It worked and continues to work beautifully. There was a video recently by Dota Alchemy about how the system applies to high a penalty if you have an accidental abandon or just raged way too hard for a few consecutive games, but you can grind up the behavior score quickly through 10 turbo games.
Quickly? A turbo game is half an hour. That would be a whole evening to get 10. Not to mention that a lot of things could go wrong in that many games.
Abandon is ruining a match for 9 other players. So 10 looks like about right.
Meh. It's one of those cases in which people will praise the decision and be like "haha stupid cheaters will cheat on each other". But the reality is, if you can tell someone is a cheater, you could have banned them. If you make decisions like this, then you are admitting that your cheat detection code isn't perfect. As a result, some people with obvious cheats will have this experience, but many people will have the…
No cheat detection code is perfect. Banning cheaters encourages the creation of new cheats (or just new accounts, since it doesn't cost anything to start playing Apex). Allowing cheaters to continue playing is a win for two reasons:
1. They don't know they've been quarantined, so cheat makers will get less feedback about which cheats have been detected.
2. Cheaters are more likely to try new cheats, so it gives a sandbox of users that can be used to find other cheats as they arise.
Counter strike has a similar system, based on a hidden score called trust factor. Matchmaking system does some optimization and tries to pair similarly skilled people with similar trust factors. They also have a neural network setup that constantly scans match logs for cheaters. If I remember correctly it is responsible for > 50% of all cheater reports (the rest comes from players). Final conviction comes from a crow…
I HATE trust factor. It's TERRIBLE for me, personally. I don't cheat but get reported a lot and my trust factor is very low. I end up playing with/against cheaters constantly.
It also doesn't seem to have any affect on casual game modes, which is really irritating.
I hope they are also keeping separate rankings for cheaters, that would be pretty cool. "I'm currently the #5 player in the cheater zone of Apex." Heck, embrace it: don't call it the cheater rankings, call it the tool-assisted rankings, by analogy with "tool-assisted speedruns". Tag people at the end of the game with which particular cheats you detected. Pop up a polite notice that "We have noticed that you seem to b…
Many moons ago I made a "chatroulette" application for the reddit community. It was surprisingly popular, but was plagued by your typical penis flasher. I added a simple "report" button. After a user had been reported a few times they were silently sandboxed with other penis flashers so they could flash their penii at each other. I love this way of dealing with bad actors - but I think that its a bad idea to let them…
Not telling the user is also problematic. If your false positive rate isnt 0% you are seriously messing with some people. The same issue exist with shadowbanning, you can make an argument to shadowban trolls who dont follow the rules of a given environment on purpose, but shadowbanning a random user is just really ugly. Wasting someones lifetime like that for no reason is just wrong.
Counter strike has a similar system, based on a hidden score called trust factor. Matchmaking system does some optimization and tries to pair similarly skilled people with similar trust factors. They also have a neural network setup that constantly scans match logs for cheaters. If I remember correctly it is responsible for > 50% of all cheater reports (the rest comes from players). Final conviction comes from a crow…
While VACNet is amazing, and completely eliminated pretty much all "exploits" that cheaters could come up with to dodge the Overwatch system, Trust Factor on the other hand has been a bit of a mess. The assigned trust values seem really random in my experience (you can't see the exact values, but if you queue with someone who has a slightly or significantly lower Trust Factor than you do, you'll get a message), tend to fluctuate wildly from one week to the next or even just between single games for some people, and there are so many players complaining about it that they had to auto filter every thread that even just contains the words Trust Factor on the CS:GO subreddit.
Because it's a non-trivial feature to develop, test and maintain. Simply using a flag (banned = true or false) is lot easier / less complex / fewer potential side effects.
I would disagree with that. When you ban someone, they know. They make a new account and come back. When you move them to a cheater queue, they don't know. They keep playing, none the wiser. It's infinitely better for everyone. The cheater still gets to play, and the non-cheaters aren't affected as much because the cheater doesn't just keep coming back. The complexity in your app is that you need to have multiple mat…
>When you move them to a cheater queue, they don't know. They keep playing, none the wiser.
When 99.9% of the people you get matched with are cheaters (many of which might be quite obvious), it's not that hard to figure out that your account was probably shadow banned. Some cheaters might not care, but it's certainly not hard to figure out.
Many moons ago I made a "chatroulette" application for the reddit community. It was surprisingly popular, but was plagued by your typical penis flasher. I added a simple "report" button. After a user had been reported a few times they were silently sandboxed with other penis flashers so they could flash their penii at each other. I love this way of dealing with bad actors - but I think that its a bad idea to let them…
Not telling the user is also problematic. If your false positive rate isnt 0% you are seriously messing with some people. The same issue exist with shadowbanning, you can make an argument to shadowban trolls who dont follow the rules of a given environment on purpose, but shadowbanning a random user is just really ugly. Wasting someones lifetime like that for no reason is just wrong.
In my case, it matched them with me for verification. Not the most pleasant job in the world :D But yeah, you shouldn't just do it at random - it must be justified.
But is that not exactly what cheaters are doing right now? I reckon it would make less sense to implement this kind of system in games which aren't free. The price of the game disincentivizes cheating and makes consecutive iterations of the cheat harder. However, Apex Legends is free. Every time a cheat is detected nothing withholds its developer of creating a new account for the next iteration.
Sometimes... VAC can kick you mid-game for very obvious things. But in other cases you may be able to play for hours until a report goes through and gets verified. That delay (as a side-effect) guarantees that you can't quickly iterate and get immediate feedback on whether you're detected or not.
Often in console land, bans come in "waves" at pre-scheduled times, so you have a hard time matching the cause with the ban