League of Legends test-running new auto-ban bots

League of Legends now has a new player reform system that will automatically ban offending players.

A highly popular game like League of Legends has the potential to attract players from far and wide. As a side effect, player harassment is also likely to increase as the player base does. A new automated system was recently rolled out on North American servers for initial testing to combat such behavior. The system is specifically aimed at tackling homophobia, racism, sexism, death threats and other forms of verbal harassment.

The system depends on player reports to work, so it does not screen chat unless prompted to by a player report. If a player participating in the game sends a report, the player reform system checks that the report is not unfounded, likely using keywords to help search the chats for abusive behavior. The case is then examined if found to have merit, based on the community’s own standards of behavior.

From racist remarks to ban hammer in 15 minutes

When the system finds a punishable offense, it then sends a reform card through email, sharing the chat log of the harassing player and the punishment. Other players’ names and chat logs are scrubbed from the log. Ideally the system would process the whole case in about fifteen minutes, either rejecting it or doling out a punishment of a two-week ban or a permanent ban.

For the first thousand or so reports the staff will be hand-reviewing each to ensure that cases are running smoothly, and there are not a high amount of false positives. False positives are thought to be in the 1 in 6000 range. 

As far as policing the behavior and speech of players, I believe that one should refer to the Terms of Use first before complaining about a violation of purported “freedom of speech.”  In section V, Code of Conduct, the Terms of Use specifically outlines that “harassing, stalking or threatening any other users in the Game,” as well as “transmitting or communicating any content, in the sole discretion of Riot Games, is deemed offensive, including but not limited to, language that is unlawful, harmful, threatening, abusive, harassing, defamatory, vulgar, obscene, sexually explicit, or racially, ethnically objectionable,” warrants disciplinary measures. 

Riot's game, Riot's rules, Riot's machine

As a commenter on Huffington Post aptly said, “Go into a restaurant right now, run around the place, swearing and screaming at people, and see how fast you get thrown outta the place.”

However, letting machinery determine whether or not a post is offensive could be prone to some bugs and possibly accidental bans. Is there not some potential for abuse of the system? Machines do not, at this point, have the same type of discretion that humans do, and will flawlessly execute commands based on the variables they are programmed for.

While removing abusers from a community is admirable, making sure that people are not falsely banned is equally important. Hopefully, Riot Games has this in mind as they move forward with their new system.

Published May. 26th 2015
View Comments

New Cache - article_comments_article_23143