Boards›Player Behavior

Trolling Experiment (why many trolls don't get banned)

Ręmędy·12/25/2016, 9:02:44 AM·10 votes·7,538 views

WARNING before you read more, this may trigger some sensitive people. Sometimes evil must be committed to destroy/reveal the greater evil.

I have used about 10+ alternate accounts and about 100+ games to conduct this experiment.

For those complaining about statistical significance, were you seriously expecting an average league player to have the ability to troll significantly? Think before you speak.

Theory:

My theory is that reports of any kind don't affect players if they manage to play the game somewhat even if they were to intentionally feed and lose you the game. It would not matter as long as they satisfy these 4 easy conditions:

  1. Have some kills + assists. (third factor that helps solidify your involvement in the game)
  2. Maintain high levels and gold, or equivalent to your teammates. (secondary factor to disable intentional feed reports as you are actively trying)
  3. Do not die as soon as you respawn. Space out your deaths between 2+ minutes. (possibly the main factor that disables intentional feeding reports)
  4. Don't be toxic in chat, or don't talk at all. (renders negative attitude and verbal abuse reports useless)

Trolling = Intentionally feeding and playing without the intention to win. If you want to avoid bans: (Get kills, assists, and gold to counter the reports) Remember that many trolls have these, not the obvious trolls that want to get banned. Average trolls rarely get banned if ever.

Testing:

All accounts played multiple games which included what is typically called 'trolling'. Several kills and assists were obtained. Levels and gold were equivalent to the team members. Never talked in chat. Spaced deaths between 2+ minutes minimum. Ran to lanes to steal cs and dive at their turrets. Stole jungler's camps.

Remember that these conditions you need to satisfy are common with all average trolls. This is not me deliberately trying to avoid the system. I'm trying to replicate the common trolls.

Some pictures of the accounts that have been through placements with trolling: https://gyazo.com/1347d74ecfb4fcf677a39f9c539bb1ce - 10 game loss. https://gyazo.com/b5a73c5281649f3653c57bc3152ba3f0 - 2-8 win/loss due to me accidentally trying to win. xD

Results

Accounts were tested for bans after a week or two. All accounts tested in this experiment successfully avoided bans and chat restrictions of all types. It is not necessarily that the report system is flawed.

Trolling is at an all time high, but I don't think anything could be done to prevent it as it's too easy to mistake it with the player being bad. Obviously, the toxic trolls will get banned, but not the ones that are smart about it and satisfy the theory above. Maybe Riot can mess with the factors above to improve their punishment system, but I highly doubt it will be possible as a player can fake being bad, and that is nearly impossible to differentiate from a bad player.

Say even if the reports were to work after awhile, it wouldn't be effective as the troll could of possibly ruined over 500+ games.

I believe that Riot doesn't conduct real reviews with real people with the report system. If you think about it, why would they waste money hiring expensive staff just to keep some measly trolls out of the game rather than investing it in obtaining more profit? I think that the tribunal may be a solution, but I'm not sure. Certainly, an automatic system is no way to effectively keep the game healthy. Maybe it just isn't possible to stop trolls after all.

64 Comments

EvilDustMan12/25/2016, 11:56:13 AM9 votes

You lost all credibility at the first paragraph where you already set yourself as combative instead of impartial.

Deep Terror Nami12/25/2016, 9:19:46 AM9 votes

Doesn't seem like you know much about how the system works. I recommend reading up, starting with this material;

https://support.riotgames.com/hc/en-us/articles/207489286-Instant-Feedback-System-FAQ-

http://boards.na.leagueoflegends.com/en/c/miscellaneous/yFLxpAcV-qa-about-player-behavior-systems-chat-log-reviews-and-feedback

http://na.leagueoflegends.com/en/news/game-updates/player-behavior/reports-how-do-they-work

http://forums.na.leagueoflegends.com/board/showthread.php?t=2411284

http://boards.oce.leagueoflegends.com/en/c/announcements/L65wrRYe-player-behaviour-faq-updated-for-instant-feedback

ModUlanopo12/25/2016, 4:49:55 PM7 votes

See, it's garbage like this that makes peer review before, during and after experimentation an absolute must. You need to learn how to science.

  • Your sample size is too small to have any meaning, especially when it's just you, which introduces bias.
  • Speaking of bias, your thesis is composed of nothing else. Well, that and unfounded assumptions.
  • Your methodology - deliberately obfuscating the fact that you're "trolling" - is designed to produce a result, not test the hypothesis.
  • Your methodology is an undefined hodgepodge. Are you testing community response to verbal toxicity, intentional feeding or both? If it's verbal toxicity, how did you correct for perceived justification (cultural permissiveness tends to increase in conditions of provocation)? How did you establish the context of your toxicity (being toxic in a random way will affect reactions)? If you fed, did you wait for a circumstance in which it could be mistaken for being outplayed or making a misplay? If you were testing people's reaction to feeding, why did you make it look as little like feeding as possible and why did you apologize?
  • You don't have a baseline from which to measure the effect of your trolling.
  • You don't establish a control group from which to measure regional differences.
  • You don't understand the difference between data and outcomes. NASA knew about ice damaging tiles during shuttle launches long before Columbia exploded, but made the mistake of believing it wasn't a problem because it hadn't exploded the shuttle previously. You're doing the reverse: inferring data from outcomes, which is just as flawed.

I could go on, but I'm pretty sure you get the idea. Don't make me pull out the Bill Nye memes.

Randomonium12/25/2016, 9:39:38 PM6 votes

I honestly don't there there will be a full proof way to separate bad players from those who troll "intelligently" like you did. Most intelligent people realize this. I don't really see the point of ruining 100 games just to prove that the system isn't perfect.

The problem with the original tribunal system was that people wouldn't look at the chat or the context and just auto punish people. If you proposed a way how to solve that issue then you might be onto something. That would have actually been a useful post and wouldn't have required you ruining 100 games.

Anomander12/25/2016, 5:45:18 PM5 votes

  1. Were these new or established accounts?

  2. What actions were "trolling"?

Your summary lacks a lot of information.

Trolling and trying to win is different then trolling and trying to lose.

Your sample size is a bit limited per account.

Riot has a lot more tools and a much larger sample size. Keep in mind it will be tuned for leniency as they don't want to punish people having a bad couple of games.

The only reason to perform this type or "study" is to troll. It lack any credibility and can actually result in your permanent removal from league.

If you want to conduct experiments within a game you should contact the company first as you are infringing upon their product.

Personally I think that Any accounts used in this study should be permanently banned as you have admitted that they trolled over multiple games in order to abuse the system and ruin other players games.

IcyPepper12/25/2016, 10:13:09 AM5 votes

Can you not.

Kei14312/25/2016, 11:31:59 AM2 votes

did you have a control where you've disabled either (or multiple) of those factors?

What is your sample size?

How do you define "trolling"?

were all 10 accounts doing the same thing? If so what is the point of spreading the games over only 10+ games per account instead of seeing if you'd get banned over 20+/30+/40+ games?