481
Epic Games is hiring a Senior Game Security Engineer for their Anti-Cheat team to champion Linux anti-cheat capabilities while working on OS internals, reverse engineering, and protecting multiplayer (lemmy.ml)
submitted 4 months ago by to c/linux_gaming@lemmy.world
130 comments fedilink hide all child comments
 

https://www.epicgames.com/site/careers/jobs/6019148004

you are viewing a single comment's thread
view the rest of the comments
[–] 28 points 4 months ago (9 children)

It also doesn't even work. There are so many DMA cheats out there that make kernel AC systems a joke. Meanwhile, Basicallyhomeless built a physical aimbot mousepad that can't be detected because it doesn't even hook into the computer. It moves the mousepad to correct your aim. It's extreme, I know, but the point is that motivated cheaters will always find ways to make AC systems completely useless. Kernel AC has already been defeated, and it's too high of a cost for something that is already defeated.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 9 child comments
    •
  • [–] 4 points 4 months ago (8 children)

    Yeah, at the physical aimbot mousepad level, the only thing that can detect that would be utilizing something like machine learning techniques for detecting/flagging accounts.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 8 child comments
    •
  • [–] 5 points 4 months ago (6 children)

    I fail to see how ML would be able to distinguish between that and a really skilled player aiming normally.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 6 child comments
    •
  • [–] 6 points 4 months ago (5 children)

    You wouldn't necessarily need machine learning, but you would need some sort of heuristics algorithm that checks a player's inputs to ensure they look like real human inputs. I'm sure the auto-aim mouse pad makes microadjustments or sudden changes that aren't feasible for a human to make, and that sort of stuff is readily detectable.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 5 child comments
    •
  • [–] 4 points 4 months ago (1 child)

    Not feasible for most humans to make, sure. I just wonder where the line gets drawn between ML and MLG, or even if it's a bright line at all.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 1 child comment
    •
  • [–] 5 points 4 months ago

    The line is gonna be fuzzy, and based on multiple systems, both automated and manual, and none of it will be perfect. But at the end of the day, it's a fucking video game. I'd rather lose occasionally to a cheater than install malware on my computer.

  • permalink
  • fedilink
  • source
  • parent
  • [–] 2 points 4 months ago* (2 children)

    until you use machine learning to collect all the game footage of pros mouseflicks so you have data for a "human-like" flicks between any 2 mouse positions, which becomes how your aimbot moves your mouse. Now the aimbot can't catch you without also getting false positives from those pros

    I have always been a believer that the best anti-cheat is a proper way to measure player skill for your SBMM, with that cheaters will naturally only be matched against other cheaters.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •
  • [–] 1 point 4 months ago (1 child)

    Lol you're really underselling how difficult it is to write cheats.

    There's still quite a few ways to detect that kind of thing though, both automatically and manually. At the end of the day, if enough people report the 3 day old account for hitting flicks like Carpe, it's gonna get banned pretty quick. And if a few escape through the cracks, well, like I told the other person, I'd rather lose to a cheater occasionally than install malware on my computer.

    SBMM can be a good solution as well, but I believe that should be a separate, opt-in thing. Matchmade games always tend to lean more competitive/sweaty/sometimes toxic, whereas games without matchmaking are usually more casual-friendly. I think it's nice when games give you the choice.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 1 child comment
    •
  • [–] 1 point 4 months ago*

    The difficulty kind of doesn't matter, because what we want here is recognizing the ceiling of cheat capabilities, and my point is that in this cat and mouse game, in the end when we approach the ceiling, the cheaters will win in the end.

    All because they control their own hardware, they have the final say of what 0s and 1s they send to you in the end.

  • permalink
  • fedilink
  • source
  • parent
  • [–] 2 points 4 months ago

    It wouldn't be difficult to randomize the bot with a noise filter to vary speed/accuracy and have it aim for center of mass, but slightly off center. Overshooting movement could help evade detection as well. Another tactic would be to let the player do most of the work, but correct it only when they would otherwise miss. Even letting them miss occasionally would throw off scrutiny. There are so many ways to hide it.

  • permalink
  • fedilink
  • source
  • parent