We’re here at the 2013 MIT-Knight Civic Media conference here at the MIT Media Lab, where the theme is Insiders/Outsiders. Across the next two days, we’re going to be looking at this theme of institutions and innovators across the areas of government, media, and disaster response. Across the event, speakers will be asking if it’s better to look for change inside institutions or try to transform things from the outside.
This session, Open Space & Place, will be split into two sections.
The first half features several speakers who will talk about civic spaces as where communication occurs, where things are shared, where a community gathers. Civic spaces can never be truly neutral territory – online or offline. With so many privately owned public spaces, we consider the shifting nature of civic space and the risks of both being visible and invisible. The liveblog below is a best-efforts summary of this section. Thanks to Chris Peterson, Erhardt Graeff, Nathan Matias, Joanna Kao for helping liveblog.
The second half explores the building and use of spaces–public vs. private, real vs. virtual–for engaging in public discourse. Attendees will break into smaller discussion groups to explore a variety of concepts within Space and Place. The second section will not be blogged as the bloggers will be otherwise engaged! You can find a list of breakout sessions here.
Andres kicks off his talk with a discussion of hashtagged collective action, citing three main case studies: #Narcotweets, #YoSoy132, #MudaBrasil. In the U.S. typical emergency response relies on the police and the press; in Mexico, however, this is not the case. There, drug cartels dictate the media coverage — threatening journalists and government officials into a complete news blackout. “If you’re a citizen in this scenario and you see shootings and car bombs in the streets but all you see on TV are soap operas, you have reason to be concerned.”
“Risky situation” is a euphemism often used in Mexico to refer to shootings, car bombs, and other dangerous occurrences. On Twitter, crowdsourced news reports fill the gaps created by these news blackouts. Citizens do the work of traditional media, creating an on-the-ground alert network. He has published “Narcotweets: Social Media in Wartime” and The new war correspondents: the rise of civic media curation in urban warfare about risky situations and social media.
Andres shows us various data visualizations of Twitter activity. Three groups emerge: people who have a lot of followers, but don’t tweet a lot (e.g. traditional news organizations like @CNNEE); average citizens who happen to be at the wrong place at the wrong time; and finally, what Andres calls “curators,” who contribute significantly to the conversation. They become the de facto hubs of information, aggregating information based on tips they receive.
If you look at the retweet network, you will see the same individuals bubbling up to the top. Andres interviewed two of them: Angela with 25k followers and 35k tweets and Claudia with 30k followers and 65k tweets. Angela describes herself as a journalist and Claudia describes herself as another citizen providing information.
Another example of Mexican coordination on social media was the YoSoy132 movement during the last presidential election.
Pena Nieto, running for president, was one of the most hated politicians in the country. He went to a private university in Mexico City, and a lot of people protested against him. The next day the press coverage reframed the story, portraying his visit to the school as a success. There was a disconnect between what you saw on mainstream media and what was on social media in tweets and YouTube videos recorded during the event. This spawned the #YoSoy132 movement in reference to the number of young people said to be in the video.
With the focus on social media, you saw political parties launching twitter bots to advocate for their candidate.
Despite the fact that this was a very organic, student-driven movement, #YoSoy132 caught the attention of the other political candidates and led to a grassroots-organized presidential debate. Pena still won the election in the end, due to aggressive campaigning in rural areas.
Twitter has also been an important organizing tool in the recent Brasilian protests that started just last week.
Andres walks us through a number of graphs based on #MudaBrasil and related hashtags, showing international proliferation of the movement and then finally a return to the original core community.
Kati London, “Entertainment for Civic Engagement”
Entertainment is something developed specifically for the purpose of keeping the audience’s attention. Within entertainment there are games.
Invoking the theme of the conference, Kati brings up Huizinga’s framework of play and the concept of “the magic circle.” Games give us the permission to fail and behave really differently, allowing us to compete and collaborate. First, she brings up the example of You Are Not Here: A Walking Tour of Gaza City on the Streets of Tel-Aviv.
But mostly today she is talking about the games she made at Area/Code, the platform-agnostic game company that eventually became Zynga NY. For example, Code of Everand was a game that taught kids about traffic rules and road-crossing safety habits, developed in cooperation with the government. Kati discusses how every aspect of the game was mapped to some part of the child’s experience: monsters were designed around the base aesthetic frame of common U.K. cars; roads were mapped to actual roads and made colorblind accessible, and generally rooted deeply in the experience it was trying to educate about.
The next project Kati brings up is Macon Money. It is a matching game where players get half of a bond and they have to find the person with the matching half. Once the player finds a match, they can redeem it for money and spend it locally. The whole process was documented and adjusted on the fly based on the connections being made.
Based on this local currency, Area/Code wanted to create a simple, ritualized process of building connections and expanding the community. In the first 6 months, over 1000 matches were made. By the end, Macon Money had 3000 unique players, with 1 in 5 playing more than once. $11,000 was disbursed to the leading vendor and a total of $65,000 distributed. But the main result was to build connections among the local community which strengthened relationships between consumers and local vendors.
Finally, Kati brings up Battle Storm, a game designed for teenagers in the Gulf Coast. The goal was to engage these youth and their families in hurricane preparedness. The game created a physical analogy for storm preparedness: one team played as the town, the other as the storm, in a game similar to Ultimate Frisbee but with additional layers of complexity, such as resources.
The community children could document their own hurricane preparedness online, and that would allow them to upload and share information with local Boys & Girls clubs. It kind of crescendoed to this large community event when they had actually been training the whole time to “play” against the hurricane.
Kati wants to leave us with the future of these opportunities. There’s a range of ways to think about insiders vs outsiders, and games can split the difference by meeting people where they are, on their own terms.
Mike Ananny – “How Associational Algorithms Do Public Work”
Mike thinks public life and public space require contestation. They require discomfort. One of the places where this may arise is through the design of algorithms.
Algorithms associate different concepts with each other. Oftentimes in ways we don’t expect, and in ways which may not have even been intended. Mike shows a series of slides which show odd associations produced by algorithms, categories, and systems; eddies in the flow which may lead to the emergence of unexpected social turbulence.
Associational algorithms create connections between two or more categories and do public work when they make categories contestable. He draws on the pragmatist philosophers in his epistemology: associations are seen to be “true” when they “work”, and their breakages are evidence of an era’s public life, where what it means to “work” must be contested and reconstructed. Public life requires contestation, but the power to describes is unevenly distributed, not only among people, but among algorithms.
Public life has always depended upon infrastructures for counting, categorizing, aggregating, and associating. For example, issues become acceptable for public debate (Hallin 1986, Herbst 2001). Forms of associational life can be imagined, citing Calhoun 1998 and Taylor 2002. Opinions are popular enough to be expressed, Igo 2007, Noelle-Neumann 1974.
Mike brings up the case study of Grindr—”it’s a way for men to hook up with men, but it’s a lot more than just a hookup service, it’s a place to figure out what it means to come out in a rural place.” When Mike went to install the app in the Android store, the top related app was Sex Offender Search, resulting in inadvertent associations. Mike wrote a piece on this association for The Atlantic, and was contacted by someone from Android. Hours later, the related apps were totally different, and Sex Offender Search was replaced with other Grindr competitors. A week later, and the related apps section for Grindr was completely gone.
The related apps are determined by an algorithm that extracts, weights, and sums the features from the app. There was a bug in the weightings, but an “engineer did not change the algorithm,” the Google rep insisted. As a result, no one person is responsible for this issue. This means it’s trick in terms of accountably.
In the comments and conversation he had with the Google rep, Mike heard folk theories of what the relation between the apps meant. Some of these included “both apps relate to sexual deviants [and] perverts” and “Get somebody in your life, and things like this become non issues”
Mills 1959/2000 writes about how folk theories of algorithms have public significance because they are “sociological imaginations” — it is a process of model-making. Mike breaks down the theories behind the Grindr-Sex Offender Search associations: “mirrors of reality,” populist indices, and unfortunate, unintended aggregation errors. “Maybe bad algorithm associations are the product of bad marketplace logic,” says Mike.” The power to re-describe is not distributed.”
John Keefe (WNYC)
John kicks off his talk by looking back at WNYC’s work on Hurricane Sandy. Leveraging open data from New York City, the WNYC team created a number of news applications that allowed people to follow the storm’s trajectory, identify evacuation zones, and track the status of transit systems in relevant, accessible ways. Other projects included Flood Gauge Watch, monitoring water levels and flood risk and the Cicada Tracker.
Cicadas emerge when the temperature in the soil 8 inches down is 64 degrees. To predict when cicadas will emerge, WNYC put together an Arduino project called Cicada Tracker that measures the temperature in the soil. They were hoping maybe 50–60 people would build these and participate — instead they got 1600 readings, and two spin-off versions of the device that were created by listeners at cheaper costs to build.
They are planning a second experiment this summer tracking particulate matter across the city using sensors toted by cyclists.
The technology is there and it’s getting cheaper. The crowd is there. The WNYC/Radiolab crowd was easily mobilized. Now it’s just a question of journalism and finding the story. Where can we go from cicadas?
Keefe wants to do a story with sensors and texting platforms to empower people in the the NYC Housing Authority. There are more people in New York public housing than in Oakland, Cleveland, New Orleans, or Minneapolis. Their complaints are notoriously backlogged. What sorts of feedback tools, sensor systems, and other networks could help public officials predict problems in public housing as well as they predicted the emergence of cicadas?
Kate: Let’s talk about the politics of being visible. For example, in many of the spaces mentioned in your work, being visible can put you at risk. Do you want to talk about the politics of visibility and risk and how they relate to questions of civic space?
Andres: It’s an interesting balance between giving visibility to the problems that are happening in the streets and the power that the criminal organizations have behind closed doors. One of the challenges I see is like the case in which one person with a Facebook account keeping track of supporters was threatened by cartels and had to escape. But people are still posting on this page with their “real” name and Facebook profile despite being warned of the dangers. People are still trying to understand what it means to be (in)visible in these spaces.
Kati: One of the groups I’ve worked with in the past is populations that self-select. There are all these voices that are present but don’t have agency. It’s a tricky dance to serve unserved needs, especially when you think about funding. The goals of the funding organization can skew where a project goes. It’s a tricky minefield that’s worthy of exploration and investigation.
Mike: I think about it in terms of two dynamics. The agency point, for sure. By engaging in some service but not wanting to be included in the narrative of another service — not just in terms of privacy. The other point is in terms of scale: there’s design for large scale, but it might not work at scales larger than what’s needed for statistical significance or advertising revenue, etc. How does scale connect to questions of agency? It can be large scale, it doesn’t have to be large scale.
John: I think in journalism when you go up and ask people to participate with your story on the microphone they have a pretty good sense that they will be on the radio. But even when we ask people to do very silly tasks like taking pictures of the snow in their backyard, and quite explicitly tell them that we will put them on the map, there’s a level of responsibility that we have there to make sure that they know that their location and their picture will be broadcast on a website. For the most part it’s not a risky situation but I can’t judge that for everybody and we have to make sure to let people know. As reporters we sometimes make the decision for people whether or not to feature them because it might endanger them, but when we give them the facility to upload for themselves, we remove ourselves from a position to do that.
Andres: Something that Sasha mentioned in the Twitter backchannel is on research around the Occupy movement. The feedback to the research was: how can we use these tools to look at the top 1% instead?
Laura Amico of Homicide Watch: Loved what John was saying about what we ask of our audiences. Hoping that the group could talk about how they phrase that “ask” of their users and what they hope to get out of it.
John: We ask people to participate every day, from calling in to tweeting to putting photos on a map. Sometimes it’s just about participating in the conversation. But the interesting thing is when we can collectively do something. Sometimes that’s just doing something cool like chase cicadas. We didn’t have to ask very hard. They just did it. There’s a 28 step process but people still did it. Depends on what people feel compelled to do, utility for themselves.
Sasha Costanza-Chock: (clarifying Andres’ point) We looked deeply at the Occupy Wall Street research. We kept hearing “we think this is really cool but can we use this to map the one percent?” But to ask a question: really interesting in developing this thought about how algorithms reproduce structural inequality. How is this issue addressed, or challenged?
Mike: I don’t have a ton of other examples to talk about right now, but I do want to broaden the concept of associational algorithms beyond just computational algorithms, because they can actually be anything that draws associations between actors. One of the things computational algorithms offer us is the way to make visible some things that have been less visible or less stated. So the city of Boston has been using accelerometers to detect potholes, but we should always ask the questions of what conditions that data are gathered; we might be mapping some set of circumstances and not the underlying phenomena that we actually care about. It makes visible a lot of inequalities that are there and potentially makes them visible to then be challenged.
Kate: As a little rejoinder to that, it’s a really important space that’s opening up. We know that a lot of algorithms are actually drawing on structural inequality to produce the results they want. For example, credits are produced in such a way to appeal to people of affluence.