Research Notes

Organized as self-defense forces, some residents of the Mexican state of Michoácan have been attempting to regain control of their towns from powerful organized criminals. Although these Mexican militias have received a fair amount of media coverage, its fascinating social media presence has not been examined. Saiph Savage, a grad student at UNAM/UCSB, and I have started to collect some data, and wanted to share some initial observations of  one of the militias’ online spaces: Valor por Michoacán, a Facebook page with more than 130,000 followers devoted to documenting the activities of the self-defense militia groups in their fight against the Knights Templar Cartel. We contrast this page with a similar one from a different state: Valor por Tamaulipas,  which has enabled residents of that state cope with the Drug War related violence.

Read the rest of this entry »

Last year, Gilad Lotan and I spent some time analyzing the #YoSoy132 protests in Mexico using data from Twitter. Several articles and even books about #YoSoy132 have come out since. For example, De Mauleón wrote an excellent piece for Nexos (in Spanish) that resembled some of our own analysis.  Sadly, Gilad and I got busy and abandoned the project, but after this recent conversation, we decided to dig out our notes and post them here in the event that they might be useful for others.

The rise and fall of the “Mexican Spring” 

Exactly a year ago, in December 2012, the newly elected Mexican President Peña Nieto took office amid violent protests. As early as May 2012, a number of massive student protests against the then candidate Peña gained a lot of attention on social media, both inside and outside Mexico. The Occupy movement and the international press called these protests the Mexican Spring for its similarities with other “hashtagged” protests. In our analysis, we only focused on the first few months of the protests. Today, #YoSoy132 is only a shadow of what it was, but during the election it was able to accomplish several important victories, including the organization of an online presidential debate (broadcast on YouTube), and the introduction of the issue of media monopolies and media bias to the forefront of the political discussion.

We focused on the origin and spread of the #YoSoy132 student protests by lookign at Twitter trending topics, follower connections, and the content of the tweets. We found that despite the common assumption that the movement appeared “out of the blue,” after an incident involving a candidate’s visit to a university, we can actually trace the movement’s gestation to several months before the trigger incident. Additionally, we found that despite the attempts to link the movement to traditional political groups, i.e. a political party, the movement actually activated typically disconnected groups of people across the political and class spectrum.

Read the rest of this entry »

Write up by Andrés Monroy-Hernández and Elena Agapie, building on the work of J. Nathan Matias

Motivated by the disappearance of local newspapers, this past summer, we started to explore new ways of supporting community news production through collaborative writing tools. The first incarnation of this is NewsPad, a system for neighborhood communities to collaboratively to report on local events such as festivals and town hall meetings.

One of the first challenges we encountered when testing NewsPad in the wild, was the difficulty of bootstrapping these collective action efforts to produce even lightweight articles in the form of lists, also referred to as listicles.

We decided to explore this challenge using on-demand, location-based labor through TaskRabbit. We were able to produce articles about the events in under an hour, and for less than $100. Here we we share some of initial reflections after running a few experiments.

Read the rest of this entry »

For the past two years, social media platforms have been rolling out machine translation in the hopes of enabling multilingual interactions. However, the people interacting in these platforms often know each other already, and have a language in common (i.e., friends). But what happens when machine translation is used to facilitate interactions among strangers, who perhaps have common interests but not a common language?

The earliest social media platform to enable machine translation was probably Facebook, which began autotranslating conversations in Facebook pages (a good place to start given that Pages are more likely to bring together heterogeneous languages). Likewise, Google+ and Twitter later released similar features, enabling, for example, Spanish-speaking Twitter users to read the tweets from the now toppled Egyptian president Muhammad Morsi, translated from Arabic to Spanish:

How often do these types of multilingual interactions occur, though? Ethan Zuckerman posed a similar question when wondering how often people use their browsers’ machine translation to pay attention to content outside their immediate reach.

Read the rest of this entry »

Along with my colleagues Shelly Farnham, and Michal Lahav—and our interns Yuheng Hu, Emma Spiro, and Nate Matias—we have been exploring ways of discovering and fostering latent neighborhood information to help people understand what’s happening in their local communities.

As part of this research, we have created Whooly an experimental mobile website that discovers and highlights neighborhood-specific information on Twitter in real-time. The system is focused, for now, on various neighborhoods of the Seattle metro area (King County to be specific). Whooly automatically discovers, extracts and summarizes hyperlocal Twitter content from these communities based on mentions of local neighborhoods and relevant keywords from tweets and profiles. One can think of Whooly as a neighborhood Twitter client.

Read the rest of this entry »

This post was written live and collaboratively during the town hall meeting using Newspad.

1. Growth in submissions, flat attendance, slightly cheaper

• 349 papers, up from 232 last year.
• Total attendance is flat.
• Increased number of lightning papers.
• No more “short paper” option (partly because lack of time for authors to turn around).
• 3 papers were in the “without publication” option (out of 10 submissions).
• Lowering price strategies: MIT dorms, no lunches, cheaper for students.

2. Where are the social scientists?

Mor Naaman (Rutgers) raised the issue that more social scientists are needed. He commended the conference for having social scientists in the program committee. He wondered if we need to convince people to attend ICWSM instead of ICA (International Communication Association). Someone else mentioned that maybe ICWSM could piggy back on a conference like ICA. Nicole Ellison (UM) thought that might be a great idea. Brian Keegan (Northeastern) raised the issue that ICA does not represent all of social science, and mentioned that other social science conference ask for more than an abstract in response to people wondering whether ICWSM should consider shorter papers. Topically, someone mentioned Sunbelt as yet another social science conference with overlapping topics. Ian Soboroff wondered if we spend too much time thinking how to make icwsm similar to other conferences. He says he attends ICWSM because it’s different. Lastly, David McDonald and Brian Keegan raised the issue of quality assessment as a mechanism to attract social scientists. David challenged the idea that rejection rates are a good proxy for quality, and asked what can, for example, tenure committees use for assessing quality. Brian asked if we could do bibliographic analysis of impact factors (e.g., if you publish at ICWSM you get more citations than other venues).

3. Submissions: Length and Additional Material

Merrie Morris pointed out that this year there was an option to allow 10 page papers instead of 8 (though 8 was still encouraged), and she wondered about the outcome of this. Emre mentioned that we do not yet have data  available  on how many people took advantage of the length increase but his guess was that most people used the 10 page option. One person mentioned that some conferences have no length limit. Winter Mason mentioned that social scientists might prefer to just submit a 2 page abstract instead of a whole paper. Bernie Hogan (Oxford) raised the issue of sharing data, and, perhaps more importantly, sharing algorithms and techniques. The use of open repositories for sharing our work was raised, Winter mentioned using Open Science Framework.

4. Format of the Conference

Brian Keegan asked for more opportunities to hack together to spark collaborations. Birds of a Feather was a good first step. Melroy De Souza (Bing), also advocated for having more opportunities for networking, and asked whether we could have virtual conferences. Emre replied that we do have an archive of all the videos of the presentations. Melroy also asked if we could have other presentations formats, such as panels with experts and practitioners. Michael Muller (IBM) commented that it was hard to share spotlight posters because the space was crowded. David McDonald (UW) commented that perhaps more than one track is necessary.

5. Future ICWSM’s

Planning to create a more formal steering committee, consisting of past papers and general chairs (though who is currently going to be on that committee not announced). Plan to begin planning conference two years out, though no announcement about who will organize 2014 or 2015 conference (they are soliciting nominees for who might do this). Informal plan to rotate more regularly between Europe and North America was mentioned, though no details on what that rotation will be like (every year? every N years?) or on locations for the upcoming conferences.

6. Thematic Suggestions

• Focus on methods, and tools.
• More applicable findings for real world (Melroy).
• Surveillance is a topic where this community could make an important contribution (Roja from UCLA)

By Andrés Monroy-Hernández and Emma Spiro

More than a million Brazilians have joined protests in over 100 cities throughout Brazil in the past few weeks. Since their early beginning as a “Revolta do Busão” (Bus rebellion) to reduce bus fares, the protests now include a much larger set of issues faced by Brazilian society. Protesters are angry about corruption and inequality. They’re also frustrated about the cost of hosting the upcoming World Cup and Olympic Games in light of economic disparity and lack of high quality basic services. Yesterday, as Brazil defeated Spain to win the Confederations Cup final, police clashed with protesters near Maracana stadium for the second time in two weeks.

People turned to social media to share what they saw on the streets and invite others to join in the protests. According to a well-known polling company, a surprising 72% of Brazilians online supported the demonstrations, and 10% claimed to have joined the protests on the streets. For a while, leftist President Rouseff maintained a high approval rate of 55%, down from 63% the year before and still one of the highest for any leader in the world. By June 29th, however, only only 30% of Brazilians considered her administration “great” or “good.”

Timeline

Although the Brazilian movement seemed to appear out of the blue the second week of June, the news about the bus fare increase first appeared in the media back in January. Furthermore, the organization behind the first protests, Movimento Passe Livre (Free Pass Movement), started 8 years ago and had organized an initial demonstration with students on May 28th in preparation for a bigger one on June 6th that attracted a few thousand people. At that point, the protest’s presence on social media seemed to have been constrained to MPL’s blog and the Facebook event for the demonstrations. This changed after the demonstrations were faced with police repression and several videos of people being injured by police were spread on social media. The movement started to gain a lot of attention on Twitter and Facebook and quickly spread to more Brazilian cities. See the following timeline for a longer list of events related to the protests.

Measuring Twitter Activity in the Brazilian Protests
In order to better understand the development of the protests in social media, Twitter in particular, we collected the full set of 1,579,824 tweets posted between June 1st and June 22nd containing the following hashtags: #VemPraRua (Come to the streets), #MudaBrasil (Change Brazil), #ChangeBrazil, #ChangeBrasil, #passelivre (Free Pass), #protestosrj (Protests Rio de Janeiro), #ogiganteacordou (the giant awoke), #copapraquem (Cup for Whom), #PimientaVsVinagre (Pepper vs Vinager), #sp17j (Sao Paulo June 17), , #consolação, and #acordabrasil (Wake Up Brazil).

Tweets per day

Above we show the total number of tweets posted each day. We continue to analyze the data, hoping to expand beyond those hashtags, but here are three things we have found so far:

1. Protests’ tweets peaked on June 17th

The peak of 96,531 tweets/hour happened specifically around 8PM local time on June 17th, 2013. This was the day protesters swarmed the Brazilian Congress. One example of a highly retweeted message this day was one from @AnonymousBrasil reporting on the protesters’ occupation of congress:

Congresso acaba de ser invadido ! #vemprarua https://t.co/vHWdNCUmsr

— Anonymous Brasil (@AnonymouBrasil) June 17, 2013

Tweets per hour – June 15th to 22nd

In the figure above, we show the hourly rate of tweets during the period of interest. Time of day seasonality is clearly visible as well as the dramatic spike in conversation on the night of June 17th. We also look at what is being talked about on Twitter. Below are some of the most commonly used words.

Most common words in the tweets of June 17th

2. International nature of protests.

Half of the tweets came from users whose time zone is set to “Brasilia” while the rest came from a wide range of other locations. The top time zones outside Brasilia were: Santiago, Greenland, Mid-Atlantic, Hawaii, Quito, Atlantic Time (Canada), Eastern Time (US & Canada), London, Pacific Time (US & Canada), Central Time (US & Canada), Istanbul and Buenos Aires.

The relatively high proportion of users from Istanbul was particularly interesting given the similar protests going on in Turkey. The actual number of tweets from Istanbul was very small (5,582 tweets posted by 3,517 different accounts), but conversation rates follow a pattern of delay compared to the bulk of the tweets, suggesting that the tweets coming from Istanbul were posted after hearing the news of what was going on in Brasil (the tweets from Istanbul peaked at 434 tweets/hour on June 18th at 2:00 PM UTC) as seen in the figure below.

Tweets per hour from users whose time zone is “Istanbul” – June 15th to 22nd

The sign says “Turkey is here”, by Juliana Spinola via Demotix

3. Interactions network returns to its beginning.

Perhaps the most fascinating finding is that the structure of the interaction network among the most active users–defined by the @mentions and retweets among the top 1% of users (those who posted at least 20 tweets in total)–exhibits cyclic behavior over the week. The interaction network begins very sparse on June 15th, grows to be more dense on June 17th, and maintains this increased density for a few days before returning to a density similar to its starting point on June 15th. The following plot shows how the volume of interactions among those in the 99% quantile grows and then shrinks.

Shapes of interaction networks over the course of 8 days (June 15th to 22nd)

Moreover, by comparing the structure of these daily interaction networks, we find that the pattern of relationships also exhibits cyclic behavior. In the second plot we show each daily snapshot of the interaction network as a point in space. The distance between points (i.e. daily interaction networks) represents the structural similarity between those networks – pairs closer in space are more similar. The plot demonstrates how the interaction network among these individuals begins in a particular configuration on June 15th/16th before changing drastically on June 17th and 18th (individuals on these days are interacting with many new contacts, with whom they did not previously communicate). By the end of the week, the network returns to a structural configuration similar to the way it began on June 15th.

Network structural dynamics diagram. Each circle represents a daily snapshot of the interaction network. The distance between points (two daily networks) represents their similarity – pairs closer in space are more similar.

Future work

This initial analysis represents an quantitative analysis of the movement’s communication on Twitter using a specific set of hashtags. More work needs to be done to not only expand to the list of hashtags beyond those we used but also to look into other communication channels such as Facebook and face to face interactions.

Future questions to investigate could focus on understanding the roles of each of those channels. Beyond that, the roles and motivations of different actors including unaffiliated individuals, students, and existing political organizations such as MPL, traditional political parties, and collectives like Anonymous.

Thanks to J. Nathan Matias for his valuable feedback during the writing this post, and Andrew Osborne for the help with some of the visuals.

This post was written with Benjamin Mako Hill. It is a summary of a paper just published in American Behavioral Scientist. You can also read the full paper: The remixing dilemma: The trade-off between generativity and originality. It is part of a series of papers I have written with Mako Hill using data from Scratch. You can find the others on my website.

Remix Diagram

Remixing — the reworking and recombination of existing creative artifacts — represents a widespread, important, and controversial form of social creativity online. Proponents of remix culture often speak of remixing in terms of rich ecosystems where creative works are novel and highly generative, however, examples like this can be difficult to find. Although there is a steady stream of media being shared freely on the web, only a tiny fraction of these projects are remixed even once. On top of this, many remixes are not very different from the works they are built upon. Why is some content more attractive to remixers? Why are some projects remixed in deeper and more transformative ways?

We try to shed light on both of these questions using data from Scratch — a large online remixing community. Although we find support for several popular theories, we also present evidence in support of a persistent trade-off that has broad practical and theoretical implications. In what we call the remixing dilemma, we suggest that characteristics of projects that are associated with higher rates of remixing are also associated with simpler and less transformative types of derivatives.

Our study is focused on two interrelated research questions. First, we ask why some projects shared in remixing communities are more or less generative than others. “Generativity” — a term we borrow from Jonathan Zittrain — describes creative works that are likely to inspire follow-on work. Several scholars have offered suggestions for why some creative works might be more generative than others. We focus on three central theories:

1. Projects that are moderately complicated are more generative. The free and open source software motto “release early and release often” suggests that simple projects will offer more obvious opportunities for contribution than more polished projects. That said, projects that are extremely simple (e.g., completely blank slates) may also uninspiring to would-be contributors.
2. Projects by prominent creators are more generative. The reasoning for this claim comes from the suggestion that remixing can act as a form of cultural conversation and that the work of popular creators can act like a common medium or language. People want to remix famous pop stars because people will be more likely to appreciate the remix if they recognize the remixed track.
3. Projects that are remixes themselves are more generative. The reasoning for this final claim comes from the idea that remixing thrives through the accumulation of contributions from groups of people building on each other’s work. Read the rest of this entry »

Even before YouTube and Twitter, incidents like the videotaping and public release of Rodney King’s case of police brutality gave a glimpse of what is now a common occurrence with social media: increased visibility of major societal issues. Examples of such issues are racism and bullying that come to light via particular incidents that gain a lot of attention due to increased access to communication channels. These issues are not necessarily new but the ability for large numbers of people to track them and to collectively reflect and react to them has become more common and at a much faster response rate.

Countries like Mexico, where deep-seated classism and abuse of power are part of everyday life, are seeing these societal issues surface through social networks. For example, in 2011, one of the first incidents of this type emerged via a YouTube video.The video showed two seemingly intoxicated young upper class women in Polanco, a posh neighborhood of Mexico City, verbally abusing some police officers–insulting them by calling them “salary men”–while the officers did not do much to defend themselves. Had it not been Polanco or those women, the situation might have been very different for the average Mexican accustomed to police abuse and corruption. The video caused indignation on social media because it highlighted the classism and impunity that is rampant in Mexican society. The event got a lot of attention on Twitter and it became a popular trending topic under the hashtag #LadiesDePolanco. The use of the English word “ladies” was a clear commentary on classism. Upper class Mexican speech often tends to replace Spanish words for English ones (for example, expensive private schools often ask their students to refer to their teachers as “Miss” and “Mister”).

In 2012, another incident with the same features surfaced on social media. This time it was a YouTube video of a middle-aged man beating a concierge at an apartment building in yet another upscale neighborhood of Mexico City called “Las Lomas.” The incident was known as the #GentelmanDeLasLomas. The same year, the daughter of then presidential candidate, Peña Nieto, was involved in a similar incident after retweeting a friend’s message using the word “prole” (from proletariat and a commonly used epithet for poor people) to attack her father’s critics. The incident was perhaps the first major incident in Peña’s campaign.

This weekend yet another incident of this kind came out on social media. This time it involved the daughter of a government official in charge of consumer protection at the Attorney General’s office. Apparently, the young woman used her influence to have inspectors visit and close a restaurant after not having received the treatment she expected. The issue exploded in social media with the hashtag #LadyProfeco (Profeco is the name of the government office her father presides). The young woman and her father were publicly criticized on Twitter, receiving more than 12,000 and 15,000 messages, respectively, on a single day on Twitter. There were more than 42,000 tweets with the hashtag #LadyProfeco.

Read the rest of this entry »