#GamerGate in (Data) Perspective Part II: GamerGate Data Versus Gaming Data
First, thanks to everyone for the feedback about last (last) week’s post. I’d say the responses covered the expected spectrum given the topic (supportive to informed criticism to defensive criticism to name-calling). Many people offered quite helpful comments; I’m very thankful for that feedback.
This week I’m going to look at two areas of #GamerGate. First, I’m going to compare #GamerGate’s Twitter posting sources to actual gaming hashtags. This comparison will expand upon my previous observations about #GamerGate’s heavy use of Twitter’s web client. Next, I’m going to plot post distribution by user for the period Nov. 1 — Nov. 7.
Quick Limitations Note
- I’m only looking at Twitter data (No Facebook, Reddit, or other site data yet, but I haven't forgotten about them).
- Except where noted, the Twitter data comes from November 1st-7th.
- My sample size from the week includes 84.9% of tweets from the period due to collection limitations (This percentage is derived from my totals versus Topsy’s totals).
- To be explicit and precise, this evaluation is only meant to offer a snapshot of #GamerGate on Twitter for the windows described below.
- At this stage I’m still looking only at the conversation around the hashtag and not separating different motivations of use.
Gaming Hashtags Versus #GamerGate
One of the more intriguing comments from last week challenged the hahstags I evaluated as poor choices because they were not gaming related. I stand by the previous selection as a good barometer of common hashtags, but if #GamerGate is abnormal due to demographic issues around “gamer” culture, then looking at other gaming hashtags, especially PC gaming culture, might help confirm the web client preference as a gamer trait.
To this end, I looked at several gaming hashtags from 11/9 including #warcraft (8,556), #starcraft (5,868), #callofduty (3,975), #callofdutyadvancedwarfare (1,535), and #minecraft (79,035). If the web client preference is a factor of the gaming or PC gaming influence, then these hashtags should show some indication of that.
I chose to run a Sunday worth of data like my last comparison, though these hashtags (except for Minecraft) are all far less popular than any of the previous hashtags.
Keep in mind that I have no firm conclusions about why the web client is so common, but I did offer three untested ideas last week: demographic, automation, and professionalization. This week offers a good test of the most likely demographic scenario—that PC gaming habits influence the use of the web client.
What I am testing here is:
Do other common gaming hashtags share #GamerGate’s web client distribution?
So, what do the hashtag sources look like? The following charts show the breakdowns for November 9th.
Of all the hashtags, the web client is most popular for Starcraft (27%) and Warcraft (19%), which are indeed clear PC games. However, in neither case does the hashtag approach the 50% usage of GamerGate for November 9th.
This 60% use is inline with what I observed previously for #GamerGate. It is also comparable to what was observed in the week long data for November 1st through the 7th.
This data fails to confirm a gamer preference for Twitter’s web client that matches the extent of #GamerGate’s preference for the web client.
I return to automation and professionalization as the leading hypotheses. However, whether professionalization or automation is the biggest contributor remains a contested and important question.
The key finding here is that gaming hashtags do not share the same web client preference as #GamerGate.
User Posts Distribution
While I’m hesitant to describe total #GamerGate participation during the week, even with a vast majority of tweets captured, I do feel confident in describing the spread of tweet intensity around the hashtag. Again, this is the entire conversation (for and against) posting with #GamerGate.
I looked at total distribution of tweets by user and number of users represented in each quartile of total tweets for the week. I’ll look at both tweets and RTs in this data set.
For the week, I captured 236,604 tweets and RTs from 22,421 accounts. That includes 78,048 tweets from 11,913 accounts and 158,556 RTs from 15,757 accounts. For the sake of thoroughness, 5,249 accounts had at least one tweet and one RT.
As that last total might suggest, both tweets and RTs worked on a sharp power curve.
In fact the curve is sharp enough that this chart doesn’t fully capture the true rate of decline. What would help is knowing how many accounts are in each quartile of total tweets for the week.
- 103 accounts created the first 19,512 tweets
- 402 accounts created the next 19,512 tweets
- 1,295 accounts created the third set of 19, 512 tweets
- 10,113 tweets account for the final 19,512 tweets
All in all, 2,138 posted an average of once a day. With 11,913 accounts tweeting, this means 18.79% were daily posters.
The curve for RTs is similar.
Again, we see the cluster of power users followed by a steep decline to a very long tail. So let’s look at the quartiles again.
- 104 accounts created the first 39,639 RTs
- 316 accounts created the second 39,639 RTs
- 974 accounts created the third 39,639 RTs
- 14,363 accounts created the final 39,639 RTs
In the case of RTs, 2,962 accounts averaged a RT a day. With 15,757 accounts this means that 18.80% of accounts were daily RTers.
A Passionate, Limited Core
In both the case of tweets and RTs about 500 accounts create half of the total volume in the conversation. Regular daily participation floats around 3,000 users. Then there’s a large body of several thousand accounts dipping a toe in the conversation.
This suggests that however organized or unorganized the movement, the conversation around #GamerGate on Twitter has a central core limited to a few hundred highly active accounts. The total mass of the conversation is in the tens of thousands, though over 80% of those members are involved on less than a daily basis.
Less than 1% of Hardcore PC Gamers Involved
These numbers represent a passionate core, but challenge the idea that “gamers” are fully represented in this conversation by either side. Blizzard’s World of Warcraft has 7.4m subscribers. Even with all 22,421 conversation members accounted for that week, #GamerGate involved one-third of one percent of PC gamers just playing World of Warcraft.
This means as many as 99.6% of gamers are completely unrepresented in the conversation around #GamerGate. It’s a sobering stat for everyone to keep in mind.
UPDATE (11/17): As explained at the start of the article, the conversation I refer to in the second half of this article is the Twitter conversation that uses the #GamerGate hashtag from 11/1–11/7. If you skip to the end, now it’s stated clearly here as well.
