Moderating language is difficult for any platform. But Valve Software is trying to take on this challenge with a new Steam feature called Chat Filtering. This enables Steam members to opt into an algorithm that obscures what Valve calls “commonly used strong profanity and slurs sent via chat.”
Chat filtering is available in beta as part of Valve’s Steam Labs experiments. And the idea is to give players control over their chatting experience in Steam Chat or in supported games (like Dota 2). And while this is a necessary and useful option, it also feels like it is potentially trying to shift responsibility away from Steam and onto individuals.
Here’s how Valve explains it:
With chat filtering, we’ve obscured the most offensive language shared on Steam. You can alter your settings to control whether profanity and slurs are displayed, and because each player’s tolerance for difficult words is unique, we’ve included the ability to add or remove words to form your personal filter. You may also upload lists of words or phrases from other sources, empowering groups and communities to work together to define and share your own sets of language guidelines. We believe this level of control is especially important given that language is constantly evolving and is used differently among various communities around the world.
I appreciate Valve’s position here, but it should acknowledge that language isn’t some personal choice. It’s a tool we use to interact with one another. And to improve a community like the one on Steam, it should worry less about the words themselves and more about the people using them to hurt others.
Words cannot hurt me
Steam’s Chat Filtering uses two lists to enable users to block words. One is a list of general profanity. This is where you’ll find variants of “fuck” and “shit.” Valve found that over 56% of instances of profanity or slurs was a variant of “fuck.” The other list includes various slurs.
It’s a good thing for Valve to give more control over to people. Maybe you want to play with your kids, and you don’t want them picking up the verbal habits of Counter-Strike players. You should have the option to shape what your chat looks like.
But why is using slurs even a choice someone can make? Well, Valve has an explanation for that:
While we do ban profanity and slurs from being displayed in more public places like user reviews, comments, forums, and broadcasts on Steam, we do not want to censor users in chat, but rather, empower them to choose what they see from others. We know marginalized groups can reclaim language for themselves, and we don’t want to stand in the way of enabling groups of Steam users from doing so when chatting with one another on Steam.
Again, I appreciate this. I don’t want Steam to ban me because I called a friend a “honky” in a knowing, joking way.
But the idea that the words are the problem isn’t accurate. People who deal with systemic racism all their lives are rarely so fragile that just seeing a word can upset them. The hurt comes from knowing that a hateful person gets to continue to share a space with you free from consequence.
Moderating chat is not like moderating the Steam store
It might help diffuse situations if I can’t see that someone is using a specific racial slur. But I doubt it. Racism often comes with some glaring context clues. And again, the problem isn’t the words themselves — it’s any targeted harassment and hatred. And I get there is a risk for false-positives when attempting to programmatically moderate millions of interactions. But it seems like if I report someone for hateful behavior, and the algorithm detects slurs in that interaction, maybe we should punish that player.
But Valve’s stance on chat is similar to how it approaches moderating the Steam store. In 2017, the company stopped permitting admission onto Steam based on content. Valve said it shouldn’t get to pick and choose the games people have the option to buy. Customers are, usually, adults, and they should get to decide how to spend their money.
And you can see how that sort of thinking might make sense for community interactions. But it’s not the same thing.
When I buy a bad or offensive game on Steam, I have some matter of recourse as a customer. I can get a refund or write a review. I also made the explicit decision to interact with that game in the first place.
Chat does not work the same way.
I did not choose who to interact with in a Dota 2 matchmaking session. Steam’s systems made that decision for me. And it should, therefore, take some responsibility for what happens within. After all, Valve and developers on Steam want me engaged with chat because that leads into more enthusiasm for the game, which could then increase my lifetime value as a customer.
Chat filtering doesn’t really minimize toxicity. If I don’t like seeing racist bigotry, filtering out certain words isn’t going to make me feel better if the offending player continues unimpeded. All that does is make me want to concede yet another space to the loudest and worst people.
Register for GamesBeat's upcoming event: Driving Game Growth & Into the Metaverse