Comment by why_at
19 hours ago
This is one use case where I think the idea of cloud gaming (e.g. google stadia) could make some sense. Having this as an alternative for linux users would be nice.
It's much harder to cheat if the game isn't running on your computer.
Or consoles. PS5 is pretty good for a cheat free environment
Generally yes, although some cheats like aim assistance would work fine on online streamed games, since they can scan your screen and adjust your mouse input to aim.
To be fair kernel anticheat can't block this completely either, it can be run on external hardware that uses a capture card to analyze your video feed and alter your mouse inputs to the computer. Generally undetectable unless the game is able to identify unnatural mouse movements.
>it can be run on external hardware that uses a capture card to analyze your video feed and alter your mouse inputs to the computer.
I think at some point defeating this becomes impossible. This sort of cheating isn't much different conceptually from just having someone who's really good at the game play for you.
Valve have been trying by training a neural net that watches every competitive cs:go match played to detect aim bots since 2017 with training data sourced from players labelling demos as cheats or no cheats, but I don't think they've hit a big breakthrough yet
That's a good idea, sadly I think gamers would reject it due to extra latency.
The ultimate "anti-cheat" is playing on some trusted party's computer. That can be a cloud machine, but I think today a game console would work just as well, turn that closed nature into an actual user-facing benefit. Console manufacturers seem focused on their traditional niche of controller couch gaming and not on appealing to high-FPS keyboard-and-mouse gamers, though.
Consoles are also vulnerable via peripherals. There are controllers that will run recoil countering scripts and things like that.
XIM fakes being a controller but is KBM. I sort of wonder whether it’s possible to use a camera to get a stream of the game and make an aimbot either by making a fake controller or a robot that manipulates a real controller.
Those are basically just scripts on inputs, pretty much nothing like aimbot or wallhacks
Yeah I don't think this would work for hardcore competitive gamers, but it would be nice to have as an option for those who are more casual. Definitely better than not being able to play at all.
It doesn't even seem very hard to implement, steam already has the ability to stream games, they could add this pretty easily as an option for any game (although there is the concern of the extra cost of running the servers).
>That's a good idea, sadly I think gamers would reject it due to extra latency.
That shouldn't be a problem if all players, regardless of the OS, are required to use the same cloud service with similar latency.
Cloud gaming is flatly non-workable for any kind of game where latency matters. This also covers most of the market for games where anti-cheats matter a lot.
> Cloud gaming is flatly non-workable for any kind of game where latency matters.
Not if only the rendering is done on the client. Look at rocket league.
Edit: of course, it is still possible to cheat in rocket league, but because all physics state is server authoritative at best a perfectly coded cheat could play like a perfect human, not supernatural.
I'm not familiar with Rocket League but server authoritative netcode is not comparable to cloud gaming. All games should be as server authoritative as possible to prevent cheating from the start. The problem is the client may have more state in memory than what you can see rendered on screen (players behind walls). Running the game on the cloud makes all of that inaccessible to cheats.
How is that a problem if everyone is using it?
People have varying distance to the server, and regardless, it would still feel really crappy for everyone.
1 reply →
Lag is the biggest issue... even a local wifi connection vs wired can make a massive difference in terms of what's acceptable lag.
Of course, to TFA's point on network code... a lot of the issues in question could come down to checking for movements that exceed human... moving faster than the speed in game, or even twitch aiming movements faster than a mouse, or a consistent level of X accuracy in shooting over time. On the last part, I'm not sure if there might be some way to mask a user's hit zone, rendering and such so that an aim-bot thinks the foot is center-mass, etc. Or if it could be randomly shifted in a test scenario.