Steam Spy, the main PC product beloved by some developers and hated by others, was recently declared dead by its creator, Sergey Galyonkin, after an update was made to Steam's privacy standards. It's coming back now, but it's going to be different.
The old Steam Spy appreciated the sales of a PC game by sweeping publicly visible Steam profiles and taking stock of their game collections. This technique was blocked in early April when Valve's Steam architects made it so that user profiles no longer automatically show which games the players own. The new version of Steam Spy will bypass this by using machine learning to predict sales based on "random data".
Galyonkin said in a blog post that he is learning to convert Steam Spy to a machine-based model. He admitted, however, that despite a wealth of data to feed his algorithm, it is "not very accurate" at this point.
"I have the data for about 70 games from different developers, and for 90% of them, the new Steam Spy is within a margin of 1
If the data could be so inaccurate, why bring Steam Spy back? Galyonkin based his decision on "over 200" messages from developers explaining how Steam Spy impacted their operations. He mentioned an indie company from Berlin who, with the help of Steam Spy's data, received state funds for his game developer in the post, he said Kotaku that it was inbetweengames, creator of the noir tactical game All Walls must fall .
"An estimate is usually better when decisions are made than no data at all."
Praise for Steam Spy was always plagued by caution earlier this month, many developers were relieved to see that it would seemingly go because of it Armchair analysts (and some annoyers) supplied data they had perceived Exactly, even if it was not true: in theory, a new, less accurate model could compound the problem. Galyonkin, however, believes that Steam Spy is worth coming back from the abyss.
"Armchair analysts do not need Stea's spy data," he said in an email. "They're fine with everything that's available: [concurrent players] the number of reviews, even a user rating." He noted that in his opinion, most players in a given game play the most games at the same time, helping them decide whether or not to buy multiple games. Valve, he added, is already making this number public.
Why developers can still benefit from the revitalized bones of Steam Spy, Galyonkin claimed that they are already happy to give reputable money for services that he believes are even more inaccurate.
"Relatively low accuracy does not prevent developers from paying $ 30 million / year for [an] AppAnnie subscriptions," he said, citing a popular app analytics service, "because making decisions is usually better as no data at all, analysts can work with data if they know that they are not exactly accurate, as long as they know the limitations and reservations. "
Galyonkin still fine tunes the new algorithm, and many of the old features of Steam Spy are not even back online yet. Country statistics, for example, are just off the table. In other words, things are early and there are still options on the table – some of them involve working with other interested parties.
"I have some offers from people with recent knowledge in the field of machine learning," he said. "I hope they will help me to improve the algorithm."
You read Steamed, Kotaku's site, which is dedicated to all things in and around Valve's popular PC gaming service. Games, culture, community creations, critiques, guides, videos – everything. If you found something cool on Steam, send us a message to let us know.