Back in 2017, I hosted a webinar called Preparing for Breaking News & Trending Topics. In it, I spoke about my time moderating large online communities at Disney Interactive, and the importance of staying on top of pop culture and culture-defining events both large and small.
In 2017, I spoke about the tragic events in Charlottesville as a cultural touchstone; an example of platform operators having to make difficult decisions about how to let their users process and discuss the attack. I shared a six-step protocol that Community Managers and Trust & Safety professionals can follow to ensure that their team is prepared to handle breaking news and trending topics.
While the COVID-19 pandemic may not be breaking news, it is an ever-evolving global event, and everyone is talking about it online, regardless of the platform. We’re seeing COVID-19 chat in mobile games, kids’ platforms, teens’ social networks, and MMOs.
With that in mind, I hope you find this six-step protocol to monitor COVID-19 chat on your platform valuable.
1. Compile vocabulary
The first step is to compile a list of words and phrases that you expect to see the community use. We’re going to use the term COVID-19 as a starting point. Obvious examples include:
- alcohol wipe
- border closing
- confirmed case
- hand sanitizer
- social distancing
- world health organization
- centers of disease control
You’ll want to ensure that you’re watching for these words in your community – and in particular, how they’re being used. Is the community simply sharing their experiences with the pandemic, or are they harassing each other and potentially spreading misinformation?
The next step is to go beyond assumptions and review how your community is actually chatting.
Are they using words and phrases that you didn’t account for in your original list? Are there common misspellings? On the internet, language can change within a matter of hours. New compound terms including “covidvacay” and “coronacation” have come out of the pandemic and this rapid adoption of languages shows no signs of slowing down.
As you go through this process, it’s critical that you and your moderation team ask yourselves difficult questions, including:
- Is quoting what could be construed as dangerous/hateful speech (kungflu, Chinese virus, wuflu, etc) acceptable for the purposes of discussing it?
- When does humor cross the line?
- How will you handle misinformation and the spread of fake and potentially dangerous news? Do you need to update your content moderation policies?
In a quick, 5 minute sampling of a single hour of chat across a variety of online communities, we saw COVID-19 referenced in a variety of different ways (spelling and grammatical errors included):
- “what if the coronavirus is fake and its part of the placebo effect”
- “so dont meme corona”
- “my grandpa died of Corona rlvirus”
- “and i have no shifts at work to pay them back cus of corona”
- “it depends on if its a serious conversation. joking about corona has become an offense. which personally i find ridiculous. who are we without our jokes”
- “well my mom is staying with me until the covid dies down so i can’t play games during the week until after 10pm -_-“
- “the whole world got corona not just Italy”
- “noone was expecting to get covid 19”
Another thing to consider is languages other than English. For example, in the Dutch language diseases are commonly used for bullying. Our Dutch Language & Culture specialist was quick to notice Dutch community members using bullying phrases like “corona child”, “corona loser”, and “corona face”.
Pay special attention to permanent UGC like usernames. You may allow users to discuss COVID-19 in chat, but do you want them to create a display name like CovidVectorGuy2020? Probably not.
Now that you know how users are chatting, it’s time to adjust your chat filter to account for these new words and phrases.
Before you make any changes, consider:
- How often was an expression used? One time in 1 million lines of chat? 20 times?
- If you adjust a rule, what’s the impact?
- Have you inadvertently created chat rules that are too strict? For example, “corona” is a kind of beer, and also refers to the circle of light around the sun or moon.
This is where using a sophisticated chat filter that recognizes context is critical.
Now that you’ve adjusted your filter, monitor your changes to ensure that you’ve avoided creating false positives and false negatives.
For example, you don’t want a phrase like “Corona means crown in Spanish” to trigger an action, whereas you would likely want “I hope you get corona” to result in moderation action (or a false send; whatever works for your community).
Tools that give you a live view of community chat can be very helpful here.
5. Analyze stats and trends
In the Two Hat content moderation platform, clients can run reports to view all chat within a specific time period, or to identify trends and common words.
Whatever reports are at your disposal, we recommend that you compile a regular report of trends and word count for all relevant stakeholders.
- How is sentiment trending? Positive or negative?
- After you’ve identified a new trending word or phrase, how often is it used? Is there an upward or downward trend?
- How many warning messages, mutes, or suspensions did you have to issue daily, weekly, and/or monthly to users who are using the topic to harass others, target someone due to their nationality, or spread misinformation?
6. Review regularly
New trends will arise. The term “social distancing” is common now, but it wasn’t two months ago.
Just today, “China’s Chernobyl” began trending on Twitter. By the time this blog is published, there will be a new trending term that you should be aware of.
At times like this, staying on top of chat trends is critical. With increased volumes as more people are in lockdown and spend more time online, it’s important to safeguard all users and ensure a positive and productive experience in your platform.
To that end, I’m currently offering free community consultations. We can use the time to discuss your content moderation approaches and policies and see if there are any opportunities to update and adapt it in this dynamic online landscape.
Request a consultation using the form below.