Twitch Trends

calming down hyperactive Twitch chats

Group chats on popular streams are extremely fast-paced and filled with redundant content. In the screenshot below, we can see three trends where users add little original content, bandwagoning onto the reactions of others. Though the chaos of Twitch chats is part of the appeal for some, we felt that slowing down the fast and furious nature of the chats would provide clarity amidst the hype.

chat spam packed into neat boxes

We created an interactive visualization that replaces the normal chat section of the website, and updates dynamically in real time while the user watches the stream.

separating needles from haystacks

In the top section, top trends are displayed. The length of each bar represents its prominence in relation to the trendiest trend.

In the bottom section, chat messages as they would appear in normal Twitch chat. Usernames are stripped since they aren’t generally useful in chats upwards of 10,000 viewers. Badges indicating status are still included, however, since knowing if a moderator authored a message might be significant.

Messages that are part of the top trends do not appear in the bottom section, which reduces both the speed and spamminess of the chat. This allows users to explore the “long tail” of chat content.

The divider between the top and bottom section can be moved vertically to adjust how many top trends are shown. Messages that are part of trends will be filtered or unfiltered from the bottom section accordingly.

viewing trend content

One may click on a trend in the top section to show only the messages related to that trend in the bottom section. This allows for more nuance on what the trends mean. Trends with opposing messages (e.g. “I like dogs” vs. “I hate dogs”) that our algorithm doesn’t split into separate trends may be clarified by using this feature.

I came up with the concept, did visual design, and coded the user interface.