League of Legends news » Initial Honor Results

It’s been over one week since we launched the Honor Initiative, one of the largest experiments ever conducted at Riot. Today, I’d like to take some time to share a few results we’ve seen, and talk a little more about how we’re preventing Honor spamming.

Honor So Far

As I mentioned before, we’ve historically reacted only to toxic behavior through systems like the Tribunal, and the Honor Initiative was our first attempt at positive reinforcement in League of Legends.

The Honor Initiative is a multi-phase experiment, and in the first phase, our goal is to see if we’re able to accurately identify the positive pillars of our community. Secondly, we hope to see if it’s possible to positively nudge “neutral” players by making it easier to be good instead of toxic. Already, we’ve seen some dramatic, global results in normal and ranked solo queue reports the past week:

• Negative Attitude reports: -29% in normals and -11% in ranked
• Offensive Language reports: -35% in normals and -20% in ranked
• Verbal Abuse reports: -41% in normals -17% in ranked

Preventing Honor Abuse

One of the major concerns we took into account when designing Honor was abuse – it’s obvious this was a concern some of you shared: “Dr. Lyte, people are going to trade Honor! This is going to be meaningless!” Rest assured, Riot has spent a lot of extra time implementing anti-abuse features into the Honor Initiative. We tasked nearly half a dozen PhDs and statisticians to this problem and let them tinker with models for weeks. What did they come up with, and what’s going on in the background?

• Pattern recognition: Honor was designed to recognize patterns. For example, Honorable players who are true positive should not be reported very often.
  • As an example, the system caught a player who was reported in 100% of his games after Honor launched, but he also got Honor in nearly every game. This obvious trader was caught.
• Social interactions: Honor was designed to analyze interactions between players. For example, the system analyzes the players who are Honoring you and the players you give Honor to in an effort to determine the quality of these exchanges.
  • In another example, the system caught a player who received 100% of his Honor from a premade friend. Both the giver and the recipient were also toxic players and both were punished for this behavior.
• Honor values: As mentioned above, Honor assesses the value of your exchanges. Receiving Honor from a stranger you’ve never played with before is valued more than Honor from a player who has played with you 100 times and Honored you multiple times. Design decisions like this were made to encourage players to be positive across a large number of games with a large variety of people—in fact, for the min-maxers, this is the optimal way to earn Honor.

So far, these abuse-prevention measures we built in have been highly effective:

• Only 0.21% of players reported for consistent trading
• 0.63% of players actively trying to trade Honor
• 0.01% spamming Honor for dishonorable reasons

These players have all had their Honor reset to zero.

What’s next for Honor?

We’re deep into our planning for the next phase of Honor and we feel that this is just the beginning of the Honor Initiative. Even now, there’s more content to discover for the truly Honorable player! I hope players have been enjoying Honor so far, and together, let’s inspire others to be more Honorable. Let’s not only make a difference in League of Legends, let’s make a difference in online gaming.