Why are there three people who all reported each other in their last game on my team?

Yenn·Oct 6, 2017·2 votes·578 views

Can we please get some improvement to 100% preventable toxic bullshit like this? It's sad that this can even happen.

12 Comments

DrugothOct 6, 20174 votes

If players could blacklist each other from being matched together just by reporting each other, queue times would gradually get worse and worse over time, the higher elos being more radically effected than lower, because of a smaller pool of players.

YennOct 6, 20171 votes

{quoted}

If players could blacklist each other from being matched together just by reporting each other, queue times would gradually get worse and worse over time, the higher elos being more radically effected than lower, because of a smaller pool of players.

There can be limitations on how long the queue can go without putting two players together. You're acting like this isn't a solvable problem.

ModWulf HelhammerOct 6, 20173 votes

You complain about autofill, and then you want queue times to get longer.

Me thinks you don't understand what you're discussing.

TrulyBlandOct 6, 20173 votes

Riot already said they won't do that, because it would bring extreme increase queue times and potential for abuse in lowly populated MMR areas.

KartagiaOct 6, 20172 votes

{quoted}

Riot already said they won't do that, because it would bring extreme increase queue times and potential for abuse in lowly populated MMR areas.

Thus RIot has said "we do not want to do anything making game less toxic, as it would lengthen queue times for rules violators"

TrulyBlandOct 6, 20172 votes

Obviously it would lengthen queue times not only for people who are blocked/reported a lot, but also for those who use the feature a lot, effectively punishing people for blocking/reporting. You also assume that people who are blocked/reported actually are rules violators, which is not necessarily true. And lastly you completely ignored the point made about potential abuse, since this kind of feature would allow people in lowly populated MMR ranges to influence who they will be matched with.

KartagiaOct 13, 20171 votes

{quoted}

Obviously it would lengthen queue times not only for people who are blocked/reported a lot, but also for those who use the feature a lot, effectively punishing people for blocking/reporting. You also assume that people who are blocked/reported actually are rules violators, which is not necessarily true. And lastly you completely ignored the point made about potential abuse, since this kind of feature would allow people in lowly populated MMR ranges to influence who they will be matched with.

Yes, I do assume blocking someone has good reason. That is usually either intentional feeding, or total toxic behavior. And lengthening the queues is not that bad. For me it is just lame excuse for RIot to ignore the Summoner's Code for profit. "You cannot be queue with someone you have blocked" would give good signal to players that they can avoid at least some of the people ruining their games. THe MMR system does not work like you believe, as the range of MMR allowed in match making is too large. IT does not also work as silver overlaps with bronze 5.

TrulyBlandOct 13, 20171 votes

{quoted}

Yes, I do assume blocking someone has good reason.

And my point is that this is a silly assumption to make, and I explained why.

THe MMR system does not work like you believe, as the range of MMR allowed in match making is too large.

"Too large" for what? Any defined interval of MMR will at some point only include a low population, that's in the nature of the normal distribution that it (roughly) follows.

To make this really mathematical: For any fixed interval size i and a value e (no matter how small), there exists an offset-MMR c for which the interval [c, c+i] contains less than e percent of the player population.

Less mathematical version: Name any percentage. There's a MMR-area that includes a lower percentage of players than that in the MMR range it uses for, say, the first ten minutes of searching for a match.

IT does not also work as silver overlaps with bronze 5.

Entirely irrelevant, because matchmaking is purely based on the matchmaking rating, not on ranks.

KartagiaOct 18, 20171 votes

{quoted}

And my point is that this is a silly assumption to make, and I explained why.

"Too large" for what? Any defined interval of MMR will at some point only include a low population, that's in the nature of the normal distribution that it (roughly) follows.

Honestly, mathematically use of average is silly and stupid idea, as average dos not contain any real information. Median is useful, and the fact that the skill level difference matters mean that large gap between players is worse than small gap, and average does not handle this. Most of the time in CS we used average squared as it emphasized the distribution side to meaningful level.

It is way more pleasant to wait for game than dodge game due someone you really do not want to play with in your team. Thus current system gives grievers more tools and more power, as they are not punished for making situation such that someone else has to dodge. That is actually worst form of grieving, as the victim is punished for dodging with longer wait time. The perpetrator is never punished, as Riot has not been able to implement pregame chat reporting system "they have been working on since season 2". Either totally incompetent or totally dishonest from Riot Games LLC.

To make this really mathematical: For any fixed interval size i and a value e (no matter how small), there exists an offset-MMR c for which the interval [c, c+i] contains less than e percent of the player population.

Normal distribution only works on really large sets, thus small number of players ruin this. Do not use statistics unless you know the axioms and premises of its use, honestly. Statistics work only on large sets of data. And by large it mean tens of thousands of valid members.

Less mathematical version: Name any percentage. There's a MMR-area that includes a lower percentage of players than that in the MMR range it uses for, say, the first ten minutes of searching for a match.

Naming such percentage is mathematically irrelevant. I have waited for games for ten minutes, and I would anytime wait for 20 mins rather than play 20 mins game which one player is intentionally ruining for his own fun, and getting away with it due RIot fear of DDoS.

Entirely irrelevant, because matchmaking is purely based on the matchmaking rating, not on ranks.

TrulyBlandOct 18, 20171 votes

Your entire premise seems to be "averages only work on large numbers, and LoL isn't large enough for that". And in doing so you completely invalidated your original point of

THe MMR system does not work like you believe, as the range of MMR allowed in match making is too large.

I mean... I don't actually agree with any of what you said, but either way the conclusion is that there are small populations within LoL.

But just to cover one of the major points anyway: MMR works on these assumptions and is given out on these assumptions. Just like IQ distribution is defined to be a normal distribution, so is the Elo system. And I'm fairly certain that even though a lot about the system has changed in the process of becoming MMR, I doubt they changed something this elementary, especially because all three of these share one very important principle: Measuring something that cannot directly be measured. You cannot measure intelligence and you cannot measure skill. You can measure your likelihood to complete tests or win games and map that on a normal distribution.

So that leaves us at the same conclusion as earlier: There are extremely lowly populated MMR ranges.

Naming such percentage is mathematically irrelevant. I have waited for games for ten minutes, and I would anytime wait for 20 mins rather than play 20 mins game which one player is intentionally ruining for his own fun, and getting away with it due RIot fear of DDoS.

Apart from the fact that I doubt that everybody would agree with this stance, you have forgotten the actual point: ABUSE. People could abuse this to increase their odds of increasing MMR/LP/Rank.

You also still have not solved the issue of, in similar ranges, people blocking others simply because they don't like their playstyle or for teh lulz, resulting in extremely long queue times for people who did nothing wrong. Given that this system would function unsupervised (in the sense that "decisions" made by it are not confirmed by a human) it would allow people to punish people for literally any reason they want as long as enough people agree on it.

You can even combine all of these points: What if everybody above a certain rank decides that the current person with rank #1 is not worthy of that rank and as such decides to unanimously block them? No more games for that person, I guess.

KartagiaOct 20, 20171 votes

{quoted}

Your entire premise seems to be "averages only work on large numbers, and LoL isn't large enough for that". And in doing so you completely invalidated your original point of

I mean... I don't actually agree with any of what you said, but either way the conclusion is that there are small populations within LoL.

But just to cover one of the major points anyway: MMR works on these assumptions and is given out on these assumptions. Just like IQ distribution is defined to be a normal distribution, so is the Elo system. And I'm fairly certain that even though a lot about the system has changed in the process of becoming MMR, I doubt they changed something this elementary, especially because all three of these share one very important principle: Measuring something that cannot directly be measured. You cannot measure intelligence and you cannot measure skill. You can measure your likelihood to complete tests or win games and map that on a normal distribution.

And Mathematically false assumption invalidates whole system. That is why I test assumptions first. That is how I have been taught to test hypothesis in university. First check assumptions and axioms (basis of the hypothesis). If they are invalid, the whole hypothesis, and the system, is invalid.

IQ is not defined by normal distribution. All people believe for IQ has been proven false. IQ is not constant. IQ can be taught, and you can learn to improve your IQ. I suggest you do check your facts before you use them.

MMR system has several really bad assumptions, but use of average as skill level of team is totally stupid. Main reason for this: role weights the skill impact. Due community way of treating supports, support skill level has way less impact, as support skill is tactical and team work, and if team does not listen support, his skill level is irrelevant. More often than not, the team goes intentional losing when support does not do things he is told to do, even if he argues with facts they are not proper way of doing things. Almost none of adcs appreciate support opinion.

So that leaves us at the same conclusion as earlier: There are extremely lowly populated MMR ranges.

Actually high population MMR ranges is bigger problem. High population on bronze/silver area has put RIot into big problem - Silver 5 overlaps with bronze 4 or 5. Thus league system does not work at all, as it has no effect in matchmaking, and thus it is no proof of skill.

THe matchmaking has also one major flaw which Riot has hidden due PR reasons. This information comes from old forum dev post. Matchmaking allows one team to have predicted win chance of 75% . I have not seen any facts proving that Riot can alter the matchmaking system. Quite contrary things seems to be like they cannot do that. Thus league system cannot work when you have quite good like hood you get 10 games in a row with 75% loss chance prediction. It is not taken into account at all in LP loss calculation.

I have also stated out that 4 vs 5 game should be considered when doing MMR update as real difficulty of the game. Riot have stated out it could be abused without stating out how. The fact is that if you make game harder for you, and you can win 4 vs 5, you should be awarded more. I have suggested long time ago that the leaver should not get anything from won, and take the difference of 5 vs 5 MMR change and 4 vs 5 MMR change as punishment. This would mean that leaver would always lose MMR, and would always lose more MMR than his team buddies. You could abuse it with hacked accounts like people are doing at present to intentionally lose games.

Apart from the fact that I doubt that everybody would agree with this stance, you have forgotten the actual point: ABUSE. People could abuse this to increase their odds of increasing MMR/LP/Rank.

People do already abuse this. Grievers and malicious part of gaming community already abuses MMR system by intentionally losing games. If system is well designed, and has solid foundation, it cannot be abused. Abuse always indicate system is flawed. This is hard to admit by people who think they made good system, just like Microsoft did with Windows NT decryption system. That system was proven to be totally useless by cryptoanalysts months later.

Assumption is mother of all fuck-ups.

You also still have not solved the issue of, in similar ranges, people blocking others simply because they don't like their playstyle or for teh lulz, resulting in extremely long queue times for people who did nothing wrong. Given that this system would function unsupervised (in the sense that "decisions" made by it are not confirmed by a human) it would allow people to punish people for literally any reason they want as long as enough people agree on it.

I have suggested solutions, which are mathematically stable. Riot has refused them. I have given them on season 2. First suggestion: Stop using average, but use average squared for skill level of team. This is mathematically way more solid. It prevents heavy spread of MMR within team, thus making teams with more equal skill level.

I have suggested that individual has different MMR for duo games for every single duo combo they have. This would remove the need for arbitrary "duo gest some benefit from duo at high level" MMR boost of duo MMR without any solid basis. Contrary RIot has admit that low MMR does not gain any benefit from duo playing, but Riot has not been able to add single IF-clause to the matchmaking removing the MMR boost from duos if both players are on low MMR.

You can even combine all of these points: What if everybody above a certain rank decides that the current person with rank #1 is not worthy of that rank and as such decides to unanimously block them? No more games for that person, I guess.

For such cases we should have RIot games support. Such bullying operations should be handled case by case by human arbitrator and bullies should be punished, unless it is proven that the person has been blocked for a good reason, and should be banned instead.