Player Behavior Design Values: Reform (More Context)
Here is the Original Blog Post.
Lyte has provided some more feedback on this one. See below!
"That is exactly my point. The reduced reports don't mean reduced toxicity. It could be reduced reports because the season's over, because people play less now that it's near Christmas, because finals happened, or less reports because people were legitimately removed. Who knows which it is. And personally, I don't think that the reports correlate to toxicity at all, since nobody ever gets reported or punished for toxic behavior, just toxic words. Do whatever you want, as long as you don't speak."
#Lyte:
We often see feedback that we don't provide enough context in some of our blogs or posts about player behavior, and that's valid. However, we generally don't go nitty gritty on a piece that millions of players are going to read because only a small % of players are going to care about the academic rigor behind the statistics.
In academic talks at universities or GDC, you'll often see me go into pretty extreme detail on some of the numbers behind experiments, and you'll want to look up those if you are interested in the nuances.
To answer your question, we don't just measure reports when measuring toxicity. We have numerous ways of measuring toxicity including sentiment analysis, which analyzes chat logs in real time to determine the level of toxicity in a particular game's chat log. In this particular blog post, we're referring to the fact that extreme toxicity (for example, specifically games where someone uttered a racist, homophobic, sexist remark or death threat) went down. We can measure this through sentiment analysis which crawls chat logs and identifies whether these phrases show up in the game.
To address your concern about seasonal changes, actually that's a pretty simple problem and numerous analysts normalize out seasonal changes by comparing any influxes they see from previous years and compensating for that in the current report. For example, we generally see changes in reports around September. For people who are currently students, that shouldn't be a surprise :)
Finally, to your last point, verbal toxicity and gameplay toxicity are both punished--you don't need to speak to be punished for gameplay toxicity. I am not sure why you believe that reports only correlate to verbal toxicity. Intentional feeders, Elo Boosters, RAF abusers, etc are all banned daily from a combination of reports and other systems.
"I got Ranked restricted yesterday and had to play Draft Pick, its terrible. The amount of toxicity is incredible. I tend to flame in heated up situations and am not an angel myself, but the amount of verbal harassment in Draft Pick at the moment is insane even for my standards."
#Lyte:
You mentioned you tend to flame up in heated situations, and I think that's part of the problem you are experiencing. We do find that players who might be toxic themselves experience more toxicity in their games, and actually do create more toxicity in the players they play with.
"I wonder what the percentage increase in toxicity in normal draft was due to all the toxic players being forced to play that. I mean it makes sense that toxicity is lower in ranked when you force all the toxic people to play another game mode."
#Lyte:
We're monitoring Normal Draft extremely closely to ensure that there isn't a meaningful increase in toxicity.
The key that makes this system different than traditional Prisoner's Island designs where you place all toxic players on a separate matchmaking island is that the majority of players in Normal Draft are still neutral to neutral positive.
By slowly and intentionally introducing a very small minority of players who may be toxic, and ensuring that they are also chat restricted allows us to ensure that Normal Draft never actually becomes a Prisoner's Island.
One of the key things we're also watching out for is how do players behave in Ranked Mode after being through a ranked restriction? Is the behavior better than before, or worse? So far we're seeing pretty positive signs that the feature is doing well.
Player Quotes: "I know of numerous people that report everyone, in every game, just to f*** with the system." Player Response: Wouldn't this just flag their account to ignore their report submission. Pretty sure Riot has said before that false reports do that to you.
#Lyte:
This. If you report inaccurately, your reports just end up worth nothing to all systems in the game.
"They don't show any data. They just make claims. Without their data being public who knows what the hell they are talking about. Besides, with Riot's PR department, you got to take everything with a grain of salt."
#Lyte:
Feel free to come to one of our university talks or GDC talks if you are in the area. Last year, we did a tour of Boston universities such as MIT and Harvard, and this year we'll be doing MIT/Harvard and adding some schools such as UCLA/USC/UC Irvine potentially.
If you want to see the full blown experiments and data behind the studies you'll have to come to one of those talks, because that's the audience that we share that level of detail with.
