. We'll have some more details as we get closer to releasing the update, but here are some highlights:
Many Curators create content for other platforms, such as YouTube, so we're making it much easier for them to show that content alongside their curations.
One suggestion from Curators that we liked was the ability for them to create personal lists of games. This will allow Curators to provide specific kinds of advice, whether that's general suggestions about which games to buy in the current sale, or more specific lists, like which games to play to follow the evolution of a particular type of game design, the body of a work from a favorite developer, or the games in a Curator's weekly Game Club.
Another big request came from both Curators and developers, who want an easier way to help Curators get pre-release access to upcoming games. It's often hard for Curators to get the attention of developers who build the specific kinds of games that a Curator covers, and it can be similarly hard for a smaller developer to find the Curators who would be interested. So we're building a system that will make that a painless process for everyone involved, which means that you should see more useful curations coming out of the Curators who like to explore newer titles.
At the same time, we're making it easier for players to use Curators to help them browse the Store. Since they're an opt-in feature, we've decided to give Curators more visibility throughout the Store as a whole, so if you're following a Curator, you'll see their thoughts in new places, and with higher prominence.
Next Steps For Steam Direct
Like all the work in the Steam Store, Steam Direct will take some iteration to get the kinks out. We're optimistic. Aiming for the low publishing fee gives every game developer a chance to get their game in front of players. The Store algorithm will do its best to make sure you see games that are worth your time to look at. Combining everyone's increased visibility into the algorithm's thinking with the human eyes of Curators will hopefully ensure that whenever that algorithm isn't working properly, we'll know about it, and have the chance to fix it.
Our next post will wrap up this series of posts, where we'll cover the sunsetting of Greenlight, and the timing for the release of Steam Direct.
In our previous post, we described what we believe a successful Steam Store would look like, and why balancing the interests of all the players and developers made it an interesting challenge. In this post, we want to talk about another group that adds further complexity: bad actors exploiting the store algorithm for financial gain.
The reason this group exists is due to various systems that add value to owning games on Steam beyond having the game itself. The best example is Steam Trading Cards, which also happens to be the primary one that these bad actors are exploiting.
We added Steam Trading Cards in 2013, and they had two main goals:
For players, they were small collectibles associated with games. They were tradeable, which meant you could collect ones for games you loved by trading away cards for the games you loved less. In effect, they were a way for you to show other players what your favorite games were. We knew some players wouldn't care about them, which was fine - they could simply throw them up on the Steam Community Market, and use the results to buy some other game.
For developers, they were an easy way to add extra value to their game, and provide rewards to their biggest fans.
After the release of Trading Cards, the number of players interested in them grew significantly, until it reached the point where the demand for cards became significant enough that there was an economic opportunity worth taking advantage of. And that's when our group of bad actors arrived, aiming to make money by releasing 'fake' games on Steam.
These fake developers take advantage of a feature we provide to all developers on Steam, which is the ability to generate Steam keys for their games. They generate many thousands of these keys and hand them out to bots running Steam accounts, which then idle away in their games to collect Trading Cards. Even if no real players ever see or buy one of these fake games, their developers make money by farming cards.
Farming Trading Cards for profit as a developer isn't rocket science. The primary difficulty is that they need to get a game up on Steam. For a while now, we've been engaged in an escalating war of disabling their latest method of gaming Greenlight's voting mechanisms, where each time we succeed, they circle around and come up with a new way. Unfortunately, this approach isn't terribly sustainable - they continue to get smarter and more large scale in their methods of generating tons of data, and we're spending more and more time fighting it.
We could restrict the ability for developers to generate Steam keys for their games, but we hate to degrade tools that legitimate developers are using to make their players happy. We're also not certain it would actually solve the problem - there are many ways a bad actor could try to get their game owned by all their bot accounts, and they just need to find a way to do it that costs less than they're making from selling their Trading Cards.
You might wonder why this is really an issue. After all, if no real players are buying their games, and their cards are being traded on the market to players who want them, where's the harm? Isn't Valve making money from the market fees on their Trading Cards? While there's truth in both of these points, the problem is that these games damage something we care about a lot, because it affects all our players - the Steam Store's algorithm.
As we mentioned in our last post, the algorithm's primary job is to chew on a lot of data about games and players, and ultimately decide which games it should show you. These Trading Card farming games produce a lot of faux data, because there's a lot of apparent player activity around them. As a result, the algorithm runs the risk of thinking that one of these games is actually a popular game that real players should see.
So we've decided to take a different approach - remove the economic incentive that's at the root of the problem.
Here's what we're doing:
Instead of starting to drop Trading Cards the moment they arrive on Steam, we're going to move to a system where games don't start to drop cards until the game has reached a confidence metric that makes it clear it's actually being bought and played by genuine users. Once a game reaches that metric, cards will drop to all users, including all the users who've played the game prior to that point. So going forward, even if you play a game before it has Trading Cards, you'll receive cards for your playtime when the developer adds cards and reaches the confidence metric.
The confidence metric is built from a variety of pieces of data, all aimed at separating legitimate games and players from fake games and bots. You might wonder why the confidence metric will succeed at identifying fake games, when we weren't being successful at using data to prevent them getting through Greenlight. The reason is that Greenlight is used by a tiny subsection of Steam's total playerbase, producing far less data overall, which makes it more easily gamed. In addition, Greenlight only allows players to vote and comment, so that data is narrow. Steam at large allows players to interact with games in many different ways, generating a broad set of data for each game, and that makes identifying fake ones an easier task.
With this change, we hope to significantly reduce the economic incentive for the bad actors to release fake games on Steam. We're hopeful that this will have little negative impact on other developers and players, with a small number of games having a delay before their Trading Cards start to drop. On the positive side, it should significantly improve the quality of the data being fed into the Store algorithms, which is a good thing for everyone.
As always, if you think we've missed something in the analysis of the tradeoffs we're working within, don't hesitate to discuss it in any major gaming forum (we read them all), or in our own Steam Community Forum.
Next post, we're going to talk about the Steam Direct publishing fee, and some other changes we'll be rolling out soon.
Whenever we announce a change to the Steam Store, we're always really interested to read the discussions that follow. Obviously we see a wide range of opinions on how good a job the Store is doing, but increasingly we're seeing that people have very different ideas of what its job even is - and what it should be.
That's understandable. One of the reasons it's so hard to make a good store - one of the reasons we've been working on it for years, and one of the reasons we think we still have years of work left to do - is that it has so many jobs. It has to serve so many players whose tastes and interests are not only different, but sometimes complete opposites.
So we thought it would be useful to define what we believe success would be for the Steam Store. That way, everyone would understand what we're trying to do, and discussions could focus on what we're trying to do separately from whether or not we're doing it well enough. This distinction also helped us realize we should be collaborating more directly with the community around improving the Steam Store.
This blog post aims to start that process by being the first in a set of three that explains our thinking around the Steam Store, and our plans for how we'll improve it with Steam Direct. We're going to talk about Store's goals, and how it executes them. In the second post, we'll cover some ways the Store is being exploited, and some changes we're making to address that. Finally, in the third we'll talk about the Steam Direct publishing fee, and some features that we'll be releasing in the coming weeks.
So what would a successful Steam Store look like? To answer that, we need to look at all the different kinds of people who use it.
Players who are highly connected to the online game community & conversations, and players who are totally unconnected
Players who browse the store looking for a game, and players who arrive already knowing the title they're looking for
Players who come to the store once a month, and players who visit multiple times a day
Players who just want to buy the latest AAA title, and players who want to search for hidden gems
Players who want to play titles earlier in their development, and get involved in their evolution
Players who want games with specific attributes, such as a type of gameplay, support for a specific technology, translation to their local language, etc
Developers with AAA titles that have large, existing fan bases, and developers who are barely known, yet have a game that would be a hit if players found it
Developers who want to build deliberately niche games, and have them find that niche audience
Developers who want to get community feedback earlier in the development process
We believe that a successful store would be one that treated all these people, both players and developers, in a manner that they would consider fair. Unfortunately, these groups often have competing interests, so it's important to understand that if we're not doing exactly what one group wants, it's probably because we're trying to weigh it against another group's interests. It might seem obvious that developers have some competing interests, but it's also true on the player side - some players specifically enjoy exploring Early Access titles, while others never want to see them.
And ultimately, that is why the Steam Store is a design challenge. We could make the problem a lot simpler by choosing to ignore some set of players or developers, but we think there are already stores that have chosen to do that, and it's much more interesting to try and figure out how to build a single store that works for everyone.
What's been done so far?
For a while now, the features we've been building have all been aimed at making the Store more successful for those groups of players and developers. Allowing the community to tag games into useful categories, and allowing players to filter the store to their tastes, let players control what they see in the Store. The Discovery updates helped players who came to browse the Store, and developers who had games that needed a certain kind of player to find them. Curators, Reviews, and Refunds all tried to help players and developers of niche or undiscovered games to find their audience.
Greenlight was a step towards opening Steam up to a wider range of games and developers, rather than us acting as gatekeepers trying to guess what people will like. We've seen huge successes from games we had no idea would be popular, and whole new communities have sprung up around genres that previously couldn't get on Steam at all. To us, that confirmed our suspicion that no single, small group of people should be sitting in judgement over what is and isn't a good game. We should do some basic checks to make sure the game works, and we now do that on every title - but not insert our own tastes as a filter between what developers want to make and what players want to play. We could serve one particular group of players that way, but Steam can and should serve a more diverse range of people and experiences than that.
AAA players and developers have probably had the least amount of new features applied to them, largely because our data showed that the store was already working well for them - but we have to be careful to not stop that being the case in our efforts to help all the other titles. As much as the online conversation is dominated by indie titles, there's a huge audience of players who just want to buy AAA titles.
These all feel like positive steps towards what we see as the goal for the Steam Store. But we know it isn't serving every type of player and developer as well as it could, so here's what we're focusing on next.
Exposing the Store's inner-workings
The algorithm behind the Store that's tasked with achieving the goals we've described above ultimately ends up producing this: the games you see when you load up the Store.
The Store is constantly trying to balance all the different interested groups of players and developers. It knows that it has a limited number of spaces it can use to show games to a player. It has some knowledge of the player, if the player is logged in and has a purchase / play history. It has some knowledge of the game, based on what the developer has told it and what previous purchasers of the game have said & done. It chews on all that data, and finally, decides which games it should show the player in all the various sections of the Store.
The problem with black box algorithms like this is that it's hard to know when they aren't working as intended. Did we not show a game to a player because the algorithm correctly guessed that the player wouldn't be interested in it? Or because there were other games it thought the player would be more interested in? Or just because of a bug?
We had similar problems in the Dota 2 matchmaking system, which was also a black box algorithm. We found that when we better exposed the data around the black box (in that case, the matchmaking ranks of the players), our players understood the black box better, and as a result, were able to better identify cases where it wasn't working correctly.
So we're going to do the same with the Steam Store. We want to show you more of what it's doing and why - and we have some features planned to help with this, starting with one we're launching today: an algorithm section on game pages that states why the Store thinks this game will (or will not) be interesting to you.
This section will let you see inside the black box, and understand what the Store is thinking. We hope it will be useful whenever you're exploring the Store, but in particular, whenever you've navigated from an external web page directly to a specific game's Store page. In those cases, this section will help you understand whether or not this game is something the Store would recommend to you. In other cases, you might be more or less interested in something the store recommends if you know exactly why it's recommending it. For instance, knowing that a particular friend or curator likes or dislikes a game might make it clearer whether you'd like it. Finally, if the store recommends something you know you're not interested in, you'll be able to see where its decision making is going wrong, and tell us about it.
Hopefully this post gives you a better understanding of what we're trying to do with the Steam Store. In our next post, we'll be covering the ways that bad actors have been gaming the Store algorithms to create revenue for themselves, which confuses our algorithms enough that it starts serving customers less effectively. We'll cover some changes that we believe should tackle the problem.
Following that, we'll talk about Steam Direct's publishing fee, and how we're approaching that decision.