4 ms·
How long before cheaters can buy a specialized robot/deep learning machine that has a high speed camera and servos that they can point at their computer screen
by wayneftw 7y ago
How long before cheaters can buy a specialized robot/deep learning machine that has a high speed camera and servos that they can point at their computer screen where it can watch the screen and work the keyboard and mouse for them? (Or, better yet, the robot plugs into USB providing it's own keyboard/mouse/controller input and video output is recorded via a simple VGA or HDMI...)
I think these physical types of cheating systems already exist for smartphones. Last time I mentioned it, some people hinted at systems you can setup for phones - https://news.ycombinator.com/item?id=21991775 https://news.ycombinator.com/item?id=21991775.
Perhaps we're just waiting on the right software for this type of revolution to occur in PC gaming. I can't think of what's lacking right now though - all the pieces seem to be in place.
- floatingatoll 7y agoIt’s impossible to stop someone who can build robotic hands, but it’s still worth stopping cheap/easy software and hardware cheats.
- ThrowawayR2 7y agoThe blocking factor for that scenario is that the hardware to do that kind of cheating isn't cheap enough that the majority of cheaters can afford it. Get it below maybe $100 or so and then it would become a real problem.
- rasz 7y agoWould be trivial for less graphically cluttered games. Your external cheat could run a copy of a game engine constantly matching player position/orientation and comparing rendering of empty map to video feed from low latency capture card. Simple visual diff is all it takes to identify even 1 pixel of enemy peeking from around the corner. You could inject visual overlay on top of external video signal and auto trigger mouse click when crosshair points at potential differences. No fancy AI/deep learning required.
- floatingatoll 7y agoIt would be cheaper to hire a gig worker to play the game for you, which is how we ended up with RMT farming as an industry in the first place. Why develop a robot when a human will do, more cheaply and with less fuss?