I was at GDC a couple weeks ago and attended Sergey’s panel where he first presented that data.
I definitely want to hear more details about the Chinese cybercafe business he describes there. If that specific detail is underestimated, it will paint a completely different picture than what the raw SteamSpy data represents.
For example, my impression throughout all of 2017 was that PUBG is what finally got Chinese players into Steam, resulting in the massive swing in Windows 7 vs. Windows 10 numbers, and along with it potentially a burgeoning new Chinese audience for developers to sell their games to on Steam.
But the presentation kind of buries the lede in there, presuming that it’s far more likely that tens of millions of new Steam accounts are being set up repeatedly in gaming cafes in China, heavily skewing the results, and of all of those tens of millions of accounts, they’re either buying nothing, or buying only PUBG.
Sergey also described in his talk at GDC that it makes economic sense for these cafes to continually buy new copies of PUBG for new accounts, over and over again, as previous accounts get banned for whatever reason - PUBG is cheap enough in Steam that only a few hours’ play in a cafe will pay for a new copy, when people get banned for cheating or other reasons it’s not that big of a deal for them.
To me, as a part-time data nerd who does a lot of data analysis for my job, this is a classic example where comprehensive data insights might be extremely misleading if that data is missing some extremely important key details.