As someone who rehosts an old game after the official servers shutdown, we have a dedicated servers for cheating and real moderators for the non-cheat ones. It works great but big corps don't way to pay for mods.

I also wonder why big companies don't do it to train ML algorithms on the cheat server data too..

[–] PlexSheep@infosec.pub 4 points 6 months ago (2 children)

Besides that, statistics is an unbeatable tool against cheaters.

[–] Sonotsugipaa@lemmy.dbzer0.com 3 points 6 months ago

Agreed. For detecting cheaters, statistics work like a Dream

[–] aksdb@lemmy.world 2 points 6 months ago (1 children)

Only if you want to cap the skill limit. Otherwise you would typically have a hand full of players that are genuinely just good or rather far outside the normal skill range. I guess with a lot of data collection one might be able to determine if there was some kind of natual progress or sudden skill jumps, but all in all it could weed out legitimate players.

[–] PlexSheep@infosec.pub 1 points 6 months ago

You can detect the hit ratio for shooters and win rates for games with matchmaking, those are really good indicators for cheating.