Last week we published an analysis on the usage of hashtags around the #OccupyWallStreet movement on Twitter, why some phrases reach Twitter’s trending topics list, while others never do. The crux of the the argument highlighted the outcomes of a purely algorithmic mechanism that Twitter uses to generate its trending topics lists.
We’ve all seen an increasing usage of the #Occupy hashtag, splintering into a wide range of sub-movements -> #OccupyBoston, #OccupyVancouver, #OccupySF and many more. While it is clear that the movement’s rise was influenced by similar demonstrations around the world, it is unclear who first decided to use the #Occupy hashtag, for what reasons and how it spread.
We’ve sifted through millions of tweets to answer precisely these questions. Some findings:
- We identified the first use of #OccupyWallStreet in a blog post published by Adbusters on July 13th.
- Many of the initial supporters are Spanish Twitter accounts.
Al-Jazeera English was the first mainstream media outlet to use the hashtag.(update: deeper look at data shows that @AJEnglish did not actually post the hashtag at that time, but rather users retweeted content that was never posted by @AJEnglish)
- As interest and momentum builds up, more media outlets cover the story.
- Jason Scott’s cat (@Sockington) is immensely influential!
Interesting? read on…
First Identified #OccupyWallStreet Use
Going many months back, there were a few folks using the #occupy hashtag on Twitter. From a UC strike (#OccupyUC) to a critique US forces in Iraq (#OccupyIraq), to outright announcements of sheer boredom (#OccupyMe), people were using #occupy in a variety of ways.
We identified the very first contextually relevant usage of the hashtag #OccupyWallStreet, published the evening of July 13th 2011, in reaction to Adbuster’s blog post titled: #OCCUPYWALLSTREET, a shift in revolutionary tactics:
The Adbuster’s blog post was a call to action, with the stated goal of bringing 20,000 people into lower Manhattan to “setup tents, kitchens, peaceful barricades and occupy Wall Street for a few months.”
We are so used to seeing rapid information spreads on Twitter, yet what we see in this case is an extremely slow buildup. Below is a network graph representing all tweets that included the #OccupyWallStreet hashtag in the first 10 days since the hashtag was first published by Adbusters. The size and color of the nodes represent their importance in spreading the word (specifically, their eigenvector centrality measure). The lighter and larger a node is, the more influential it is in terms of spreading the information.
Some of the very first tweets:
@Adbusters: “Can we get 20,000 people to barricade Wall Street until their demand for real democracy is met? http://bit.ly/re9ENL #occupywallstreet”
@OccupyWallStNYC: “Can we get 100,000 invites on the Facebook #occupywallstreet page before Friday at Noon? http://t.co/yeYdPJ2”
@jnjcasper: “Is #tahrir coming to America? Call on Sept17 to #occupywallstreet. Interesting. Scary. So far, weak, but don’t dismiss. http://t.co/mTb73FZ”
@GuyAitchison: “I wonder if this will catch on – a call to Occupy Wall Street on 17th September http://bit.ly/re9ENL via @adbusters #occupywallstreet”
In the period between July 13th and July 23rd we saw a total of 568 tweets published by 386 different users. Both @Adbusters and @OccupyWallStNYC are (not surprisingly) the most central in discussions and mentions of the planned event. Accounts such as @AmirMoeSoliman and @MsVeronicaJay have such central locations in the flow since their content was retweeted by @OccupyWallStNYC.
The network branch above is a view into the top right region of the graph, representing much of the Spanish contingent that reposted and shared the hashtag at a very early stage. These users were most likely influenced by the events happening across the Iberian peninsula over the summer, tweeting their enthusiasm hoping to see an outcome of their struggle crystalize somehow in the United States. This Spanish contingent helped post and spread the word about #OccupyWallStreet before much of the US have ever heard about it! An example tweet:
A popular, highly visible movement
On October 13th, news came out that Brookfield Properties, owners of Zuccotti park #OccupyWallStreet’s main location, requested that NYPD clear protesters out of the park in order to clean and sanitize it. The movement suspected this was an attempt to crush the demonstration, pleading its supporters to physically come to the square early in the morning, before the NYPD deadline. Information about the turn of events spread like wildfire on Twitter, users asking others to physically come to the square and support the movement. The following graph represents over 6,000 tweets posted by close to 5,000 twitter users, tracking and spreading verified and unverified information about the impeding standoff between the protesters and NYPD.
The graph is organized by centrality – the larger and lighter a node, the more important it is in affecting others to respond, effectively getting information to spread in the network. It is easily noticeable that @OccupyWallStNYC is the most important information sources, while in its “neighborhood” one can notice sibling accounts, such as @OccupyWallSt, @OccupyLA, @OpWallStreet, @The99Percenters, @WeOccupyAmerica, and so on.
But unlike the early information flow at the beginning of this post, this one is filled with media outlets participating in and helping spread information about the turn of events. Some of the major media outlets to participate are @HuffingtonPost, @AP, @NewYorkPost, @CNNbrk, along with news aggregators such as @BreakingNews.
The cleanup was called off an hour before the requested deadline, sending messages far and wide across the network. @jncatron cited the following message that came in through the AP: “NYC official: Protest cleanup is being postponed: By KAREN MATTHEWS AP NEW YORK”. Below is a section of the visualization highlighting all the people who reposted that message.
Additionally, Jason Scott’s cat (@sockington) played quite an influential role in spreading word about the postponed cleanup (visualized below):
@sockington: “CONTINUING MY #OCCUPYLITTERBOX WELL INTO THE MORNING glad to hear cleanup was postponed”
Questions, comments, thoughts? Ping me on Twitter – @gilgul
Oh – and links to hi-res, svg versions of the network graphs can be found below: