DataDownload: Change is hard.

8 min readJul 25, 2020

DataDownload: Change is hard. A weekly summary of all things Media, Data, Emerging Tech View this email in your browser

Today — the world of media is changing. We’re looking deeply into how social platforms treat people of color differently. Twitter is taking QAnon offline. And, at the same time, Disney is using AI to identify animated characters — with no shortage of interesting outcomes. The New York Times is in the news… both for adopting first-party ad data, and for the announcement that its new CEO is Meredith Kopit Levien, who is currently COO. Congratulations Meredith- a well deserved and timely promotion.

Finally, a word about what it takes to ride the rollercoaster that certainly lies ahead. We like this TED Talk about the powerful impact of grit. Or, should we say GRIT. The power of passion and perseverance. I like the word, and I like Angela Lee Duckworth.

Media isn’t changing because it’s fun, or easy, or simple. We’re changing because we must.

Change is hard. Change is good. Change is necessary.

Tell us your change story. We’d love to learn and share what you’re doing to build the future.

Steve (steve@nycmedialab.org)

Steven Rosenbaum
Managing Director
The NYC Media Lab
Steve@NYCMediaLab.org Must-Read Black Instagram Users Were 50% More Likely Than White Users to Have Their Accounts Automatically Disabled, Internal Research Reportedly Showed

Current and former Facebook employees have told Business Insider that an internal 2019 study found that an automatic moderation algorithm was 50% more likely to auto-ban Black Instagram users than white users. When researchers informed higher-ups, they were told to stop the research and not to discuss the findings with anyone.

To further fuel the flames, Instagram reportedly used a different moderation algorithm but informed staff they could not research potential racial bias. Facebook didn’t deny things, but said the researchers’ methodology was flawed. Speaking with Business Insider, one frustrated engineer said that “I’ve seen people be driven insane as leadership ignores them or outright shuts them down and commits us, again and again, to doubling down on this same path.”

2 min read

Over the years, Twitter has warped and shifted its views on free speech. Where once it referred to itself (probably jokingly) as the “free speech wing of the free speech party,” the platform is now seriously considering the costs of such a stance.

On Tuesday, the company announced it was cracking down on QAnon, “a sprawling online community loosely organized around the belief that Donald Trump is waging a secret war against an elite cabal that engages in devil worship and pedophilia.” The conspiracy theory/cult has led to COVID-19 misinformation and coordinated harassment — and worse.

Joan Donovan, research director at Harvard’s Shorenstein Center, said this is “an important marker that Twitter is recognizing how it is being manipulated.” The company has already removed 7k accounts around QAnon — including prominent figures in the “movement”. It won’t make them dissapear, but “Donovan predicted it would dramatically reduce the group’s ability to spread.”

5 min read

Read More Tech+Media Disney’s New AI Is Facial Recognition for Animation Disney researchers have developed the Content Genome, which populates knowledge graphs with Disney content metadata — sort of like what you see in Google when you search for prominent figures or companies. From there, AI applications can use this data to search for characters, scenes, or objects in Disney’s vast archive. The Engadget piece dives into the team’s creation of a robust taxonomy and automated tagging pipeline for the Genome, and the use of deep learning to detect non-human cartoon faces — fascinating stuff.

“‘So if an animator working on a new season of Clone Wars wants to find a specific type of explosion that happened three seasons ago or as a reference to make something for this current season, that person had to spend hours on YouTube going through video because you can’t find that by just looking at episode titles.’ But with the help of this platform, the animator will be able to simply search for the requisite metadata.”

6 min read Read More Inside the New York Times’ First-Party Data Play NY Times Senior director of ad platforms Sasha Heroy sat down with Digiday to expand on the Times’ move to first-party data for ad targeting. The publication has already built 45 proprietary audience segments, split into a handful of categories: age, income, business, demographics, and interest. Here’s what Heroy had to say on building up a first-party audience business:

“Our first hurdle was proving we could invite readers to share information that would be useful to our advertisers, showing in a transparent way how we would use that data. In the first proof-of-concept, many more readers than anticipated were willing to share information as a truth-set for how we use these audience models. Having 6M subscribers and many registered users is valuable for a first-party audience business, but an enormous amount of data is used to inform models that are gained from repeat visits and reading habits from registered and anonymous users in the model.”

5 min read Read More “Look at the Robot as Your New Colleague”: What Automation Can Do for Sports Reporting

United Robots has generated over 1M news articles for its clients since 2015. Using structured data, like real estate sales or traffic incidents, the company automatically generates content. To understand why readers turn to certain articles for sports coverage, United Robots CPO Henning Johannesson studied the 1k most-read sports pieces over the course of a year, finding these were breaking news and top league team stories. Low engagement match reports — that’s something AI can do, Johannesson says.

“Go and write about the team’s equipment manager who has worked with the team for 50 years, who has repaired boards and football boots. Do that story instead of writing about the dull game that automation can do.”

3 min read

Read More What We’re Watching Grit: The Power of Passion and Perseverance

Angela Lee Duckworth left a high-paying job in consulting to teach math to seventh graders in an NY public school. “She quickly realized that IQ wasn’t the only thing separating the successful students from those who struggled. Here, she explains her theory of ‘grit’ as a predictor of success.”

6 min watch

Watch Now What We’re Listening To Podcast: Antitrust 3: Big Tech

NPR’s Planet Money podcast asks, “what does the rise of dominant tech companies say about competition and the state of antitrust law?”

Listen Now Virtual Events Virtual Event: Machines+Media — The Disinformation Industrial Complex
Date: July 28, 11AM
From disinformation and privacy to health side effects and anti-trust concerns, the social media giants face big questions ahead. How can we understand this emerging, shadowy industry in an election year? And what is the right role for government in solving these problems? What are the priorities and outlook for change in the US and abroad? Register Here.

Virtual Event: Jorge Camoes — Dataviz Lessons From the Pandemic
Date: July 28, 6PM-7:30PM
Jorge Camoes argues that the pandemic forced us to take a look behind the scenes, and realized the importance of data cleaning, clear concepts, statistical and subject data expertise, and find ways to communicate a complex reality. Register Here. A Deeper Look How to Understand COVID-19 Numbers

If you’ve been following the COVID-19 widgets on the NY Times’ front page, you already know the hoped-for lull in cases during summer hasn’t arrived. In fact, things seem to have been getting worse. And with this influx, there have been debates, arguments, and debunking around the numbers: it seems data is being used out of context by both the masses and national leaders.

How bad are things, really? ProPublica’s deep-dive into coronavirus numbers explores common metrics and how to interpret them. Here’s what we’ve learned:

To understand what’s going on in your state: look at the number of cases, case positivity rates, and number of deaths.
Positivity rates measure “the percentage of total tests conducted that are coming back positive.”