/dev: Tackling AFKs
Discussing new changes to reduce the frequency of AFKs.
This week I figured it would be good to focus on inting/afking, given it’s been a topic of quite a lot of conversation recently. It’s also something we should be talking about more, since it’s an ongoing pain point for many players.
TL;DR : Deliberate game-ruining behavior is something we need to prioritize addressing more. In the near future we’ll improve how report feedback works and introduce reporting in Champ Select. In the moderate term we’ll test more responsive detection of deliberate inting/pseudo afking. We're still considering steps after that but the plan is to keep working on this longer term. We’ll keep you updated each month on how internal progress and tests being run in specific regions are going.
First off, let’s define exactly what we’re talking about:
Ok, so here's where are our heads are at:
That’s pretty general, so what actual changes are we planning?
We’ve just started trialing improvements to how our report and notification systems work. For now, those are live on NA only, if this test goes well we’ll roll them out worldwide in a couple of weeks. Those should result in a significant increase in the number of punishments being surfaced.
One component of that is changes to how punishment notifications are generated. Previously you’d only be notified that a user you'd reported had been punished if that punishment was triggered immediately after the game you were reporting them for and the punishment was for the specific category you’d reported them in. The actual punishment system, however, looks at a player’s games over time and aggregates reports from different categories together. As a result, notifications were substantially lower than actual punishments. The revised notification system, by contrast, will notify you if a user was punished regardless of category of report and will do so if the player gets punished after any of their next dozen or so games.
Disruptive behavior in Champ Select is a problem that players have very few ways to deal with. Starting around late Q2 we're going to give you the ability to report disruptive players in Champ Select. First, these reports will be used to establish a data foundation for champ select behavior. Then, once we've got enough data to identify different types of behavior accurately, we'll deploy a punishment system.
After the above changes, we want to go back and reexamine our previous stance on rapid, automated detection of players who are trying to lose a game. We’ve been very cautious about it historically due to the risk of falsely identifying, and therefore punishing, a player who isn’t actually trolling. Avoiding unjust temp bans is important to offer a good player experience. Having said that, so is controlling deliberate trolling, self-sabotage, etc. In retrospect we may have prioritized avoiding one type of player pain too much at the expense of another type, so want to look at our options here again.
One thing that would be helpful as part of that is understanding where your heads are at as LoL players. What’s an acceptable rate of incorrect bans if it results in a noticeable drop in deliberate inting/afking? Is it worth accidentally giving a two-week ban to one player who was genuinely trying if that means 19 trolls also get banned? What if the ban is undeserved 1/100 times? 1/1000? Are incorrect bans never acceptable at all?
Beyond that, we’re still assessing which approaches to investigate next. We plan on doing more work in this space than just what’s listed above once we’ve got some of those projects shipped. We’ll update you folks on how things are going within a month.