On the evening of Friday, Nov. 13, I was writing a report about a research project and I needed to confirm some facts. Although it had been a long week, and we had finished late that afternoon, I emailed my team at the CulturePlex Lab.
The project deals with the perception of entrepreneurship transmitted by big media – the New York Times and Financial Times – that we had been doing with colleagues at the Ivey Business School since last summer. It involves collecting, extracting and classifying nearly 2 million sentences from the digital versions of the papers. It also requires we analyze the emotions associated to those sentences using several techniques from artificial intelligence, including sentiment analysis.
The email exchange derived into a note from Javier de la Rosa, a PhD student and software lead in my lab, saying: “It is a good moment to collect tweets: I have set it up to ‘Paris/Japan,’ because of the earthquake, and a 100 million tweets limit. Let’s see what we capture.”
It would still be a few hours before we learned about the dimensions of the Paris attacks, the number of dead and wounded, and the reactions of the public and governments to the new war enemy.
The following week started as usual – attending to classes and paying attention to grading, meetings and grant applications. On Thursday, the whole lab met to do a design thinking session on another project, an ongoing collaboration with the Central Bank of Colombia about the peace process in the country. Again, part of the project deals with assessing the emotions of the population about the peace agreement with the guerrilla and how this evolves as Colombia goes into a new historical phase after 50 years of internal war.
Someone suggested using the transcripts from radio broadcasts, while some other mentioned Twitter. The problem, we said, is Twitter does not let you collect tweets from the past unless you reach a special agreement with them. On Twitter – as opposed to Facebook, that has a more restrictive policy – you can collect anything you want provided it is public and happens one second after you start collecting. And you have to have enough storage.
At that very moment, Javier mentioned we had the collections from Paris and we had not done anything with it yet. For the next three days, including another long weekend, David Brown, PhD student and self-taught programmer, devoted himself to parsing, cleaning, analyzing and visualizing all the information from the Paris attacks.
In a period of 29 hours following the attacks, we fished 4.3 million tweets by more than 2.5 million users who had used the keyword ‘Paris’ in their messages.
But, what does all this mean?
This image shows a network composed of tweets and hashtags. Tweets are represented by colored nodes (coloured according to country of origin), and are positioned according to where they were produced. These tweets are then connected by edges to the hashtags they use, which are represented by white nodes.
First, social media is here to stay. Never before in human history there had been a device – in this case Twitter, but the same is applicable to Facebook, Whatsapp, Kik, Weibo and Wechat – capable to convey and gather such a wide and immediate response. Being able to bring together such large numbers of people through the use of language and other cultural artifacts implies having a lot power, making possible to form communities and enabling the building of huge cultural networks.
Second, what happens in these places is for real. There is a movement that tries to play down the important of social networks – it does not matter, they say. However, ISIS recruits in social media. People buy, talk, date, study, create, listen to music and organize their trips on social media. Basically, humankind has created a new form of doing all things human.
Third, we don’t know enough about how these networks work and disseminate ideas and behaviors.
In order to analyze the tweets we collected, we built a network connecting users to hashtags. The resulting network had 13,615,818 edges. This is quite a large structure, almost impossible to visualize in graphic form, but smaller than the network of the brain, many ecosystems, the one connecting the 1.3 billion people actively using Facebook all around the world, or the human networks that powerful viruses use to create a global pandemia. In order to study them, you need a lot of computing power, knowledge of network physics, the mechanisms of community formation and the language and cultural elements that make groups of people tip their behavior one way or another.
And we need to learn more about all this.
Fourth, the world is not so flat. The analysis of the network of tweets shows that these structures of communication are easy to join in but difficult to control centrally. They show a complex organization with some general rules and many regional behaviors, islands and groups. One-fits-it-all approaches only take you that far in understanding and affecting people attached to networks. Many of the issues surrounding human behavior in social networks have to do with cultures and how our brains interact with cultural objects. The different degrees of reaction to similar attacks – Mali got 400,000 tweets in double the time – prove some things, and one of them is that there is room for variation and difference in global networks. They have various topographic shapes.
Fifth, celebrities and social media specialists know how powerful these networks are to create buzz and market their products. One tweet from One Direction member Louis Tomlinson during the Paris attacks summarily proved the true impact of celebrity and the existence of the global connectivity. One tweet by Tomlinson at 6:47 a.m. Nov. 14 was the most active and widely shared Twitter message related to the event with more than 173,000 retweets and 208,000 likes. Super-connectors like Tomlinson, who has 20.7 million Twitter followers (and counting), have an unbelievable amount of power to persuade.
Social media makes it a perfect mechanism to plants ideas and purchasing behaviors, from music and entertainment to politics and gossip. Young people are especially vulnerable because they live there. The fact the most retweeted message (38,199 retweets) around Paris was by a ‘boy band’ member represents a wake-up call to the influence celebrities have these days – and the responsibility that comes with it.
In sum, cultural networks like the one immediately formed around the Paris attacks on Twitter have become a vehicle to our understanding of humankind and human behavior. If humans have always coalesced around words, rituals, books, music and images, the main change we are living today is these networks are both global in reach and very fast in its formation.
We need to better understand how they work if we want to understand and affect the way we, humans, live in the 21st Century.
Digital Humanities professor Juan-Luis Suarez is the Director of the CulturePlex Lab.
