webleads-tracker

RH100

Editors Lab - The New York Times / Reuters Hackdays

29 March 2014 - 30 March 2014

An incredible Editors Lab at The New York Times

On March 29 - 30, 2014, the best US Newsrooms gathered at The New York Times' headquarters to develop innovative news prototypes. 

The theme of this Hackdays was ‘News as a conversation’: Exploring innovating methods to discover and visualize stories buried in news-related data that engage audiences, invite user interactions and encourage collaboration across the newsroom. Think breaking news, specific events (elections) and opinion polls.

Our White Paper offers an in-depth look at the best ideas developed during the event and provides an overview of its highlights. 

A great partnership with Reuters

Reuters was official partner for the GEN Editors Lab Hackdays in The New York Times and brought invaluable knowledge and hands-on coaching for all the participating teams for the Hackdays. As the world’s largest international news agency providing real-time, award-winning multimedia news content, Reuters was a natural fit to partner with GEN as we drive newsroom innovation together. 

Reuters opened access to their API 'Polling Explorer'. The explorer, http://polling.reuters.com/,  displays the opinions of hundreds of thousands of Americans on a multitude of topics. All surveys are conducted online using a pool of pre-screened participants. The data available was:

2012 exit polling results

2012 results for presidential election

2012-2013 polling data on all topics from politics to lifestyle

 

Mauricetamman

Maurice Tamman (Reuters) assisting the WNYC team

Participating teams

Wall-Street-Journal bbcsquare the guardian logo test huffington-post-logoretouché wnyc-logo-2 Narratively LOGO NO NYC1 CUNY-logo NewYorkdailylogo ColumbiaJschool

And the winner was...

 

NYDAILYNEWS

 

New York Daily News won the Hackdays with NewsQs (see the prototype here).

 ‘NewsQs’ is a tool that lets readers interact with New York Daily News in a whole new way. ‘News Qs’ allows the readers have a ‘live chat’ with the New York Daily News website as if they were talking to a human. 

According to the team, ‘We have data and reporting. As a newsroom we make decisions on how to present it, but we want to let our readers talk to us about what they want so they have curation control, contact and interaction. We embraced the hack theme of news as a conversation. Our idea is to let readers literally converse with our website turning the concept of a live chat into a method of site navigation.' 

The users type a question on the homepage (for example, 'who won the election in my district?'), and the website answers it directly and provide links for more information. The team joked about getting some inspiration from Clippy, the Microsoft Word help tool. 

NewsQs’ project stemmed from the successful live reader chats the New York Daily News has hosted on their website with celebrities. But the tool can be used in a many different ways. It could be used not only to cover elections but also live events such as sports events or awards ceremonies. 

On the technical side, the team explained that ‘the project is written primarily in Javascript, using associative arrays as the main data structure, and leveraging the regular expressions to match questions from the user. We used two main js libraries, highcharts and jQuery to get information from the Reuters polling and elections API, as well as individual data aggregation from the New York State Board of Elections.’

The jury was impressed by the versatility of this tool and its user- friendly interface. They appreciated how well NewsQs fits the New York Daily News’ specific market reaching a young and popular audience that could greatly benefit from this simple, direct and quick way to access information.  

The Guardian

The Guardian was awarded an honorable mention by the jury and won the prize given by Reuters and the audience’s award. 

The team created Pollie, an interface for exploring polling data (see the prototype here). When using Pollie, the users start by logging in with Facebook. Then they are asked questions such as : ‘When you think about the rights of same-sex couples, which of the following comes closest to your personal opinion?’ and they have to choose between four answers. First, their answers are compared with the exit polls answers. Secondly, Pollie identifies a couple of the users’ Facebook friends that are most likely to have extreme and opposing views on the question, based on how respondents demographically similar to them answered the survey. According to the team, ‘the presentation of polling data is often too complex and not relatable. But Pollie allows readers to interact with poll results in a personalized and conversational manner. It asks the reader for information about themselves and their friends and uses that to guide them through a tailored view of the data. We believe Pollie makes data more accessible, and can be adapted for use with any any poll.’

 The team used a Javascript library (Ractive.js) to build Pollie. From Facebook, they pulled demographic information about the users and their Facebook friends. From Reuters, they gathered exit poll answers to questions nationally and for each demographic set. There are two versions of Pollie. In the first one, the users click their answers. In the second one, thanks to Speech recognition, the users can ‘talk’ directly to Pollie. The team used Speech recognition to create a new experience and to increase accessibility for people with disabilities.

According to the team, in the long run ‘it would also be ideal to create an interface where newsroom editors can easily select the questions and polling data to be used in their Pollie project.’

The Jury was impressed by the way Pollie transformed data exploration into a truly entertaining experience. The jury members also praised the neat interface of the project and the technical level it displayed.

 

TEAMSATWORK2

 

Teams at work

Narratively

Narratively built Stac, a tool that allows the readers to select information from an article and easily create with this information a striking infographic and then share it on social media.  According to the team, ‘important news is often bursting with numbers—big, boring, daunting numbers. As a result, data-heavy stories are often hard to get through and even harder to share. With Stac, publishers and readers now have an easy and engaging way to create eye-popping and highly shareable and personalized infographics directly from an article. And consumers have a quick and exciting way to digest the news.’ Facebook and Twitter give the opportunity to their users to share articles in a rather impersonal way. But Stac give them the opportunity to add their own personal twist. 

The jury gave an honorable mention to Narratively. It praised this project as a smart and innovative way to encourage article-sharing and reading and appreciated its sleek interface.

 

 

The Wall Street Journal

The Wall Street Journal developed ‘The Bias Meter’, a quiz that invites its users to see whether their preconceived notions of the American public are right or wrong.To see the prototype, please click here. The users have to guess which group they think is likely to answer a question a certain way. For example, they are invited to answer the question ‘who is more likely to say abortion should be illegal in all cases?’ and need to guess whether it is the male or the female population. The result reveals if their selection was right. Then they have the opportunity to click a link that leads them to the latest news on that topic. On the technical side, according to them, they ‘used JavaScript with jQuery and Underscore to build a library that takes question responses and demographic categories of our choosing, interfaces with Reuters polling API, and creates a static JSON file. Then, (they) used JavaScript with jQuery to create a functional front-end prototype powered by the JSON file.’

 

Wsjnyt (1)

The Wall Street Journal team

The Huffington Post

The Huffington Post built ‘Opinion Googles’, a quiz that allows the users to see how well do they know different demographic groups. They are asked questions such as ‘you are 18-34, female and employed full-time, what percentage of people like you voted for Barack Obama in 2012?’. The question is created by random combination. After having answered, they can visualize the distance between his guess and the demographic group’s actual response. That is how the team describes its project : ‘Many users find open- ended presentations of polling data uninteresting and difficult to understand. Polling data can also be used to reinforce commonly shared opinions instead of exposing differing points of view. Opinion Goggles solves both by asking users to put themselves in other people’s shoes and estimate how those people feel about a particular subject or question. Our project is built in HTML, CSS, and JavaScript, using the D3 library to visualize where a poll result and a user’s guess fall on a spectrum. We generate random combinations of demographic characteristics, and use the Reuters polling API via JSONP to present questions to the reader.’ 

 

BBC

BBC’s project, ‘Quiz: The exit poll explorer’ allows the users to test their knowledge of the American voting public. The users are invited to answer questions such as ‘What proportion of the over 55s have voted for the first time?’. Then they are encouraged to explain the exit poll answer: ‘But what would make someone vote for the first time after nearly 30 years? Any suggestions, let us know @exitpoll’.

The team describes its project in this way :“How can we squeeze the most news goodness out of Reuter's exit poll 2012? This is no easy task as election night is the busiest night in the newsroom. So, we want to inspire curiosity about the exit poll in our readers so they can help us crowd source the survey. We will do this with a quiz which will lead into an API, over which have built a simple UI for our readers.” 

 

 

WNYC

WNYC built Demographic 8-ball, a quiz that allows its users to discover how much do their backgrounds shape their opinions and their habits and where do they fit in to “America's demographic mixing bowl”. To see the prototype, please click here. The user answers a few short questions and then discovers what is his demographic profile according to the answers he gave. He can find out to what extent his background has influenced his political opinions. The teams used HTML/CSS/JS and D3 for charts. Data was pulled from the JSONP endpoint provided by Reuters. The questions were rendered via Handlebars and the Question pagination were done using simple jQuery & CSS animations. 

 

 

CUNY

CUNY University built ‘Surprise! You're 36% Conservative’, an app that, according to the team, ‘uses the Reuter's API cross tabs to identify surprising connections between political affiliation and stances on issues. To see the prototype, please click here. It uses quizzes to place a user in a political spectrum based on how closely their answers match the 2012 exit poll answers. It creates a profile that evolves as the user does more quizzes.’  

CUNYNYT

 

The CUNY team

Columbia University

Columbia University created ‘Vote for your Values’, an interactive map that allows users to see how different political issues had influenced the outcomes the presidential elections. The team used the Reuters API to identify the key issues of the election. According to them, ‘overall, it was great working with the Reuters data set. We had to adapt to wrangling data, play with different models (for analysis) and create visualizations from scratch. We definitely learned a lot from this experience and would highly recommend it.’ 

Nythackdays

 

Brainstorming session at the beginning of the event

Lessons Learned

This competition demonstrated the versatility and potential of the Reuters Polling API. The teams have used it in very innovative anddiverse ways and have avoided the usual pitfalls encountered when presenting polling data. The challenge was to engage the readers with the data. Whenpresenting polling results, journalists tend to give a neutral and general overview of the key findings that is not attractive to thereaders and lacks relevance to their specific interests. To avoid thispitfall, most of the teams have relied on personalization and gamification. The Guardian provided the best example of “polling data’s personalisation’. It achieved to turn the data exploration into a truly personal experience by connecting the polling data with the readers’ personal Facebook data. This team managed to create a real bridge between the polling results and the personal sphere of the users. Most of the teams have built games in an attempt to transform the exploration of the polls’ results into an entertaining experience. Teams such as The Guardian or WNYC seemed to have created the best news-gaming experience because they succeeded in balancing information density with a sleek and a playful design.

 

Exploring the data 

Polling results are dense and difficult to read. Interesting detailed data is buried in polling results. But when looking at the polls’ results,the readers do not have access to it because they do not haveneither the time nor the data literacy to discover it. Reading articles about the polls do not help them find out this detailed data, as most of them focus only on top-line numbers. During election season, journalists are bombarded with data from polls and tend to concentrate on the most obvious findings and overlook the wealth of data buried in exit polls. At the Editors Lab, the teams have attempted to tackle this problem by building innovative exploring features that lets the users make themost of the incredible richness of the API. They have explored the data in two different ways. BBC and The Huffington Post have invited their users to guess the opinions of a certain demographic group, whereas WNYC and CUNY university encourage their reader to identify demographic groups based on their of them focus only on top-line numbers. During election season, journalists are bombarded with data from polls and tend to concentrate on the most obvious findings and overlook the wealth of data buried in exit polls. At the Editors Lab, the teams have attempted to tackle this problemby building innovative exploring features that lets the users make the most of the incredible richness of the API. They have explored the data in two different ways. BBC and The Huffington Post have invited their users to guess the opinions of a certain demographic group, whereas WNYC and CUNY university encourage their reader to identify demographic groups based on their opinions.

 

Interacting with the data 

The challenge was to engage the readers with the data. Whenpresenting polling results, journalists tend to give a neutral andgeneral overview of the key findings that is not attractive to thereaders and lacks relevance to their specific interests. To avoid this pitfall, most of the teams have relied on personalization and gamification.The Guardian provided the best example of “polling data’spersonalisation’. It achieved to turn the data exploration into a truly personal experience by connecting the polling data with the readers’personal Facebook data. This team managed to create a real bridge between the polling results and the personal sphere of the users. Most of the teams have built games in an attempt to transform the exploration of the polls’ results into an entertaining experience. Teams such as The Guardian or WNYC seemed to have created the bestnews-gaming experience because they succeeded in balancinginformation density with a sleek and a playful design. The Wall Street Journal invited theplayers to read an article about the issue tackled in the questionbefore coming back to the game. By adding some context to theirreporting of the results, they helped their users make sense of thedata together with increasing the stickiness of their website.The New York Daily News uses a similar way to transform the polling results in a gateway for more news.

 

Contextualising the polls’ results 

Polling results and news articles are mutually beneficial. Polling results can serve as an entry point to news articles and news articles are essential to make the most of polling results. After giving the answer to each question of its quiz The Wall Street Journal invited the players to read an article about the issue tackled in the questionbefore coming back to the game. By adding some context to theirreporting of the results, they helped their users make sense of the data together with increasing the stickiness of their website.The New York Daily News uses a similar way to transform the polling results ina gateway for more news.

 

Predicting elections and narrativizing the data 

There are other ways in which this API could be used in the future : Predicting elections and narrativizing the data. Reuters API could provide a great resource to build prediction models for future elections. The timeframe of the event was probably too short to allow them to create a statistical model sophisticated enough to predict the results. But the Reuters polling API with updated results could be a perfect tool to create results’ predictions in the same way that Nate Silver did in FiveThirtyEight. Following the theme of the competition: “News as a conversation”, the teams focused on building interactive features but did not explore how polling results could be integrated into a storytelling experience. However this could be a very promising way to make the most of the Reuters Polling API.

Live-blog

The essentials

The Speakers

Charlieszymanski

Charlie Szymanski

Reuters

Coreyfrang

Corey Frang

Bocoup

Maryanne Muray

Maryanne Murray

Reuters

Mauricetamman

Maurice Tamman

Reuters

The Jury

Dannyschechter

Danny Schechter

MediaChannel.org

Davidstolarsky

David Stolarsky

The New York Times

Edebourgoing

Évangéline de Bourgoing