UK United Kingdom

Did Twitter censor Occupy Wall Street?

In the age of social media, if we’re able to get online, engage with a few tools and connect our ideas to others who sympathise, we’re able to initiate social change. Or that’s the theory. The world begins…

Have the #occupy protestors been gagged? AFP/Timothy A. Clary

In the age of social media, if we’re able to get online, engage with a few tools and connect our ideas to others who sympathise, we’re able to initiate social change. Or that’s the theory.

The world begins at our keyboards, and the possibilities are endless: social movements, protests, collective action, revolutions. I mean, people are even naming their children “Facebook” and “Like”.

It’s almost glamorous, really. Not to mention empowering. We can create spreadable hashtags for our #ideas and share them on posters, through speech and online.

Emmanuel Dunand/AFP

But in reality, things might be a bit different. Our trust of, and romance with, social media has been shaken by the #occupy protests that started in New York with Occupy Wall Street. Organisers say the protests have spread to more than 1,400 cities across the globe.

It’s almost as though social media might have cheated on us. But we can’t prove it – we can only dig for evidence where we think it might exist.

Occupy Wall Street, which started on September 17 and continues to gain momentum, is a peaceful display of public frustration around wealth inequality, corporate greed, and the mishandling of the economy in the United States as it emerges from a deep recession. The protesters refer to themselves as the “99%”.

Through media such as Livestream, Twitter and Facebook, as well as in alternative media outlets such as Adbusters, #occupywallstreet and its tens of thousands of participants have only recently broken into the mainstream media.

Even user-uploaded videos of police conflict and mass arrests have had arguably little impact on media attention in the US for such a large-scale, homegrown movement.

In this glorified age of socially-empowered media change, movements such as Occupy Wall Street must still be legitimised by mass media in order to gain a sympathetic audience.

But how do social media help movements such as Occupy Wall Street with their most pivotal need – to be brought to potentially sympathetic audiences?

The case of the missing hashtag

Hashtags are a user-created phenomenon, a system for ad hoc organisation. They help “smart mobs” assemble, and can be easily modified: #occupyeverywhere, #occupyfacebook, #occupySF, even #occupyramsaystreet has popped up.

Hashtags such as #occupywallstreet are not infinitely dispensable, though. And they might not be as “democratic” as people are inclined to believe.

Adrian Kinloch

As the Brooklyn Bridge incident, which saw 700 protesters arrested, occurred on October 1, many Twitter users were outraged that #occupywallstreet, the official, and by far the most frequently used hashtag of the movement, never showed in the official Twitter “trending” list – an aggregation of the most popular hash tags used on Twitter. Some demanded an explanation.

It led many to ask for Twitter to be more transparent, but that’s not part of the service the company offers.

“Though I think Twitter should be transparent with its statistics, we don’t need it to be, as Topsy, Trendsmap, and Trendistic can count for us,” Jeff Jarvis, a prominent American media scholar and technology journalist, has argued in response to demands for more information.

Taking this lead, many Twitter users circulated an image and links to a blog that provided “evidence” Twitter was censoring the #occupywallstreet tags in North America using data sources outside of Twitter.

But why did the #occupywallstreet tag trend on these outside sources but not on Twitter itself?

Although #occupywallstreet participants probably divided their efforts between various iterations of different hashtags, some probably misspelling words in the process, this alone cannot explain the lack of “trending” attention.

“It’s most likely because they don’t have access to our trends algorithm and may not have access to the full firehose of tweets,” Twitter’s vice president of communications, Sean Garnett said by way of explanation.

This may seem reasonable, but further investigation provides conflicting evidence.

Rob Sheridan

To really understand what happened, we need to look at the new rules of critical mass media:

1) Users provide Twitter, Facebook and Google with data: what we are thinking, saying, and doing: what we’ve done; what we plan to do; what we’re thinking about doing. Then, through a complex process, they tell “us” what they think we’re doing. It’s a conversation but in the end Twitter, Facebook, and Google still have the final word.

2) Twitter, as with other online social media, both disperses and attracts audience attention. It is collective yet ephemeral, fluctuating as people tune in and log out.

Regardless of what people think, we don’t know which Twitter topics will trend, why some do not, and how this is really determined. Just like Google’s search, it a complex, protected formula – a secret sauce.

Tracing the trends

So what can we do? We can use those “outside sources” mentioned above, where statistics are available.

Oddly, there’s really only one real statistic-based trending analytic tool around: Trendistic

Using Trendistic, I plotted #occupywallstreet on October 1 (at its peak activity to date) against two of the “top” daily trending topics – #whatyouknowaboutme and #october.

These statistics are over the exact same time frame, when #occupywallstreet appeared to go viral on Twitter, directly before and during the NYPD Brooklyn Bridge protest.


The #Occupywallstreet tag never appeared on Twitter but, according to this graph, it conflicted with two topics that did.

This is by no means the final say or a complete statistical analysis. But it’s interesting that the data above contradict the two primary arguments about how topics trend on Twitter, according to the site itself:

Mat McDermott

1) “It’s already been around”. #October “trended” for most of the day and wasn’t novel by any means. October had a both a higher total percentage and longer baseline and existing time period than #occupywallstreet.

2) “It needs to be very concentrated”. #Occupywallstreet’s percentage differential (or baseline to peak percent of total tweets) was steeper than #whatyoushouldknowaboutme, which “trended” for nearly all of the day.

Unless these data are wrong (taken directly from Trendistic, with the same percentages appearing on other Twitter statistic sites), the stats speak for themselves.

It’s not a conspiracy, but the fact is, as Twitter’s spokespeople stated themselves, algorithms and full access to user’s activity data are not public. Nor do they want it to be. Twitter is a business that must protect its intellectual property.

Revolutionaries, it’s time to leave Twitter behind

What lesson can take from this? In the online world, it’s easy to get a movement started, but it’s harder to interpret what we’ve started or what we’ve accomplished.

It’s not just Twitter, it’s all “transparent” social media.

This means future movements that might be controversial or “revolutionary” are likely to be better off building their own organisation and social media applications.

This is exactly what Occupy Wall Street has done by launching a free Android app as well as Vibe, an anonymous, open version of Twitter.

In the end, if we want to effect social change, we can’t always rely on social media to help us.

Information for this article was taken from publicly available data on Trendistic.

Join the conversation

10 Comments sorted by

  1. Sean Garrett

    logged in via Twitter

    Let me say it again: Twitter did not block (and will not block) any Occupy related topics from trending.

    Correct, Twitter's algorithms are not public. If they were, they would be constantly gamed. Plus, we change them to get provide a better experience for users. We went from a model that was focused on popularity to one focused on what was "breaking" because people engaged with the latter more. It doesn't make it as easy for people to find validation via trends, but it does make most people on Twitter happier with the service.

    Twitter does provide all of its public data to third parties who pay for it. It's called the "firehose". Many more services pay for less. Researchers tap into this, too.

    We will eventually provide the full firehose of data to the US Library of Congress for access by researchers. Perhaps, someday, you can research this matter deeper with this access.

    1. Jonathan Albright

      PhD Candidate at University of Auckland

      In reply to Sean Garrett

      Hi Sean, there is a lot of misinformation about this issue circulating right now, as I'm sure you well know.

      This story is meant to clarify some of this, as well as bring attention to the fallacy when people become outraged (and believe me, many are) when they assume social media, particularly algorithms and search visibility on properties like Twitter and Google, to be completely transparent when they're actually not.

      You're absolutely right: People would quickly find ways to exploit Twitter…

      Read more
    2. Martijn Boersma

      logged in via Twitter

      In reply to Sean Garrett

      Sean, can you elaborate on the difference between popularity and breaking?

      What I gather from your comment, and from the definition of 'breaking' is that a peak in hash-tags needs to be sudden in order to be trending. It sounds like it needs to be as a gradual build up of hash-tag usage doesn't constitute breaking news but popularity. In addition, if a hash-tag is trending for a certain period of time, the definition breaking implies that it will cease to be a trending topic as the over time the hash-tag can be considered as popular and not as breaking any more.

      However, this contrasts with the analysis on Trendistic, as #OccupyWallstreet reaches a sudden peak of 0,57% in a shorter period of time than #whatyoushouldknowaboutme (0.59%) and #October (0,54%).

    3. Jonathan Albright

      PhD Candidate at University of Auckland

      In reply to Martijn Boersma

      Here's data from pro data analysts who wrote their own tracking scripts for Twitter #occupywallstreet: Twitter has 82.5% of all online mentions by network (FB 2.8%).

      "After the start of occupation on 9/17 and up until 9/23, average mentions per day increased by a whopping 2,004%. The following week had a 97% increase over the week prior, and the week after the Brooklyn Bridge arrests saw a 216% increase in average mentions per day"

      Look at the spike in traffic from bottom to peak on: It makes the graphs I used from Trendistic pale in comparison. The jump between Oct 1 - 2 unbelievable.

    4. Martijn Boersma

      logged in via Twitter

      In reply to Jonathan Albright

      I would love to see the logic behind an algorithm that ignores such a significant spike in traffic explained.

  2. Sean Garrett

    logged in via Twitter

    hey all: I recommend checking out this article/discussion via Social Flow:

    Social Flow has access to the full Twitter firehose. Trendistic, for example, has access to our "garden hose" (10% of tweets).

    As for the logic behind the algorithm, we explain that pretty well in the blog post that this this post links to ("To Trend or Not to Trend"). But, to summarize, people who use Twitter get more engaged in Trends that are new and emerging than ones that they have seen before.

    1. Martijn Boersma

      logged in via Twitter

      In reply to Sean Garrett

      Thanks Sean.

      I see my reasoning about what constitutes 'popularity' and 'breaking' reflected in "To Trend or Not to Trend": Twitter favors novelty over popularity, topics break into the trends list when the volume of tweets about that topic at a given moment dramatically increases, and topics don't become trending or cease to be trending because the velocity of conversation isn’t increasing quickly enough, relative to the baseline level of conversation happening on an average day.

      My confusion…

      Read more
  3. David Glance

    Director of Innovation, Faculty of Arts, Director of Centre for Software Practice at University of Western Australia

    My issue with this is the drastic over-simplification of the news and analysis coverage of the Occupy Movement.

    The focus on conspiracy by Twitter (and Google and other) users reflects the eagerness of people to believe that what they are doing or talking about has relevance and seeing this reflected in certain key measures.

    When it isn't, they commonly turn to conspiracy - after all, who could believe that OccupyWallStreet would not be as interesting as the latest Kim Kardashian event? Also…

    Read more
    1. Jonathan Albright

      PhD Candidate at University of Auckland

      In reply to David Glance

      Thanks David. This piece isn't meant to oversimplify or focus on conspiracies, but rather to draw attention to the underlying issue, which is the false transparency of social media.

      There will be research papers and dissertations which detail every facet of the OWS and Occupy movements soon enough.

      You're right, people often assume that Twitter, FB, etc are advocates because they work for their causes in many regards, then becoming outraged after they feel it does not reflects their views and…

      Read more
  4. Tarleton Gillespie

    Associate Professor

    For me, the interesting question is not whether Twitter is censoring its Trends list. The interesting question is, what do we think the Trends list is, what it represents and how it works, that we can presume to hold it accountable when we think it is “wrong?” What are these algorithms, and what do we want them to be?

    I took Lotan's post as a starting point for a similar discussion, that might be interesting to all of you:

    Can an algorithm be wrong? Twitter Trends, the specter of censorship, and our faith in the algorithms around us
    Oct 19, 2011

    originally posted on Culture Digitally:
    reposted on Salon as "Our misplaced faith in Twitter Trends":