Since there's no native Linux binary of the game, I assume the real issue here is Windows users telling the game "you're running under Proton, no kernel anticheat please!" and bypassing all of those additional checks.

Might be selection bias? How do they detect cheaters / what is "infection"?

If they had a reliable way to detect cheaters they would kick them immediately, how can they have a reliable metric of what they can't catch?

Just FYI those aren't scare/sarcastic quotes. It really did work and for fairly straightforward reasons.

(Which isn't to say I'm happy about it, as a SteamDeck owner.)

I'm not convinced by their graph. The decline was already apparent before the Linux ban. There's a slight additional decline. Also, I never trust graphs that don't show you the y-axis.

Some people in the comments were saying that they were also deploying additional security measures at the same time but seem to be blaming the decline completely on Linux bans.

You can also see an uptick in cheating after a certain point, leading me to believe that potentially some of their additional measures were effective, until cheat makers learned how to get around them.

As others have mentioned, this meets all the criteria for selection bias as well as a number of other forms of bias.

You can't derive meaningful numbers without doing a few things, like having non-ordinal metrics at scale on a graph.

Gaming companies have a long sordid history of providing graphs that do not reflect reality especially in press releases, to support a narrative that is more likely to be false than true historically.

The metrics aren't shown, or properly defined in any meaningful way. This translates down to them saying, We've done something about the cheaters once and for all, it was the linux people, trust us.

There's a fairly heavy conflict of interest in these cases, and most would rightfully say this in its current state meets a criteria for misinformation.