Behavioral Systems: August 2020 Update

Sharing our changes coming in the next few patches.

DevAuthorRiot Codebear
  • Copied to clipboard

Hello everyone! We're back with another monthly update on our progress toward improving League's behavioral systems. As usual, we would like to give you some insight into what the team has been working on this month and what you should see rolling out across the next few patches.

In ourJuly update we called out a few specific areas around disruptive behavior that we would be focusing on: Intent to lose actions and leaving or going idle in game.

These include:

  • Intentionally feeding opponents
  • Sabotage through disruptive gameplay actions (aka “griefing”)
  • Intentionally leaving the game
  • Going idle in game
  • Game disconnects

Our goals are to:

  • Reduce occurrences of the above behaviors in all queues
  • Mitigate the impact to players when they experience disruptive behaviors in their games

Intentional Feeding and Idle/AFK Detection

After 100s of depressing manual VOD reviews and verification exercises, we think we’re in a good enough place to start getting improvements to how we detect intentional feeding and idle/AFK behaviors out for testing. These changes will allow us to:

  • Identify these behaviors at DOUBLE THE RATE we are currently
  • Take action against those negative behaviors more consistently
  • Let players know when we’ve caught negative behaviors in their games and work towards alleviating that pain

Our first detection improvements are already live in 10.15, but we’ll continue to push new iterations across the next couple of patches. We’re working towards a world where we detect these behaviors as they occur, mitigate them for affected players during those games, and take action against the offenders immediately after the game.

Idle and AFK Actions

If we're detecting disruptive behavior more often, how are we actioning against it? In order to reduce the overall footprint of idle and AFK behavior in League, we've outlined a few options that tackle the goals of:

  • Preventing this behavior from occurring again.
  • Deterring this behavior from occurring in the first place.
  • Mitigating the impact of this behavior when it does occur.

We’re looking to dig more into the following solutions to address AFKs in the short-term:

  • Early surrender options when we identify an AFK early in your game
  • LP mitigation for those in ranked games that have an AFK or intent to lose behaviors on their team during a loss
  • Stricter Queue lockouts for those that disconnect / AFK
  • Pre-queue warnings for those that have disrupted games previously

We’re looking for feedback from players on these implementations, so we’ll be doing them in the same live experimentation fashion as we did with Champ Select Reporting. We don’t have a hard date for delivery just yet, but we should have at least one experiment live in the next month.

Speaking of Champ Select Reporting...

Champ Select Reporting

Two patches ago we launched Champion Select Reporting and Muting features. We’ve been able to gather significant data around disruptive behaviors in Champion Select and have seen success in giving players a pre-emptive way of avoiding verbal abuse in Champion Select and in game. Rather than turn these features off after the two-patch experiment as originally planned, we'll leave them on going forward.

The two biggest trends we saw in reports were players whose champs were banned by their own teammates and reports of negative chat. Though we know these aren’t all the issues, it’s enough for us to start developing improvements based on our experiment's results.

We’re committed to shipping the following to further develop the Champion Select experience:

  • Ban confirmation and detection: We understand that there's strategic value and personal preference for certain bans, and we want to maintain your ability to pick which champ you don't want to play against. Based on report data, though, we want to make it clearer when your ban choice overlaps with what a teammate wants to play. We’ll be implementing better automated detection to double-check that you actually want to ban a champion that your teammate is currently hovering.
  • Penalty System Integration: Though we can’t confidently act on behaviors directly in Champion Select just yet, we are looking to find and implement the best post-game actions we can to common issues, similar to End of Game reporting.
With each experiment we run, your feedback and experience with these new features helps guide the direction of League's behavioral systems. Thanks for sticking with us as we work to make playing League a better experience.

As always, see you on the Rift.


  • Copied to clipboard

Related Articles
Related Articles