The Civic Project Carousel | MIT Center for Civic Media

The Civic Project Carousel

Rahul Bhargava introduces the Civic Media Bingo session, a whirlwind introduction to Civic’s many interesting projects.

Nathan Matias starts off with NewsPad, a tool he created with Andres Monroy Hernandez at Microsoft, and Eventful, which Andres and Elena Agapie created to carry on the work. Nathan sees NewsPad filling the need for events that don’t currently get reported on, like neighborhood yard sales. Nathan cites Wikipedia’s cooperative online news reporting as the exception in a field of single-user news curation tools like Storify. Newspad is designed to let curators pull together a seed post that additional contributors can join.

The app structures the posts as listicle-style entries in the spirit of modern news tactics. Every section of the article allows you to invite friends and comrades to contribute. This summer, the team tested the platform in Seattle.

With, Andres and Elana reverse-engineered the news article and broke it into discrete tasks that can be assigned to a team of contributors. These tasks can even be delegated to TaskRabbit, where the dynamically allocated labor pool helps ensure that tasks like interviews are completed. The mobile app provides a checklist of remaining tasks.

Microsoft Fuse Labs has open sourced NewsPad and Nathan is continuing to develop it.

Catherine D’Ignazio is interested in the “where” of journalism. With Rahul, Catherine built CLIFF, an open library for accurate geoparsing of text. Catherine’s applied geoparsing to studying the Boston Globe’s geographic reach, international coverage of the Boston Marathon bombing, and finally, to engineer serendipity. The internet arrived with hopes and dreams for engagement and cosmopolitanism, and we do have unprecedented access to information. But our tendency to group ourselves with others like us (homophily) combined with corporate algorithms designed to give us only what we immediately desire has undercut the original promise of the web.

Catherine’s project, Terra Incognita (assisted by yours truly), lets you explore 1,000 of the world’s cities through the lens of the news. When you open a new tab, the Chrome extension presents you with a randomly selected city and some articles about that place. The articles come from a wide range of sources, like Global Voices, Wikipedia, and traditional news sites.

Sourcing news recommendations for 1,000 global cities proved a challenge, which Catherine solved with content from Instapaper, human curation, and Bitly. It’s surprisingly difficult to find English-language content about many places in the world.

The results of the user study are starting to come in, and 62.5% of users showed an increase in geographic diversity in their newsreading after installing Terra Incognita. Catherine’s working on releasing a public alpha and is looking forward to collaborating with you.

Billions of people watch and post online videos each month. Ed Platt (@elplatt) took YouTube’s country-specific trending lists to learn about each nation’s culture. Some trends span entire geographic regions, and their popularity tells us something about common cultural touchstones.

Ed and Rahul built an online video explorer, What We Watch, to help us browse trending videos around the world. We can even compare nations to find surprising similarities in very different places. Looking at 57 countries over 7 months, the team found a decentralized but highly-connected mesh of nations, where each country shares at least one video trend with each other country. There are no particularly trend-setting nations, and videos are definitely crossing boundaries.

When we zoom in on the differences between national clusters, we find existing cultural divides represented in the data. Countries with high migration, like Australia, serve as bridges to other parts of the world. One of the most universally popular videos is a baby cycling through the full range of human emotion.

Chelsea Barabas (@chels_bar) cites an old New York Times poll that found that 57% of Americans still believed in upward mobility, and surprisingly, this belief has remained strong despite the economic ruin of the recession. The cost of a college degree has skyrocketed while the return on that investment has diminished. Debt and unemployment are at all-time highs. But these hardships aren’t evenly distributed across the US population. There’s a disturbing trend of downwards mobility for college graduates in the lowest income bracket.

We share a cultural narrative of the tech startup, and the kids who teach themselves to code and go on to found Microsoft and Apple and Facebook. There are lots of online resources for anyone looking to learn to code, and a strong sense of meritocracy in the tech sector. But as we now know from Google and Yahoo!’s employee diversity figures, some groups have definitely benefitted more from the ascendance of the technology economy than others.

Companies like CodeSchool and Udacity promise upward mobility by providing part-time education in economically strong careers like software development. Chelsea’s partnering with Code2040 and their 27 fellows to learn how they apply their newly acquired tech chops to find employment upon finishing. Chelsea’s going to investigate how Silicon Valley companies think about diversity and meritocracy in their hiring practices.

Sands and Ali (@sandsfish and @alihashmi01) look at the problem of overwhelming, ubiquitous data that doesn’t have an obvious narrative. They’ve developed an algorithm that discovers clusters of topics and uses unsupervised machine learning to detect key themes from unstructured text. The tool works with the Media Cloud platform, where they’ve applied it to coverage of NSA surveillance and net neutrality debates. At first glance, we get broad frames like “government” and “Obama” and technical themes like “access” and “data”. Eventually, we get the vocabulary used to convey the issues to the general public. Suddenly we see a lot of mentions of Netflix, as the streaming video service is used to explain net neutrality. The tool is available on the Media Meter Dashboard.

Media Cloud is a huge database of news articles and blog posts collected over years to facilitate media research. It lets researchers understand things like the Russian blogosphere and the Trayvon Martin story and the SOPA-PIPA debate. This research is hard to conduct because collecting the data and parsing it requires a high learning curve. Media Meter is a suite of front-facing tools built on top of Media Cloud that allows researchers to conduct media analysis research on their own.

Media Meter Dashboard allows you to construct your own queries of the massive database, limit your date range, and filter your results by source. It’s available in limited beta at the moment.

Mentions finds your subject across online media. Pulse tracks the ebb and flow of mentions of your subject over time. Frequency, developed by Deborah Chen, finds the related words mentioned alongside your subject to help you understand the context within which your topic is discussed. All three are coming soon.

Over the summer and fall, each of these widgets will become a full-fledged tool to help scaffold quantitative media research. Catherine’s CLIFF geoparsing work will become Media Meter Focus to help researchers explore media and geography. You can sign up for the limited beta on the Media Meter website while the team stress tests the hardware behind these tools.

Tal Achituv is a HAM radio operator and was intrigued by the gap between the information found on police scanners and the information that gets reported on in the news. When a big event happens, reporters go and ask the police for the radio transcripts, but there are issues with this approach.

If we want to be aware of what’s happening around us, how do we allow access to this public information? When it’s originally broadcast, this information is public domain. But once it’s been recorded, only the police have copies. This becomes problematic if we want to hold the police accountable. There are commercial solutions that allow police departments to record their channels, but these services aren’t open to collaborative analysis.

Tal’s building an open source hardware and software platform, RadioDVR, to allow for archiving, collaboration, and eventually transcription, search, and other analytics. You could search for a given address. He’s looking to work with journalists to make it a useful tool for newsrooms.

Tal shows us screenshots of a less dramatic use case: his daughter’s baby monitor. He’s visualized heatmaps of chatter on each channel, so you can see if you’re missing conversations elsewhere.

Ali Hashmi and Julia Belluz (@alihashmi01 and @JuliaofToronto) want to use big data to re-calibrate health journalism. Health reporters don’t often do a great job translating scientific findings to their popular audiences. According to the CDC, cancer, heart diseases, and chronic diseases are the biggest killers of Americans. Cancer gets a lot of play in the press, but emphysema, one of the leading causes of death in America, has been mentioned only four times in the last year of The New York Times’s coverage.

The Health Gap is a prototype to visualize the disparity between a disease’s actual death toll and the amount of attention it receives in The New York Times. They’ve also looked at the Times of India, where the neonatal disorders that kill many people receive hardly any media attention. They’re looking to expand their focus to more countries and more media sources, and allow users to upload data from their own websites to self-audit. In this way they hope journalism will self-correct to better allocate their research and spotlight.

William Lee (@williampli) is a PhD student in Computer Science at MIT and built a tool called BillTracer in Ethan’s Participatory News course. The Dodd Frank Act is 848 pages. Have you read it?

William mapped the votes leading up to the passing of the Troubled Asset Relief Program in 2008, and is able to map the votes and bills leading up to it to see where its content originated. We get a visual sense of the ideas Congress considered in various pieces of legislation and a clear picture of when these ideas made it into law. We can expand our search to whitepapers and other documents to trace the ideas in our legislation back to their sources.

Government data often lives in opaque PDFs. But the open government community has had success obtaining, cleaning, and parsing government data. William’s interested in using computer science to recover the structures lost to transliterated data with automatic structure recovery. He’s also conducting new forms of analysis, like using machine language analysis to unmask the authorship of unsigned Supreme Court decisions.

NGO 2.0 (@NGO20) works with Chinese NGOs to provide communicates and technology solutions. There are ten projects, but Yu will focus on two. They’ve created a collective map of NGOs including their events and partnerships. The NGOs who began using this map soon began finding one another. Companies have used the map to share their Corporate Social Responsibility efforts.

NGO 2.0 also runs technology workshops with a curriculum including a wide range of cloud-based social media, CRM, and communications tools.

Ed Bice (@edbice) works with the Meedan team across San Francisco, Vancouver, Cairo, Tuscon, Brazil, and several other corners of the world. Ed’s presentation is titled 21,000 Miles of Social Media. “Like many wonderful things in my life, this project is downstream of an email from Ethan”.

Paul Salopek is literally walking the path of human migration over 7 years and 21,000 miles (see National Geographic’s Out of Eden walk coverage). Meedan is following Paul along the way, collecting 100-mile social media samples at each location. Ed describes it as a form of modern anthropology.

They’re using Checkdesk, another Meedan project, to translate and embed the social media posts. 10,000 social media posts become 100 translated, contextualized artifacts. The audience gets a thin section of humanity across the sample points along Paul’s global journey.

Ed brings up Mona Seif’s Facebook Page: “If you want to understand Egypt today, you need to be able to understand this page.” They also translated poetic Farsi tweets into English with a helpful [trans] denotation preceding the original tweet.

Like Paul’s walk, this is a 7-year project. They hope to build a more participatory platform and robust translation support tools. As Paul traces humanity’s path, Meedan’s tracing the arc of creating the cross-lingual web. “All that connects me to you is a bunch of electronic devices. If humanity goes back to its primitive state, I would lose you forever.” language is just as important as connectivity. The world wide web is a thousand siloed webs right now, and won’t be open until translation is baked into the foundational platforms of the web: Facebook, Mozilla, and others.

Dan Sinker (@danswinker) shares the results of the Knight-Mozilla OpenNews hack weekend that took place at the Media Lab preceding this conference.

When Russia annexed Crimea, there was some media discussion about the fact that different populations of Google Maps users saw different versions of a border line between Crimea and the Ukraine. Alyson Hurt and her team built a site to help compare the different versions of border lines that Google serves up to users in different countries. It’s the Filter Bubble applied to geography. All the team had to do was load the country-specific Google Maps URL (like to change Google’s representation of nationality. The code is up at Github.

Keyblur is open source file sync and share inspired by remote rural networks built with the ability to deliver data by truck when the Internet went down, as it frequently did. The project combines OpenWRT, Git, and a tool similar to Google’s Pond project to encrypt data in physical locations, like a dead drop.

You can stay up to date with the Center for Civic Media’s projects on this page and by joining our email list.