Data Journalism and Disinformation: Notes from SXSW 2018

By Christian Marc Schmidt

Spending several days at this year’s SXSW Interactive Festival, I focused on panels and presentations revolving around data journalism and “fake news” in social media. I sat in on a number of fascinating sessions that shed light on today’s media environment and the issues we face, as well as the opportunities. Below are some of the notes I took down.

Disinformation and the Effects on Democracy

This panel brought together Mark Jacobson of Georgetown University, Samantha Bradshaw of Oxford University, as well as John Sipher and Molly McKew of New Media Frontier. They discussed the proliferation of disinformation in social media, and how it is affecting our culture and democracy.

Instead of bringing us together, social media is splitting us apart. It has created a world where people trust their friends more than they trust credible sources, and this is having a polarizing effect on society.

Social media amplifies disinformation and propaganda, which now have a reach that is both wider (reaching many more people) and deeper (targeting specific micro-groups of people during times when they are most vulnerable). This is affecting society at its core, including the outcome of our elections.

Traditional media is also caught in the attention economy. Its principle is that where the attention goes, the money will inevitably follow. In this economy value is measured in clicks and eyeballs, and serious journalism is being replaced by clickbait.

We need to become more sophisticated about how we receive information and move away from punditry. The solution to the problem is not technology, since this is essentially a human problem. Just taking down fake accounts isn’t going to solve the problem. Legislating against the proliferation of fake news is only putting a band-aid on the problem.

While education is the solution in the long-run, the immediate, short-term solution is that we need good local storytelling. We’re all looking for people like us to listen to. We need to give local voices a bigger platform, to explore issues we are all affected by.

Hacking our Democracy and Discourse

In this segment, Brian Barrett of Wired interviewed Senator Mark Warner on the effect that Russian cyber attacks and disinformation campaigns had on our 2016 election.

The facts are that Russia massively intervened in our elections to support Trump’s candidacy, and sabotage Clinton — something Warner called the “Weaponization of information”. It is also known that Russia and its agents tried to intervene with 21 State’s electoral systems, and that they used social media in an unprecedented way to influence public opinion. One of the most interesting aspects of what happened is that it was (and remains) unclear who needs to take responsibility. Are Facebook and Twitter responsible, or the federal or local governments?

It is clear that while platform companies need to start engaging more in the debate and taking responsibility for content published on their platforms, the government also needs to step up their game on a “cyber doctrine” and measures to prevent electoral fraud that enable foreign powers to influence the results of American elections.

Trust in “The Media”: What do news consumers want?

Joy Mayer of the Trusting News Project talked about how journalists can reclaim the trust of their audience, which in the US is currently at an all time low with only 32% of people stating that they trusted the media in a recent 2018 Knight Foundation/Gallup poll.

She tells journalists to focus on building trust via the following methods:

• Differentiating themselves from “the media” by communicating their values and explaining their motivations and goals;
• Describing their ethics, as well as funding sources, including how they check for accuracy and handle corrections, as well as explaining where the money comes from and what it is used for;
• Explaining their process and demonstrating balance, i.e. why their reporting is fair and accurate;
• Showing who they are, and being accessible to their audience — people don’t trust people who they don’t understand and who aren’t responsive.

In summary, journalists have to accept that actively earning trust is part of the job. Journalists can help people decide where to “spend their trust” by imbuing the trust building messages into the journalism.

Investigative Journalism in the Social Media Age

Sonya Gavankar of the Newseum interviewed Washington Post journalist David Fahrenthold. Fahrenthold won a Pulitzer Prize for his reporting on Donald Trump’s charitable giving, for which he turned to Twitter to fact-check his investigation. Twitter helped him get to the truth behind where Trump’s charitable donations had gone, after his conventional methods and sources didn’t yield any results.

According to him, social media—while often a platform for snarky punditry—also has the power to demystify journalism and give people more trust in the work of journalists. Journalists “solve mysteries,” and that is why people are interested in their work. Fahrenthold recognized he could use social media to amplify questions he is exploring, to get help as well as different perspectives. Social media also gives him the opportunity to share his process with others, and transparency builds trust.

Social media is also changing the way stories are delivered. Line of coverage is important. You can draw out the process of unraveling a mystery and keep a story in-front of people, as opposed to doing it all at once in a single story.

Data Journalism: We’re All News Nerds Now

Focusing on data journalism, this panel moderated by Jon Loyens of data.world assembled Allison McCann of VICE News, formerly of FiveThirtyEight, Sarah Cohen of the Walter Cronkite School Of Journalism, formerly of the New York Times and Washington Post, as well as Troy Thibodeaux of the Associated Press.

They discussed how journalists might approach the challenges posed by a lack of data literacy. Cohen mentioned that Data Journalism is a lot like being a scientist — you’re in the field where you collect your field notes, then in the lab where you analyze them. This means that if there’s a widespread feeling about something in the world, one should be able to see it in the data. Thibodeaux told the audience that journalists had an obligation to explain things to the audience in a way they could understand. This obligation needs to be balanced with the novelty of the analysis and the visualization approach.

The principle of reproducibility was also a point of discussion. Work needs to be replicable. With the right methodology, others should be able to follow the creator’s steps of analysis and get the same results. This principle has mainly resulted in journalists putting work up on Github and providing documentation to help people recreate something.

On fact checking data journalism, Cohen said that most newsrooms have very stringent fact checking processes, and while much of data journalism is organic, certain aspects of her process are formalized, such as maintaining a data diary and notes that are made available to others to review and verify. Thibodeaux floated the idea of what would it mean to write automated tests in your code for your data analysis, the same way that developers write automated tests for their code.

McCann finally talked about how data journalism can also be fun and viral. While much of data journalism has focused on sports and politics, entertainment and culture are areas that deserve more focus and could lead to large impact. She also spoke about the current trend of data visualization swinging back towards simplicity, to a large extent based on getting back to graphics that work well on a phone and are overall simpler to understand.

Data-Driven Storytelling: Perspectives & Paradigms

During this panel, Amy Yu from Viacom moderated a discussion on data visualization and data storytelling between Russell Goldenberg of The Pudding, Renee Lightner of Viacom, and Alex Simoes of Datawheel. It was an interesting set of perspectives, from Goldenberg and Lightner who focus on creative, one-off pieces involving custom code, to Simoes whose organization focused more on automation and a platform approach to data storytelling.

Simoes talked about how today, looking for data on sites like data.gov is like shopping in the warehouse section of IKEA. As a user, you need to open up the data to see what’s in it. Datawheel’s goal is to build a data showcase, so that you can see what you can find. Instead of the paradigm “data as files”, the paradigm becomes “data as stories.” Simoes talked about his platform approach to visualizing data, where his company is trying to put themselves out of business by increasingly automating the tasks they are doing to enable scaling up.

For Lightner, it is important to think of data stories as experiences. Words are not enough — people want to be immersed in the data. It’s important to be original and have good aesthetics. Data storytelling on its own is hard, but to make an impact it needs to have an original “spin” on it—maybe a new way of visualizing the data.

Goldenberg mentioned that great data visualizations are those where people can see trends on a macro level, but can also dive into the data to find their own conclusions as well. In terms of his process, he tries to get into prototyping in code as quickly as possible, to start answering real questions and not to get excited about something that can’t be manifested. Having multiple options in the design phase is necessary, to leave your options open as you iterate in case the data doesn’t bear them out.

Facts, Truth and Dataviz

Finally, Scott Berinato of the Harvard Business Review and the author of Good Charts spoke about that there is no such a thing as an objective data visualization. Every chart is a manipulation. Every chart is a collection of decisions about what to show, and what not to show.

Different visualizations of the same data can show different “truths.” Visualizations automatically cause the viewer to begin looking for stories. However when it comes to visualizations, we always see a meaning that is different from the data. The facts are not the whole truth — they are only part of the truth.

What you show in a graph can significantly change the story. There are charts that have “bad intentions” (and good or bad execution). There are also charts that have “good intentions” (and good or bad execution). Even though a visualization may represent the facts, it doesn’t necessarily show what is “true.”

People may not be able to convey the data points after looking at a chart, but they can remember how they felt when they saw it. First we feel (e.g. is it red, big, steep), then we try to relate (“have I seen this before?”), then we think (“what does it mean”).

In the end, what we need is data visual literacy. This entails three things: A comfort level with numbers and statistics, being able to gauge one’s own emotional response to a chart, and building confidence at reading charts.

Christian Marc Schmidt is Founder and Principal of Schema, a research and design firm based in Seattle.