Language not available Language not available Language not available

This post is not available in your language. Here are some other options:

Visual Mapping of Twitch and Our Communities, ’Cause Science!

Feb 4 2015

Twitch has grown so quickly this year that it’s hard to keep track of all the amazing subgroups and communities that call Twitch home. To illustrate this, our Science team has recently been building visual maps of the Twitch world and we’re thrilled to share them with you!

NOTE: Visualization and layout were completed using an open source tool called Gephi.

In this map, each circle is a specific channel on Twitch. The lines between channels represent the amount of overlap between the audiences of those channels; each time a specific viewer watched two different channels during the time period this data draws from, it makes the connection between those channels a little stronger. Because we’re only drawing on a short time period in December, not all channels are represented here, and sizing only approximates activity in that time period.

The structure of this network can give us some insights into how the world of Twitch Communities is organized; we see that our most popular games (League of Legends, Dota 2, CS:GO, WoW, Hearthstone, etc) all have their own dense communities of channels. The large red portion of the map contains channels that play a variety of games or games with communities that have a lot of overlap with other games.

Since this data comes from December when there weren’t many big League of Legends tournaments, the League cluster is dominated by current and former pro players and other charismatic and entertaining players. There are also two very distinct offshoot clusters less tightly connected to the main cluster. These clusters represent League broadcasters in Asia (specifically Taiwan and Korea), as well as a cluster of the largest League broadcasters in Central and South America. It makes sense that language-specific sub-communities have a lot of overlap, since viewers tend to prefer channels in their native language, plus they are more likely to be broadcasting in convenient times for people living in that timezone.

Unlike League of Legends, the Dota 2 community is more tightly connected, far less diverse (no large sub communities), and during the time period surveyed was less connected to our larger Twitch community. The lack of connectivity to other major gaming clusters is illustrative perhaps of some isolation of Dota 2 in the gaming community, or perhaps the singular focus of Dota fans on Dota above all other games.

Three of the most connected gaming communities represented on this map are Hearthstone , World of Warcraft , and Starcraft 2. These are all Blizzard games that benefit from extensive cross-promotion, plus Hearthstone and Warcraft share a common fiction.

We also see strong overlap between some of the top broadcasters in each group, and anecdotally we have found that many of the channels in the intervening region have broadcast other Blizzard games at some point in their content. The ‘sc’ in Trumpsc’s name, for example, is a remnant of his early Starcraft broadcasting.

The center of our map contains what we could accurately describe as the “mainstream,” channels that are viewed frequently and in large overlap with other popular channels. Most of our largest broadcasters are found in this region, specifically the largest of those streamers who were not classified into a particular top game. Perhaps the heavy promotion of channels with lots of viewers encourages popular channels to have strong overlap with other popular channels; when a channel has lots of viewers people have a tendency to check it out and see what the big deal is. Dota is the one big exception here. There are many large Dota channels, but they have less overlap with other major channels and don’t appear in the core of the map.

The remaining segment of our map represents the broadcasters that were unclassified into a specific community based upon one of our popular games. Even within this spread out cluster, we still see individual micro-communities that have developed and are both internally connected and well-connected to other nearby communities. This is also the section where we would expect to find communities not collected around games, but rather by content type and expression, including speedrunning, channels playing new releases, role playing, and other similar groups.

Still with us? Want to help us do work like this every day? We’re hiring researchers and data scientists to join our Science team! Check out our job posting and learn more about our team. Shoot us an email if you’d like to learn more about the role or apply!

In other news