Bug fixes, new data, new maps, and new leaderboards...it has been a busy summer so far!

It has been a little over a month since we launched version 3 of Game Oracle but there was a lot of stuff left on the cutting room floor that we feel deserved to make it...maybe there were also some bugs to fix 👀 So we've been hard at work cracking out a new minor update for you all. Let's sum up what v3.1 has in store:

Squashing Bugs

There were a number of issues we picked up on after our last release that needed addressing and we're glad to say those bugs have been squashed! Here are the major ones we've resolved:

• When searching for similar games in Data Explorer the distance between a query and a game wasn't being shown correctly, and sometimes the distance would disappear from the results table when altering your table settings. Clearing this one up actually (in a strange way) revealed an opportunity to redesign some of the logic behind our table component which improved performance and made re-introducing table search a possibility - what a happy accident!
• There were a few issues with rendering menus and card elements on some mobile displays - our relationship with Tailwind CSS is slowly improving, one day we may even be friends.
• In some edge cases, e.g. with Battlefield 3 for some reason, the developers lifetime revenue was not being calculated. Turns out there was a bug in our code for aggregating a developer's revenue over time. This has been corrected and we've improved our tests to prevent future mishaps like this one.

Reflecting on this work, we have introduced a new self-service feature in our customer dashboard. You will now find a "Report/Suggest" toggle in the top right-hand corner of the dashboard. When you click this button you will see a form where you can report bugs and offer suggestions for us to act upon. We will do our best to address these problems/suggestions in a timely manner. 

IGDB Data

Great news everyone! We're now partnered with IGDB, the community powered gaming knowledge base, allowing us to bring you even more data about the PC marketplace. In v3.1 we've shipped new data assets associated with every game on the Game Oracle platform:

• Game Engines: about 15% of Steam games on our platform now have their game engine listed offering insight into the technologies behind games in the market segments that interest you most
• Platforms: now you can see what other platforms games are being distributed on as well as Steam e.g. Xbox, Playstation, Switch etc.
• Social Media: for each game we now provide a full list of their social media platforms offering you the opportunity to investigate the common marketing strategies applied by your competitors
• Franchises: new flags that identify franchises, helping you understand their design pivots and revenue growth
• Remakes/Remasters: new flags to identify games that are remakes/remasters granting valuable insight into the impact of brand revivals
• Time-to-Beat: for almost 6000 games we now have statistics on how long players reported it took for them to complete the game which can help you gauge expectations for particular genres and mechanics

We also now offer the ability to filter your search results in Data Explorer for particular game engines, helping you focus on technologies that are relevant to your development.

 

Leaderboards

We've been discussing these internally for a while now so I am excited to announce that v3.1 delivers two free-to-access leaderboards:

1. The Outlier List
2. Game Engines Leaderboard

The Outlier List

Our biggest asset at Game Oracle is our map of Steam, a statistical model of how similar games are to each other. The Steam map is great at identifying games that exist in uncrowded markets. A TLDR for those unfamiliar, but we can measure the 'distance' between games on our map which tells us how unique a game might be. The 'distance' is captured by a metric we developed called the 'Saturation Score', a measure of how saturated the immediate market around a game is. 

We wanted a way to celebrate the wackiest and most wonderfully different games on Steam, so we decided to create The Outlier List.

The idea is simple:

1. We select games with at least 100 reviews from indies and small studios, excluding big triple As
2. Then, we filter for games whose Saturation Score is in the bottom 1%
3. Finally, we ranked the games by their SteamDB score which is a balanced measure of player reception (this is also why we only include games with at least 100 reviews)

We hope you enjoy The Outlier List and that it offers a nice source of inspiration for the types of left-field games that find success in the PC gaming market.

Game Engine Leaderboard

Following our partnership with IGDB we thought it would be really interesting to examine the popularity of different games engines and how the games created by those engines perform on average.

There are many caveats going into this one, which we explain in the "important caveats" section of the Game Engine Leaderboard. The most notable caveat is that the choice of game engine does not correlate with the quality of games the engine produces; different engines excel in different applications, but the quality of a game will be impacted by a host of different factors.

Nevertheless we still thought it would be interesting and possibly helpful to display this information in an interactive manner. In the leaderboard you will find both free-to-use and proprietary game engines. You can rank those engines according to:

• Average SteamDB Score: this will rank the engines by the average SteamDB score of games created by each engine, providing insight into player reception of those games
• Most Popular Engine: this is simply the percentage of games associated with each engine (note that this is a percentage of games where the game engine is known, which is currently around 15% of games on Steam)
• Estimated Total Units Sold: estimated using the Boxleiter method, this is the number of units sold in total across all games associated with a game engine
• Estimated Average Units Sold: estimated using the Boxleiter method, this is the average number of units sold amongst all games associated with a game engine

Another Map Update

Last but not least in our list of changes coming with v3.1 is another update to our map. As we have mentioned before, we are constantly improving our map to try and provide the most accurate model of game similarity and PC gaming markets. 

Following some feedback from Moss Monkey Games and Vigenère Games we made some upgrades to how we allow the model to learn the metric that defines game similarity. Long story short, the model was over fitting to specific keywords in some market segments and wasn't learning the essence of what makes games similar in these cases.

Additionally, we found the model was not giving enough weight to differences in visual style which often led to thematically similar 2D and 3D games being placed in the same neighbourhoods.

After many sleepless nights and jugs of coffee consumed, we managed to develop an updated approach which is has greatly improved the model's understanding of what defines a market segment.

TLDR: v3.1 is ready!

So in short, v3.1 is now available and brings a host of new exciting changes:

• Multiple bug fixes
• New data kindly provided by IGDB
• The Outlier List and Game Engine Leaderboard
• A new and improved map