Games using anti-cheats and their compatibility with GNU/Linux or Wine/Proton

Post by kalaksi » Mon, Dec 01, 2025, 10:41 AM UTC

kachapopopow wrote: post #5

hastily3114 wrote: post #4

Is there really no way to make anti-cheat on Linux that can't be bypassed? I don't know much about this, but it seems very difficult to make an anti-cheat for a platform where you can make changes in the kernel.

Linux explicitely allows you to do things that makes cheating *really* easy. There is also complete lack of secure boot and a way to validate that your kernel hasn't been compromised. I mean seriously, making a cheat for a proton supported game that no anticheat has any hopes of detecting are in 100 lines of a kmod driver and 1 console command: insmod. On windows you at least need to use scuffed tools like KDU to byp…

Post by doener » Mon, Dec 01, 2025, 10:42 AM UTC

givemeethekeys wrote: post #17

Between Windows being so unbearably bloated and no way to make anti-cheat really work on Linux, it looks like the consoles win!

Post by soloridindan » Mon, Dec 01, 2025, 10:42 AM UTC

vkou wrote: post #16

soloridindan wrote: post #13

Earlier quoted context omitted.

I think anything that relies on reflex alone is flawed design. You can design around this, by for example in Dota2 it doesen't matter how fast you click an entity, because the turnrate of your character is limited, so a person clicking reasonably fast and a bot clicking in 0.01ms both arrive there at the same time. Precision also doesn't matter, because a player can click the icon of the enemy instead of trying to ma…

Unless DOTA2 is running at a ~3 tick rate (Which it's not), even taking account processing delays and action batching, a bot will always have faster reaction times than an actual player. It will also never misclick. This problem is magnified in a shooter game, which would be unplayable with that kind of batching, but where a cheater with an aimbot is actually impossible for a legitimate player to beat.

Post by ChocolateGod » Mon, Dec 01, 2025, 10:43 AM UTC

soloridindan wrote: post #18

vkou wrote: post #15

Earlier quoted context omitted.

That only solves half of the cheating problem - illegal inputs from clients. The other half is much harder to solve . For a simple example - my client knows that there is an enemy player around a corner. It knows exactly where that player is, because that player is walking, and making noise. A cheats could allow the cheater to see his opponent's player's model through the wall. For a more blatant example, consider ch…

You already know where an enemy is if you hear them behind the wall, you don't need a cheat to tell you that there is noise coming from other side of the wall. The server also doesen't need to tell you they are behind it if they're sneaking. A game that allows zero home-in time sounds like a flaw in the game and something solvable on the serverside. You can replace a playermodel with wider "sound coming from around h…

Post by soloridindan » Mon, Dec 01, 2025, 10:46 AM UTC

brettermeier wrote: post #14

soloridindan wrote: post #8

Earlier quoted context omitted.

I think the moment you accept data from the client as truth you've lost the battle already, everything else is just damage control. Loads of games have realized this and kept checking game rules on the serverside and reveal data on a need-to-know basis. This makes it nearly impossible for cheats to be made because anything you know you should know, and everythin you act is parsed by the backend according to rules alr…

However, this only solves the cheat problem to a minimal extent. There is a lot of important data that players should not be directly aware of, but which is important for the game. For example, it is important for calculating sounds to know where enemies are nearby, even though you cannot see them, which makes wall hacks possible, etc.