Is Twitter Censoring the Revolution?

(Originally posted on the Involver blog at http://blog.involver.com/2011/09/29/is-twitter-censoring-the-revolution/)

At this point pretty much everyone who works in Social Media has read at least a couple articles talking about the role played by Twitter and other social media sites in coordinating and publicizing protest movements around the world. Social media has been alternately credited and blamed for everything from the ouster of Mubarak in Egypt to the recent riots in London. Now personally I take these claims with a very large grain of salt, but there’s a lot of truth to them as well. Today, just about everyone with a message to spread is using social media and Involver has customers scattered all over the political spectrum.

One of the interesting things about social media is the neutrality of the mediums themselves. Now of course if you talk to just about any activist you’ll hear that “the media” is biased against them. Conservatives think that the media is too liberal and Progressives think it’s too conservative. But up until now it’s been generally assumed that social media is a relatively even playing field because anyone can broadcast on it and build an audience.

Barack Obama may have 10x as many followers as republican front-runner Mitt Romney (10,252,333 to 108,426 as of this writing) but Romney’s numbers are rising fast – he just hit 10k on 9/23/2011 and is looking to at least double it in the next 60 days. The thing is, you don’t need to be either Obama or Romney to build an audience using social media, you just have to say things that people find interesting and want to hear more about. And because social media content is created and promoted by users it’s not subject to the same accusations of reporter bias that traditional media is – there’s essentially no gatekeeper. So when I saw tweets last week claiming that Twitter was taking on a gatekeeper role and censoring hash tags related to the execution of Troy Davis, I figured it was worth investigating.

The accusation was that Twitter had admitted blocking the hash tags #TroyDavis and #TooMuchDoubt from trending because the tags were “offensive.” As you can imagine, that set off all sorts of warning bells. Digging deeper , however, I found hundreds of tweets repeating the claim of censorship but none with a link to a reputable source. Now that’s not proof that Twitter didn’t censor the tag, but it certainly doesn’t help prove that they did. It’s particularly sticky because while Twitter’s spokespeople are adamant that they don’t censor political news, they’re forthright in admitting that they do censor out “offensive” topics. But offensive to whom? And by what criteria?

Accusations that social media sites like twitter censor results are common, but are they accurate?


A quick Google search for “twitter censorship troy davis” reveals a dozen blogs taking various positions on the subject – some adamant that the #TroyDavis was actively censored and others equally adamant that no such censorship took place. None offered much in the way of evidence but that didn’t stop people from tweeting and repeating those stories and citing them as proof. This week I saw similar claims that Twitter was censoring #OccupyWallSt (the ongoing protest camp in NYC based on the model pioneered in Egypt during the Tehran Square protests against Mubarak). Thinking back, I can recall claims during the last election about Ron Paul being censored. A broader search revealed blog posts claiming that twitter had censored trending topics related to #wikileaks, the aid flotilla to Palestine several months back, and several other topics.

Now Involver provides a development platform that’s used by campaigns, companies, and brands with wildly diverging interests and ideologies. Our role isn’t to advance any particular agenda but to build some of the best social media tools in the marketplace and empower our clients to effectively achieve their goals, whatever those may be. In the aggregate, all of those different ideas and ideologies enrich everyone because they contribute to the marketplace – both the literal marketplace and the marketplace of ideas. So if the companies that provide the core platforms upon which our industry is built take it upon themselves to actively distort that marketplace of ideas it has an obvious impact on our clients. The thing is, after hours of digging I was unable to find proof that any censorship had occurred. Which was both reassuring as someone working in the industry and disappointing because I wanted a juicy story for this blog. Unfortunately for me, facts got in the way.

Part of the issue is that claims of censorship are very difficult to verify or disprove and many people don’t let facts get in their way. In the Guardian story linked above, the #flotilla topic disappearing from the trends list was put down to a faulty algorithm and a bit of coincidence. Apparently, there had been another trending topic about a different #flotilla the week before and twitter automatically downgrades recurring tags. Twitter strongly denied it was practicing censorship, and the Guardian writers judged that their explanation made sense. The Wikileaks post shows a lot of aggregate data comparing third party tools with Twitter’s own reporting and that looked like the strongest potential documentation of censorship, except that trending isn’t just based on how many times a hash tag is retweeted but also by how many people and the rate of change. That ‘Rate of Change’ factor is critical especially for grassroots political causes since their activity tends to build slowly over time. Because the frequency graphs for tags like #occupywallst and #troydavis look like gently rolling hills with long starts and trailoffs instead of sharp changes, they are much less likely to trend – even though their total numbers may end up being much higher then the currently trending topics. That’s by design because it allows Twitter to prevent Justin Bieber and other celebrities from staying at the top of the trending list 24/7.

Now you can argue that the end effect of the algorithm is inherently unfair to grassroots movements, but you can’t argue from that that twitter is actively censoring those movements. It’s a subtle distinction but a crucial one. In real terms, the extreme difficulty of getting a topic to trend based on organic spreading of the topic without media assistance is an (ironic) side effect of our media-obsessed culture. All of which is hopefully useful information for any of you who are working on campaigns that leverage Twitter! For their part, Twitter routinely denies all charges of censorship and, at least until #Wikileaks runs an expose on Twitter’s secret plans for world domination, I’m inclined to take them at their word.

I suppose the point of this story comes down to credibility. Tweeting that you’re being censored because you don’t have the audience that you want does not help one’s brand or cause, in fact all it does is damage your credibility. It’s one thing to say CNN should be covering something that it isn’t, there are all sorts of plausible cases that can be made and argued about an opinion-based judgement like that. It’s another thing entirely to claim that a social media site is manipulating the playing field when you don’t have the hard evidence required to back up such a claim of fact. Incautious tweets are much more likely to damage your personal brand then they are to damage Twitter’s credibility. Worse yet, it makes it increasingly likely that genuine instances of censorship will be ignored as just another user crying wolf. That said, if you have evidence and can make a convincing case please do us all a favor and come forward.

What are your experiences? Have you seen topics pulled that you expected to continue to trend? Do you have raw data that can help prove or disprove these claims? We’d love to hear about it!