do inconclusive reports lower weight?
examples. reporting for undetectable but legitimate forms of trolling
examples. reporting for undetectable but legitimate forms of trolling
I mean it greatly depends on what you reported them for as well as if you leaved a detailed comment on the matter. If trolls are getting creatives when it comes doing what they do. The system won't automatically know why you were reporting them.
Like say someone is following you in the jungle trying to steal your farm. Just reporting them for trolling wouldn't be enough since the system alone cannot check someone following you and more than likely needs to be manually checked!
you could try that but no one is reading the reports. its a computer program. you report them it scans their chat for toxicity, it checks their clicks per mintue for afk, it checks their k/d for feeding and then if it cant find anything it lowers your report weight.
The only statement here that is correct (but mostly because it's extremely vague), is that the program scans their chat for toxicity. But again, this is really vague and sounds way more simple than it really is.
There still are manual reviews, specifically when it comes to trolling, as that system is still learning. But generally, there are manual reviews to find loopholes in the automated systems. Those people do not only read the comments on the report, but those comments also ensure they're looking at the right thing when they make their decision.
Clicks per minute is probably a part of afk detection but not all, as players have been afk-bustered when they were clicking around and moving in base. The exact things they track is intentionally kept secret. Riot even has confirmed that they intentionally do not have every detected afk/leaver show up with a red name, so while it is definitely plausible, it isn't even a confirmed thing that it looks at apm at all.
K/D is only a minor part in the feederbuster detection, as there have been cases of people getting banned for feeding at the end of games (or even just a limited amount of kills) while actually maintaining a reasonable and sometimes even positive K/D. At the same time some people have horrible games with a horrible K/D, who are not banned. Only looking at K/D is actually something that Riot explicitely denied, as that was one of the main concerns of the community when they announced it.
all your corrections are made up nonsense. there are millions of reports every day. there is not enough time in a day for rioters to read every one and check on it. lets say they could read the report and check to confirm it in 10 mins and pretend there are only 1 million reports in a day. (way low with 60 million active accounts). that would be 10 million minutes = 166,666 hours. per day. it aint happening.
there are no doubt occasional manual reviews but they are like finding a unicorn.
i believe clicks per minute and xp per minute are both taken into account for afk. but yes this one is a bit foggy.
if you are hoping to see people get banned with a decent k/d, you are hoping to find another unicorn. the only way they are getting caught like this is if there is a rioter or a friend of a rioter actually in the game with them.
i am glad i could educate you.
there is not enough time in a day for rioters to read every one and check on it.
Never said that. But manual reviews are mainly there to teach the system how to detect new "patterns". That's why manual reviews will mostly occur for very frequently reported players that do not exhibit (yet) detectable negative behaviour. So if you are worried that your report might not count because they are so low-key trolling, your chance of a manual review are actually significantly higher. Unless you are wrong and their behaviour is detectable. Then it doesn't really matter, does it now?
if you are hoping to see people get banned with a decent k/d, you are hoping to find another unicorn.
W00t! Guys! Guys! I've seen a unicorn! Seriously though, I've seen exactly that happen. It wasn't even based on a manual review, but the automated feeder buster. It's been too long for me to believe I can find the post again, so feel free not believe me.
tl;dr right upfront: This is speculation. I don't know how the system works, I can only (roughly) explain how I would do it. If you're not interested in speculation, downvote and skip it.
The key (in my hypothetical system) would be confidence. If the system has a high enough confidence that somebody did do something punishable after being reported (judging by in-game data), that person will be punished, and the reporters would have their weight increased if below maximum weight.
If the system has a high enough confidence that somebody issued a false report (judging by in-game data and report data), that person will get their report weight lowered if above zero weight.
In the entire middle area nothing happens to the reported or the reporter; unless the report data suggests the reported may be doing something under the radar, at which point there is a manual review.
So far I have not seen any evidence of Riot not going by this (again, very rough) idea, and I've seen a few posts pointing into this general direction.
So for your example, in my hypothetical system: The troll/feeder system still seems fairly bad at detecting things, which I'm certain is quantifiable to some degree, which means the system has low confidence in both directions. Could be a troll, could be not. So however rarely the system finds a definitive troll, that would roughly also be the scale of how rarely it will find a definitive false report for trolling. Or differently put: You will have to be very very blatantly trolling to get punished, and you will have to be very very blatantly false reporting to get your report weight lowered.
Hope that somewhat helped.
Why the hell downvote a question? Makes no sense at all. Sometimes I feel like people just scroll down on the main menu downvoting every topic without even reading the title.
yes.