Hacker News Reader

Comment by saagarjha | Hacker News Reader

Humans are notoriously bad at this. See countless examples of people claiming “hax” when they’re against a player that’s much better than them.

View on HN

Those people are not looking at the other players’ screens. CS Overwatch works by saving replays and is pretty accurate

I haven't played CS OW but based off the description I just read on the replays, it seems like they are probably storing the interactions as text data and then using it to show the visuals later. That can be fudged.

Back when I played a lot more, I would usually call it a bad day if I didn't get accused of cheating a few times.

I guess it's the Dunning-Kruger effect, where players think they're a lot better than they really are, so they can't imagine how someone can be _that_ much better than them.

This brings me back. When I used to play competitive BF2 in one of the tops clans in TWL, we'd start our thrice weekly practises by squading up in a pub server and play until we got votekicked and then switch to our private server.

IDK if it's still the same these days, but in BF2, 8 people in a squad all working together and doing their role within the squad could easily roll over 32 randoms on the other team.

That's a double edged sword though. Humans being bad at telling the difference between cheating and skill is not good for training your AI to spot cheaters but at the same time it lowers the bar for how good your AI needs to be. The metric for whether or not your anti-cheat works is "whether or not humans can tell others are cheating". It doesn't matter so much if it can't identify the subtler ones if the players can't either.

Reply on news.ycombinator.com