08 Oct

Hashtags day before election

#azvote13 mid-September – October 8

Interesting changes here. Group 1, the pro-government forces, are a really tight cluster now. They used to be much less close. Seems like they’re talking to each other more now.
Group 3 is a VERY tight cluster of oppositionists. Strong ties exist between Group 3 and Group 4, centered around muntezir. Group 5 is foreigners and Azerbaijanis that hang out with foreigners.

#azvote13 October 2 – October 8

#azvote13 October 7 – October 8

#secki2013 October 2 – October 8

This is much less contentious than #azvote13.

#secki2013 October 7 – October 8

06 Oct

Election week hashtag analyses

I’ll be continuing to monitor the hashtags, but here is where we stand on Monday morning before the Wednesday election:

#secki2013 for the weekend – pretty dead, but probably will pick up in the coming days
full analysis

#secki2013 for the past 10 days
full analysis

Opposition blogger muntezir is the most powerful tweeter in the network, but groups 3 and 4 are pro-government forces and are really loud on this hashtag.

#azvote13 for the weekend
full analysis

In this hashtag, the pro-government forces (group 2) are “winning” – being louder than the opposition forces. Although muntezir in group 4 and group 3 being a lot of oppositionists and their foreign friends are still holding strong.

#azvote13 since mid-September
full analysis

This tells a great story – while the pro-government forces (group 1) are very loud on this hashtag, all the other big clusters are very well connected and spreading information widely.

06 Aug

Ten thousand signatures felt like ten thousand hands, they carry me

An online petition has been recently been circulating through Azerbaijani Internet circles asking people to support freeing youth activists who have been arrested in 2013, in light of the October 2013 presidential election.

As of early August, 2000 people have signed. (The hashtag suggests they need 10,000 signatures.)

I’m pretty meh about online petitions, although I have recently changed my tuned, as they do provide evidence for “citizen support” in Azerbaijan where public opinion is difficult to ascertain.

Regardless, I’ve done a hashtag analysis of the efforts on Twitter.

Only July 31 when the campaign began, this was the hashtag map.


Popular Azerbaijani Internet personalities Habib Muntezir and Bakhtiyar Hajiyev each have their own clusters of followers and retweeters. The youth organization N!DA (from whom many of the detained activists come) and its leadership make up the largest cluster though.

I did another analysis on August 5 and things look a little different.


Now those key users (Muntezir, Hajiyev, and the official Twitter account of N!DA) from the first analysis are all in cluster 1 together. Cluster 2 includes the leadership of N!DA, separate from the official N!DA account. My guess is that those individuals associated with N!DA are communicating with each other, perhaps about other things, more frequently than they are with the rest of the network, so they have become clustered together.

I don’t entirely understand clusters 3 and 4. Cluster 3 includes some foreigners who tweet about Azerbaijan (including myself). Suggestions on this are welcome.

In my estimation, a lot more of the discussion of this issues takes place on Facebook rather than Twitter. However, analysis of Facebook hashtags is slow in coming.

03 May

#evnvote13 before the election weekend

Hey, let’s check in on #evnvote13, shall we?

This is the Yerevan mayor election. On Sunday May 5 people will vote.

I’m trying out a tool called TAGS explorer. What I like about it is that it automatically archives Tweets and the visualization is constantly updated. Also some of the different metrics are a bit more accessible than NodeXL, in terms of a public audience’s use.

Check out TAGS here

The hashtag is getting a lot busier, huh?

But of course I also am doing #nodexl analyses, which are much more powerful and give clusters.

Here’s the #evnvote13 hashtag analysis on Friday May 3 at 8:30pm Yerevan time.


And in the cluster analysis, it is sort of cool that (unlike normally) all the Hayastantsis and foreigners are all mixed up. It seems like group 3 is more likely to Tweet in Armenian.

We’ll see how things change on Sunday. 🙂

24 Apr

Why is it impossible to hijack #armeniangenocide on Twitter? (Hye jack? ;))

Today is April 24th, which is the Armenian Genocide day of commemoration. All around the world Armenians hold events to raise awareness and remember their ancestors.

In the age of social media, April 24th has also moved online. For the last few years, Twitter hashtags of #armeniangenocide and #recognizethearmeniangenocide and #recognizearmeniangenocide have been common. This year I’ve also seen a growth in image memes or inspirational images posted to Facebook (not surprising, as that trend has been occurring on Facebook more widely.

In the past few months, we’ve seen a number of hashtag hijacks. Here’s my report on a hijacking of #armvote13, #protestbaku, and an attack on an individual.

In all of these cases, a group of individuals “take over” a hashtag by posting messages unrelated to the “spirit” of the hashtag. For example, #armvote13 was about reports of election violations, election results, etc, and people were using it to write things against Armenia and Armenians.

The result is that the hashtag becomes useless because the useful information is hard to find.

So, back to April 24.

Efforts were made to hijack the #armeniangenocide hashtag but this time the attempts were quite unsuccessful for a few reasons:

1. Unlike #protestbaku or #armvote13 – hashtags that were first being used by a small(ish) network of people to share information and second to get information out, #armeniangenocide is entirely an awareness-building hashtag with no information sharing or coordination purpose.

2. To use #protestbaku or #armvote13 you actually had to KNOW that the hashtag existed – someone had to tell you, or people you know on Twitter were using it. (And in both cases, the hashtag was decided upon by small groups of well known users and announced before the event.) #armeniangenocide is a no-brainer! This is the term that people use – it isn’t something that someone had to invent. So any person, veteran Twitter user or not, can get on Twitter, write a message “Hey, don’t forget the #armeniangenocide” and never use the hashtag again.

3. Similarly, it isn’t useful to follow the hashtag #armeniangenocide — it isn’t like #armvote13 where someone may be sitting at home waiting to hear news, or a hashtag like #ica13 (that I use for an academic conference) where people want information on hotel deals or something. There is no purpose in following #armeniangenocide as a hashtag.

4. Star Power! Armenians happen to have a TON of celebrities amongst their people. The most (in)famous in the 21st century is Kim Kardashian (and her siblings and their various significant others and their quarter Armenian spawn.) Kim has 17 MILLION Twitter followers. As my colleague Sarah Kendzior eloquently wrote last fall, when the Kardashians tweet about something, a LOT of people see it. (Sister Khloe and Kourtney have 8 million followers each, brother Rob has 4 million.)


This morning when Kim tweeted about the Armenian Genocide, she had over 1000 retweets every minute.

Given all of this, trying to hijack the #armeniangenocide hashtag is a waste of time.

I did a few nodexl hashtag analyses today, but the hashtag was so busy, there was no way to really track it.

#armeniangenocide 10am ET April 24
#RecognizeArmenianGenocide 11am ET April 24
#RecognizeArmenianGenocide 11:20am ET April 24
#armeniangenocide 2:30pm ET April 24
#recognizearmeniangenocide 2:30pm ET April 24

There is one other reason that the efforts to hijack the #armeniangenocide hashtag were futile.

Some of the Twitter accounts that tried to take over the hashtag were either spam/fake accounts or the real people that were trying to take over the hashtag had, amongst their followers, many fake accounts. As a result, it LOOKS like the person is more powerful on Twitter than they really are because they can make more “impressions” (more people will “see” what they wrote.)

Moreover, one can see using particular tools when there is an unusual growth in followers. For example, the below image.


If one were to try to acquire a bunch of Twitter followers in order to look more important or powerful or influential, one should try to be less obvious about it.

Normally Twitter users have written at least one tweet and follow some people. Furthermore, no native English speaker would write their location as CANADA, Regina or USA, Connecticut. Mine is Seattle, WA. At a quick glance of my friends, almost every American and Canadian writes their location in the way that we do on letters: Ann Arbor (city), MI (state abbreviation).


If you’d like to look at the Excel file of this particular Twitter user, contact me @katypearce on Twitter and I’ll send it to you. Even though this is all publicly available data, in honor of privacy, I won’t post it here.

10 Apr

#barevolution hashtag analysis April 9 10pm and April 10 1am and April 10 10pm

#barevolution is on!

Here’s a hashtag analysis for the last week (this includes following relationships).


Full report

3 main groups here – foreigners, diaspora, and Hayastantsi.

Here’s the analysis for just replies and mentions.


Full report

The groups here are more interesting. There is a group of people all talking to/with @onewmphoto, another with @goldentent and @unzippedblog, another with @mkdotam, and another with @writepudding.

More on this soon!

Here’s 1am April 10

with follows


full report

just replies/mentions


full report

Here’s 10pm April 10

with follows


full report

just replies/mentions


full report

11 Mar

#protestbaku analysis – the day after

ProtestBaku was the “official” hashtag of the protests in Baku over the past few months. It has been used by both the opposition and pro-government groups.

I’ve done a number of hashtag analyses on it.

Throughout the last few months, the pro-government tweeple have both hijacked the ProtestBaku hashtag and started a few of their own. One such example is #molotovlugenclik (Molotov Cocktail Youth), which I wrote about here.

I’m not a fan of the “impressions” or “top tweet” metrics, but I’d like to look at the networks themselves.

There are two ways to conduct Twitter analysis – one is looking at followers (who follows whom) and the other is to look at replies (and just use of the hashtags).

I’m going to focus this analysis on use of the hashtag.

With protestbaku, there were 471 Twitter accounts that used the phrase, and 529 connections (replies and retweets).

The density (0-1 how interconnected the twitterers are), is 0.001 — so, while it might seem “bad” that the interconnectedness is low, considering that the point of this hashtag is to spread information, low density is a good thing.

The average distance between any two tweeple in the map is 4.09. So you can think about this in this way: for information to go from 1 person to another, it would have to go from X…X…X…X – two people in between to get there.

full report



So, on this second analysis it takes into consideration people that follow each other. This tells you less about the spread of information and more about the relationships between people on Twitter.

However, there were 468 tweeple, with 9925 connections between them. The average distance was 2.25, so very close. Density was .05 – so tighter than the model without followers, but still information flows widely.

The most connected users are:

So there are 3 main groups.

Group 1 is foreigners and anti-government.

Group 2 includes mostly pro-government Twitter users.

Group 3 focuses on diasporan Azerbaijani tweeple like muntezir and JamalAli.

full report



#protestbaku did a very good job of spreading information widely, especially compared to other similar hashtags both in Azerbaijan and in different protests globally.

Despite hijacking efforts, the hashtag was “controlled” by those who use it properly.

10 Mar

Hashtag analysis #protestbaku #10mart – March 10

Lots of action in Azerbaijan right now – I can barely begin to describe.

But here’s some hashtag analysis.

#protestbaku from 10 March 3am
full report


I don’t have much to say about this right now, but more will come tomorrow.

#10mart full report

10 March 3pm full report

10 March 4pm full report

10 March 5pm full report

10 March 7pm full report

10 March 9pm full report

08 Mar

#protestbaku got interesting again

I’ve been collecting the tweets for #protestbaku but things haven’t been interesting in awhile. But let’s look at this analysis from March 8 at 5am.


full report

I don’t really get these clusters.

Groups 1 and 2 tweet in Azerbaijani mostly. And somehow I am in group 2. I am not sure what the differences between groups 1 and 2 are.

Group 3 is a mix of foreigners and locals, an English and Azerbaijani.

But this weekend is a new protest, so the hashtag should liven up. Maybe this is a good pre-protest sense of what is happening.