[dannersy]

Kernel level anti-cheat seems to be ultimately ineffective. I've seen this with my own eyes with backend data when I worked for a company with a moderately successful shooter. Kernel level anti-cheat is serving a similar function to something like EAC these days, which basically just keeps the lazy people out. With any decently sufficient backend data analysis, you catch these folks just as easy.

I am of the belief that if you cannot completely control the environment, like a user's physical hardware, you're fighting a losing battle. Forcing them to install spyware is, in my opinion, just marketing for your competitive shooter and too invasive for the perceived gain.

[ChocolateGod]

> With any decently sufficient backend data analysis, you catch these folks just as easy.

No you can't. Using server side analysis of behaviour will always get false positives, especially as ML networks are getting better and better at reproducing normal player behaviour.

Your server side logic can't detect wall hacks, ESPs etc.

> which basically just keeps the lazy people out

That's better than nothing. Cheating will always be possible (see DMA devices), but as long as you make it reasonably hard to put most people off it, that's a success.

> cannot completely control the environment

But applications should be able to attest the environment is not modified, which is what desktop Linux lacks. Windows, macOS and Android do have methods to do this.

> like a user's physical hardware, you're fighting a losing battle

The alternative is no releases on PC.

[dannersy]

> The alternative is no releases on PC.

If you think there aren't hackers on console, you're deluded.

> Your server side logic can't detect wall hacks, ESPs etc.

Correct, neither do the invasive tools. You can get these things without modifying memory or the game, which is the majority of what the kernel level stuff keeps on eye on.

> especially as ML networks are getting better and better at reproducing normal player behaviour.

True, there is no answer here. Again, if the provider is this sophisticated, then EAC isn't helping here. Then it is just the usual cat and mouse, and the kernel level is moot.

> But applications should be able to attest the environment is not modified, which is what desktop Linux lacks. Windows, macOS and Android do have methods to do this.

Also true, but they are frequently circumvented anyway, so ultimately you're at the same level of Linux from a functional standpoint.

I know you're probably thinking that we should just hit all vectors, and I agree to an extent. The problem is that those who charge for cheats easily circumvent (yes, easily, they are literally within 24hrs sometimes of adjusting to modifications by anti-cheat software) the methods you're claiming help. While the data you get from users on the server end doesn't paint a whole picture, it is the source of truth of what the game decided happened, so it becomes the only thing you can rely on. I know folks who specialize in this area who basically believe that online competitive shooters are basically lost and it isn't worth the investment to fight it.

I am of the opinion we should just go back to dedicated servers and have self moderation. Sure, you get false positives if the owner or their mods suck at the game, but at least the players can do something about it. CS2 is a fucking nightmare at the moment.

[not_a9]

EAC is a kernel mode anti-cheat on Windows, though?

And given the whining on UC it seems to be fairly effective.