Update | Click here for an update to this post. The more I dig, the more I think something weird is going on here.
A week ago I wrote a piece on why the hashtag #Wikileaks wasn’t appearing on Twitter’s trending topics lists, concluding that Wikileaks’ failure to trend was an artifact of an actually-quite-reasonable-really algorithm, and shouldn’t be taken as evidence of anything nefarious.
Okay. That was then. But in the intervening week, Wikileaks has seen a truly staggering amount of traffic on the site, and still hasn’t trended once.
How staggering? This staggering:
Back in the summer, the title of the movie Inception peaked at 0.4% of Twitter traffic, soon stabilizing at about half that. It trended for a month and a half. Wikileaks has broken 2.0% of total traffic twice in the last week, and hasn’t yet dipped below Inception’s all-time high.
Want more? Here’s more. Wikileaks had more traffic than “world aids,” the week’s longest-trending phrase, for all but three hours of the week. It out-trafficked Bieber and Obama by wide margins. It was more popular than Jesus by an order of magnitude. At its peak, it appeared more frequently on Twitter than the word “or.”
And here’s my favorite … on five separate occasions in the last week, including eight hours on Friday and two hours on Saturday, “wikileaks” was getting more traffic than “twitter.” On Twitter.
Okay. So does all this prove that Twitter is blocking Wikileaks from trending?
No.
Not exactly. Not quite.
For one thing, as I posted last week, Twitter’s trending topics don’t — and shouldn’t — just reflect raw traffic. As the company explained back in May, when they overhauled their algorithm, it’s designed to catch “topics that are immediately popular, rather than topics that have been popular for a while or on a daily basis.” Since Wikileaks has trended before, it’s now got a higher bar for trending again. (This explains why the alternate Wikileaks hashtag #cablegate trended for a total of 22 hours last week, despite having consistently lower traffic than Wikileaks itself. Because it was a new tag, it got the algorithm’s attention.)
For another thing, not all tweets are equal in Twitter’s eyes. As the site’s trending topics FAQ notes, they take a dim view of folks who try to game the TT list by flooding the service. “The most important thing,” they say, “is to make sure your Tweets are genuine thoughts or impressions and not attempts to insert yourself into a trend.”
This is likely a partial explanation for the incredible trending run that Inception had this summer. Inception wasn’t just a popular movie, it was a controversial and confusing one — one that got people discussing, not just tweeting. I haven’t gone back and checked, but I have a very strong suspicion that an unusually high proportion of Inception’s tweets during its reign were brand-new content — much higher than that of recent Wikileaks tweets.
It’s also possible, given Twitter’s warnings on its FAQ page against “repeatedly tweeting the same topic/hashtag without adding value to the conversation in an attempt to get the topic trending/trending higher,” that efforts to get Wikileaks to trend — or even discussions of why Wikileaks wasn’t trending — could have pushed the topic down on the TT lists. (Which means, for instance, that the folks who are tweeting about Wikileaks’ failure to trend under the #thingsimiss hashtag are likely hurting Wikileaks’ chances of trending.)
Finally, I have to at least acknowledge Twitter’s denial that they muck with the trending topics list for political purposes. A few days ago, a similar controversy arose in relation to the failure of #demo2010, the hashtag of Britain’s student protesters, to trend in that country. Asked for comment, a Twitter spokesperson told The Guardian that there was “absolutely no truth” to the charges. “We have not,” he said, “and will not, do anything to stand in the way of people using Twitter for the open exchange of information.”
So does that settle it? A week ago I thought so. But I have to admit that the sheer scale of the Wikileaks traffic since then gives me pause.
I still think the vast majority of TT conspiracy theories are bunk. Twitter wasn’t censoring #demo2010, I’m certain. There’s no evidence that it censored #cablegate. It’s not censoring #imwikileaks, as some have charged — that hashtag just hasn’t taken off in any serious way.
I’m also unconvinced that Wikileaks trending on Twitter would be all that big of a deal. Huge numbers of people are participating in Wikileaks discussions on the service — if the goal was to stifle conversation, it failed. Trending on Twitter is an easy measure of a subject’s influence, and in a weird way a sort of badge of honor, but it’s not obvious to me that it has a huge effect on a story like this.
To be honest, this is another reason that I tend to doubt that Wikileaks was censored — to do so would pose a huge risk to Twitter’s credibility with its audience, and for very little reward.
But if Wikileaks’ failure to trend is an artifact of Twitter’s algorithm — and that’s still my default guess — it’s a really screwed up algorithm. Because Wikileaks has been the biggest story by far on Twitter this week, by any measure. It’s a huge story. It’s an important story. It’s a breaking story. It’s an evolving story.
It’s the very definition of a trending topic.
So why isn’t it trending?
Update | As I noted at the top of this post, I’ve done a visual comparison and anaylsis of traffic for “Wikileaks” and “Sundays.” You can find that post here, and I expect you’ll find it as startling as I do.
17 comments
Comments feed for this article
December 5, 2010 at 1:24 pm
crguna
This happened earlier with other topics like #barkhagate . So this is not specific to wikileaks. We were told then that the trending topic stays on the list upto a week and stops being listed as trending after that.
December 5, 2010 at 1:41 pm
Angus Johnston
It’s not as simple as a one-week cutoff — as I noted in the post, “inception” trended off and on for a month and a half this summer, including a total of 39 days on the TT lists.
Another theory that has been offered, and which I addressed in my previous post, is that Twitter doesn’t allow words that are also usernames to trend. But they do, as a quick check will show.
December 5, 2010 at 1:54 pm
crguna
Could it be that Trending depends upon proportion and number of ‘new users’ posting on the #topic ?If most of the posts are from the same users who have posted before, maybe it is not really ‘trending”
December 5, 2010 at 3:30 pm
Ciaran
This is an absolute disgrace.
First – Paypal cuts donations to Wikileaks.
Who the fuck are Paypal? Who do they think they are? So if someone goes up against the government, they can just pull the funding plug?
Is this democracy?
Then Amazon removes Wikileaks as a paying customer from it’s servers, saying that it doesn’t work with people “engaged in illegal activities.”
And yet it still lists Hillary Clinton’s autobiography after the state department approved the hiding of illegal cluster munitions on British soil? With British complicity? No.
That’s NOT fair. This has nothing to do with illegal activity. This is businesses doing favours for politicians.
Why?
WHY?
Because they are corrupt. Corruption Wikileaks has the courage to expose.
Is this democracy?
IS THIS DEMOCRACY?
No.
No it’s not.
I’ve started a facebook group – I started it yesterday, we’ve already got over 700 members. Please help us publicise it.
It is called the Julian Assange Couchsurf 2010.
This is the link:
http://www.facebook.com/home.php?sk=group_128056727255913&ap=1
It’s anyone who’ll lend a couch to a man on the run, being hounded by the authorities over spurious accusations, hounded for an act of incredible courage and principle.
It’s acting as a stunning resource for some extremely worrying articles – such as one featuring an interview with one of Assange’s alleged victims, in which she says this:
“It is quite wrong that we were afraid of him. He is not violent and I do not feel threatened by him.”
Are these the words of a rape victim? How would this farce play out in a British court?
This is a stunning and sickening assault by the political establishment on a man of incredible courage. It is imperative that people stand together and demonstrate that not all of us are so easily used by a system that cares only about the degree to which we consume.
I’ve also just written a big blog article about this, I’ve drawn in a lot of sources and done some proper research. I’d really appreciate any help anyone will give me, or the facebook group, in getting this message out.
December 5, 2010 at 3:31 pm
Ciaran
That blog article link is:
http://ruthlesstruthdotcom.blogspot.com/2010/12/julian-assange-couchsurf-2010.html
Sorry for the breach of blog etiquette, I feel very strongly about this but I understand completely if you want to take this link down.
December 5, 2010 at 7:12 pm
Top Posts — WordPress.com
[…] New Questions About Why #Wikileaks Isn’t Trending On Twitter Update | Click here for a comparison of traffic between “Wikileaks” and “Sundays,” which trended for most of […] […]
December 6, 2010 at 7:48 am
Twitter is censoring the discussion of #Wikileaks | safety first
[…] added to my own thought with material I encountered on Student Activist. I warmly recommend reading his posts on this issue as well as his interactions with the Twitter Representative. He goes into greater […]
December 6, 2010 at 7:50 am
#Wikileaks auf Twitter? Fehlanzeige! (Update) : netzpolitik.org
[…] Johnston hat das Ganze mal ein bisschen genauer analysiert und zwei Artikel dazu geschrieben. Hier und hier. Ähnliches hat bubbloy bei safetyfirst […]
December 6, 2010 at 8:13 am
Jillian C. York
You may find this helpful; we did an analysis of Twitter trends this past summer when complaints arose that they were censoring “flotilla” as a hashtag:
http://opennet.net/blog/2010/06/iranelection-censored-evaluating-twitters-trending-topics
December 6, 2010 at 9:51 am
Liz
The BP Gulf Oil Spill is not a good example because it’s a trend that actually got MORE popular over time, not less. At What the Trend we saw it move from #30 in our weekly Top 100 each week until it rose to #6 over the span of five weeks. It’s a counterexample of the phenomena you’re talking about
This result is partially because of the timeline of the oil spill: from 20 April 2010 to 15 July 2010. Twitter changed their algorithm (from what was most “popular” to trends that are “newly emerging”) at the end of May.
That’s when everything shifted and became much more ambiguous and arbitrary. It why the Brazilian presidential candidates were in the Trending Topic Top 10 in August but not in September or October (the election was Oct. 3rd) when the political conversation was getting louder. If this system had been in place in 2008, you would have seen Obama & McCain trending in June but not making the trending topics in November 2008.
The way around this is the childish but effective changing of terms and hashtags so there is this novelty factor. The users who do this most effectively are fans of Korean pop bands who, nearly every week, get their trends in the Top 10 because they are continually inventing new hashtags once their older ones no longer make the top trends list.
This is why Wikileaks didn’t make Twitter’s top trends but #cablegate spent 22 hours on the Top 10, most of that time at the #1 spot. In fact, it was such a strong trend that it catapulted the topic of Wikileaks to the #20 spot in our Week in Review (http://www.whatthetrend.com/week_in_review_20101203).
As for why this doesn’t effect the “Sundays” trend, I don’t know. But this novelty factor, as you call it, doesn’t just affect Wikileaks, it affects #NowPlaying, Justin Bieber, Harry Potter, Glee, #FollowFriday, Kobe Bryant, Super Junior, basically any term that has trended exceptionally high for a certain period of time. How many hours/days it takes to have the trending threshold raised (the number of mentioned required to make it into the Top 10), is unknown to those outside of Twitter.
It would be great if Twitter was more transparent but I’m guessing they think if they made their super secret algorithm formula public, that regular users as well as spammers would try to game the situation.
It’s less than an ideal situation as the example of the Brazilian Presidential Election shows. Movies that are newly released into theaters trend opening week and then are never seen again even if they continue to be popular. And some topics are important and newsworthy for more than 24 hours and they don’t have fan clubs to keep inventing new hashtags. Right now, that is the only solution around this problem that I know of.
Liz Pullen
What the Trend
http://www.whatthetrend.com
December 6, 2010 at 10:16 am
Wikileaks Isn’t A Trending Topic on Twitter: Censorship, Miscalculation, or Just Not Worthy? - AllTwitter
[…] the course of two more posts last week (here and here) about the issue, Johnston brings to light some in-depth statistics about Twitter traffic […]
December 6, 2010 at 1:17 pm
Liz
P.S. I’ve never seen a username as a trending topic and I’ve seen tens of thousands of trends. At least not with a @ sign. So, you’d never see @BarackObama trend but you could see Barack, Obama and Barack Obama all trending simultaneously.
Sometimes Twitter lumps trends together and sometimes it separates out phrases. For example, right now Mark Gonzalez, who is a soccer player, is a trending topic. But Twitter is also counting any Tweets that include both “Mark” and “Gonzalez” in the update including Tweets I’ve seen mentioning Red Sox players Adrian Gonzalez and Mark Teixeira.
It’s not a perfect system. It gets very confusing when a trend is a word that has different meanings but the same spelling in several languages. There is a lot of confusion about why the word is trending and sometimes it just seems like some fluke on Twitter’s part.
December 6, 2010 at 2:06 pm
Chris
Angus Johnston, what’s surprising to me is that you are this naive! You claim that there’s nothing it in for Twitter to underreport the Wikileak traffic on its service. Get real man! There was also nothing in it for Amazon and Paypal to screw with Wikileaks. They did it because the CIA, NSA and US government have a lot of leverage over them. Several years ago, CNN backed off from reporting illegal US government activities after the government threatened to cut them off from vital communications satellite services controlled by the government. As Wikileaks clearly proves, this kind of strongarming from the US government is a daily occurrence. It’s ironic that your story is about Wikileaks, but you seem to have learned nothing from the leaks themselves.
December 6, 2010 at 2:08 pm
champak
this buastard is going to start 3rd world war
December 12, 2010 at 9:11 am
vis4.net/blog - über Infografiken und Datenvisualisierungen in Flash, HTML5, SVG und Processing
[…] New Questions About Why #Wikileaks Isn’t Trending On Twitter […]
December 14, 2010 at 8:39 pm
Perché #Wikileaks non può essere Trend Topic « Gilda35
[…] New Questions About Why #Wikileaks Isn’t Trending On Twitter […]
November 24, 2014 at 9:27 am
Warum ist Wikileaks kein Trending Topic bei Twitter? - zurPolitik.com - zurPolitik.com
[…] Kürzlich schrieb ich über die Idee, dass Twitter aktiv daran arbeitet, Wikileaks aus den Trending Topics zu halten. Gestern führte ich das weiter und drückte meine Skepsis aus, dass ein fauler Zauber stattfindet – auch wenn die Daten seltsam aussahen. […]