This is epic. Protesters in Tahrir . . .
G. R. Boynton

In October of 2011 OccupyOakland set up camp in a park in Oakland. After a couple of weeks the mayor had had it. On October 25 the police, 500 strong, attacked the Occupy camp with tear gas and batons. It was not pretty. One of the casualties was a Marine veteran who had survived two tours in Iraq only to be severely wounded by being hit by a tear gas cannister and then had to be rushed to the hospital.

The police claimed that there was no brutaility in their actions, but the images of brutality were everywhere. On the 26th every news outlet in the country carried the photos or video, and the criticism poured in. The mayor's Facebook page got thousands of critical posts until she turned comments off.

On the 27th they did what they had to do. The police chief resigned -- saving the mayor from having to take responsibility. And the mayor apologized, said she was a supporter of the 99%, and said OccupyOakland could return to the park.

And on the 27th there was a huge outpouring of messages via Twitter. It was just over 44,000 messages. That many lines of text can be daunting. However, there is a standard feature of Twitter messages about politics that reduces the magnitude of the job.

There is a high level of redundancy in Twitter messages about politics. As much as 40% to 70% of the messages may be retweets, where a retweet is an attributed quotation. A person reads a messages and sends it on, including the username of the original writer, to the person's followers. In this collection, for example, the number of retweets was 30,411 of the tweets or 68% of the messages. Retweets vary rather systematically. In situations of stress, as was the case for OccupyOakland, the magnitude is greater. In communication about political candidates, such as Rick Perry in October, 2011, the retweets were 36,973 out of 94,970 or 39%.

This brief note introduces two tools for working with this kind of redundancy.

One step is reducing 30,411 messages to just over 2,000 messages with a count of the number of times each message appeared. The top ten messages for October 27 in Oakland were:

One message was retweeted 1,071 times. Another message about Google assumes less background information from readers.

RT @ZsofiaT: #Google Refuses Law Enforcement Request 2 Remove Videos Of Police Brutality From #YouTube http://t.co/pjUQ1opq #OWS #OccupyOakland #OccupySF

The protestors claimed police brutality. The police claimed there was no brutality, but they also tried to get Google to remove the videos of police action. For OccupyOakland it was a great victory that Google refused.

The second most frequently retweeted was

This is epic. Protesters in Tahrir tomorrow will march to the US embassy to protest the violent crackdown of #occupyoakland protests.

This message was retweeted 816 times. @thefineprintuf said "breaking news." @tokimonster said "Very cool!" @ThatSarahSnyder said "This gives me chills/tears." The exact phrase was retweeted 816 times. But if you search for Tahrir you find many variants on this news. It was a moment of having made it on the world stage. The message about Egyptians appeared 1,400+ times in the 44,000+ messages. (Boynton, 2/8/2012)

And that makes it foreign policy. Foreign policy being exercised by the protestors of one nation in response to protestors of another. People are beginning to take foreign policy into their own hands.

But my favorite Twitter message did not make it into the top ten.

RT @TheAndrewFoley: So LET ME GET THIS STRAIGHT.The #OPD attacked the #OccupyOakland camp & shot an Iraq vet IN THE HEAD to *prevent* violence and injuries...

To reduce 30K messages to 2K messages I used a program written specifically for the task. It is retweets.exe, runs on a Windows computer [with a Mac version soon], and is available to anyone who would like to use it. Forty-four thousand is pretty unmanageable; 2,400+ is not.

So, what to make of all this redundancy? The message stream of October 27 was filled with messages that appeared over and over. In what way might this be important?

Twitter is designed to let individuals choose to follow others. If you follow a person you receive all of the messages that person writes or retweets. You may not read them, but you have the opportunity to read them. President Obama is the star of the political world with 12,818,020 followers; it changes every day so this was as of March 4, 2012. No other political figure has over a million in followers. However, this is a very loosely connected network. Very few people follow the same people or set of people. In a brief note about Sarah Palin I examined the pattern of retweeting this message. (Boynton, 3/16/2010)

Sarah Palin Crossed Border For Canadian Health Care: http://bit.ly/cRRdAU

The messages was retweeted 366 times. If you count the followers of the original writer and the 366 retweeters the number who had access to the message was 366,636 for a thousand to one ratio. One possibility is that many of the 366,636 received the message from more than one person, which would reduce the total number receiving the message. However, the joint membership was only 4.4%. So 95.6% received the message from only one person.

To show how this worked with the messages of October 27 I looked at the retweeting of my favorite message. The message was retweeted 136 times. Two persons who retweeted it had no followers which makes retweeting a bit bizarre. Seven accounts have disappeared. So I computed the followers of the remaining 127 retweeters.

I used Twiangulate.com, a software system online, that will give the followers of three Twitter accounts at a time, the number of shared followers, and other bits of information about the accounts. TheAndrewFoley is followed by 495 persons. The followers of the persons retweeting the original message was 642,784; that is 5,061 followers on average. The account that produces this very large number is ebertchicago. It is the account of the noted film critic and political blogger Robert Ebert. He has 605,640 followers, which is the overwhelming majority of all followers who saw TheAndrewFoley's message. The remaining 126 retweeters had 37,639 followers, which is an average of 299 followers per person. But it was 642,784 who had access to the message, and that is a big number.

When the number of tweeters is large the matrix needed to calculate common followers is quite large; it is N + (N*N-1)/2. This is half a symmetric square matrix with the diagonal added. For this Twitter message the number of comparisions is 8256. We need software that will chug through that many comparisons without requiring us to enter usernames over and over.

Redundancy is a regular part of Twitter messages about politics. Political communication is, in this sense quite unusual. For the population at large the number of retweets is much smaller. Therefore we need tools to help reduce the redundancy, and this has been about one such tool.

References

Boynton, G.R. (2/8/2012) What they had to say

Boynton, G.R. (3/16/2010) Sarah Palin did what? The Importance of Redundancy

Carlson, Nicholas (3/31/3011) How Many Users Does Twitter Really Have?