92% accuracy by only voting punish

12345 ... 8
Comment below rating threshold, click here to show it.

Gimpyloser

Senior Member

01-11-2013

I am not a scumbag so I of course don't just click punish, but I went through my last 100 cases, did the math, and had I done nothing but punish, I'd have 92% accuracy. Based on my review of the cases I pardon that are punished, it is clear that a large number of people are using the spam punish tactic for improved accuracy and rating. I'll keep taking the accuracy hits and keep doing the right thing, but this concerns me greatly. What is also concerning are the more recent comments by Wookie, regarding a few cases in which the cases presented to the tribunal did not warrant punishment. Wookie often researches the punished player and finds games which are not in the tribunal that would warrant a punishment in order to justify the failure of the tribunal system. The Tribunal is suppose to vote based on the evidence presented to them and finding games outside of the tribunal's level of access to justify why a player was incorrectly punished is not really a valid argument in my opinion. The problem here seems to be tribunal members spamming punish.


Comment below rating threshold, click here to show it.

OmenOfPlague

Member

01-11-2013

I think you are right about the "spam punish" but .... if wookie can take more casses to see if a summoner is toxic or not to the comunity i think at last the tribunal do his work, taking action against a guilty summoner ( my english is bad but im traying)


Comment below rating threshold, click here to show it.

magooomba

Senior Member

01-11-2013

Quote:
Originally Posted by Gimpyloser View Post
I am not a scumbag so I of course don't just click punish, but I went through my last 100 cases, did the math, and had I done nothing but punish, I'd have 92% accuracy. Based on my review of the cases I pardon that are punished, it is clear that a large number of people are using the spam punish tactic for improved accuracy and rating. I'll keep taking the accuracy hits and keep doing the right thing, but this concerns me greatly. What is also concerning are the more recent comments by Wookie, regarding a few cases in which the cases presented to the tribunal did not warrant punishment. Wookie often researches the punished player and finds games which are not in the tribunal that would warrant a punishment in order to justify the failure of the tribunal system. The Tribunal is suppose to vote based on the evidence presented to them and finding games outside of the tribunal's level of access to justify why a player was incorrectly punished is not really a valid argument in my opinion. The problem here seems to be tribunal members spamming punish.
Why does this argument always seem to pop up? Yet again, let me explain why this reasoning is faulty. You are concluding that the Tribunal is full of a "large number of people are using the spam punish tactic for improved accuracy and rating." In concluding this, you have already disregarded a much more likely conclusion: that most people who enter the Tribunal merit punishment. The Tribunal cases are not an assortment of random people in random situations. Rather, they are players who have been reported above the threshold needed to enter the system to begin with. It would be extremely faulty to assume that these players are simply a random collection and would show an average punishment level when the method used to get this sample is hardly random at all. As such, simply concluding that X% of players are punished, thus the system is broken is completely devoid of any analysis: it's simply your presumption.

You assume that your punishment level should be less than 92%, without any reason why. If I were to do the Tribunal and assume that my punishment level should be 100%, using your reasoning I would conclude that the Tribunal is full of people who spam pardon so as to not punish players. Of course this would be faulty logic, but it's the same that you are employing. You can't say that X% of players being punished shows anything more than X% of players were punished because there is no reference point to show anything else. And that's not even going into your incredibly low sample size, but that's another topic entirely.


And as to your point about WookieCookie using the tools available, Riot has consistently said that the Tribunal is merely one method of punishment. It is certainly not the only one used. Your proposals for what the Tribunal should be are clearly not what the Tribunal is: one of a handful of tools used by Riot to ensure that those players who negatively influence the game are dealt with.


Comment below rating threshold, click here to show it.

Gimpyloser

Senior Member

01-11-2013

Quote:
Originally Posted by OmenOfPlague View Post
I think you are right about the "spam punish" but .... if wookie can take more casses to see if a summoner is toxic or not to the comunity i think at last the tribunal do his work, taking action against a guilty summoner ( my english is bad but im traying)
You could go through about anyones account and find a game that warranted a punishment if you looked hard enough, but that doesn't mean the system is working. The only saving grace is that probably 80% of the cases presented to the tribunal deserve punishment. So by default the decisions are fair 8/10 times.


Comment below rating threshold, click here to show it.

Gimpyloser

Senior Member

01-11-2013

Quote:
Originally Posted by magooomba View Post
Why does this argument always seem to pop up? Yet again, let me explain why this reasoning is faulty. You are concluding that the Tribunal is full of a "large number of people are using the spam punish tactic for improved accuracy and rating." In concluding this, you have already disregarded a much more likely conclusion: that most people who enter the Tribunal merit punishment. The Tribunal cases are not an assortment of random people in random situations. Rather, they are players who have been reported above the threshold needed to enter the system to begin with. It would be extremely faulty to assume that these players are simply a random collection and would show an average punishment level when the method used to get this sample is hardly random at all. As such, simply concluding that X% of players are punished, thus the system is broken is completely devoid of any analysis: it's simply your presumption.

You assume that your punishment level should be less than 92%, without any reason why. If I were to do the Tribunal and assume that my punishment level should be 100%, using your reasoning I would conclude that the Tribunal is full of people who spam pardon so as to not punish players. Of course this would be faulty logic, but it's the same that you are employing. You can't say that X% of players being punished shows anything more than X% of players were punished because there is no reference point to show anything else. And that's not even going into your incredibly low sample size, but that's another topic entirely.


And as to your point about WookieCookie using the tools available, Riot has consistently said that the Tribunal is merely one method of punishment. It is certainly not the only one used. Your proposals for what the Tribunal should be are clearly not what the Tribunal is: one of a handful of tools used by Riot to ensure that those players who negatively influence the game are dealt with.
You always troll me by attempting to destroy the credibility of anything I post. In this case you are somewhat correct. The sample size of 100 cases is too small for a real and accurate study. Clearly this isn't a research project or I'd have done more work. Also I did not link specific cases in which I felt the player was unfairly punished to support the idea that the punishment rate should be less than 92%, although they did exist within the sample. I could have done more to present this topic as a more valid argument, but I did not. I may improve it later if I have time to build a case worthy of the court of Magoo.


Comment below rating threshold, click here to show it.

PenguinKillBear

Senior Member

01-11-2013

Seeing as all players showing up in the tribunal have been reported for having a toxic presence in the game, why is it wrong that 92% of them were actually guilty of just that?


Comment below rating threshold, click here to show it.

Calys Teneb

Senior Member

01-11-2013

What's hilarious is I have 92% accuracy and I pardon and punish appropriately, and I get nearly every pardon correct with the majority.

Come at me OP.


Comment below rating threshold, click here to show it.

Kabusabe

This user has referred a friend to League of Legends, click for more information

Senior Member

01-11-2013

I'm wondering what the OP expects for the % of the cases as being pardonable.


Comment below rating threshold, click here to show it.

Gimpyloser

Senior Member

01-11-2013

Quote:
Originally Posted by PenguinKillBear View Post
Seeing as all players showing up in the tribunal have been reported for having a toxic presence in the game, why is it wrong that 92% of them were actually guilty of just that?
It isn't. If indeed 92% of the players are guilty then they should be punished. In my experience it's closer to 80% or 8/10 cases I review. Of course guilt and innocence are subjective so for you 92% may warrant punishment and for me only 80%. I think the problem lies in that if you simply click punish, you will have an extremely high accuracy rating. If you are right that 92% deserve to be punished then I think that is too high and actually encourages people to simply click punish without investigating the cases. Perhaps Riot could implement a system in which they automatically generate pardon cases into the system to decrease that percentage. 1/5 cases is an obvious pardon for nonexistent player... Just a thought.


Comment below rating threshold, click here to show it.

gnfnrf

Senior Member

01-11-2013

The overall punishment rate for the Tribunal is around 76% right now.

I didn't check 100 cases to get this, I checked 10,000.

Another guy checked 14,000 and got a slightly higher value (I think closer to 80%).


12345 ... 8