Aug 24, 2021

Axios Login

Isn't it weird how all parents think their kid is the loudest? I mean, weird because all the other parents are just wrong.

Today's newsletter is 1,417 words, a 5-minute read.

1 big thing: The limits to Facebook's transparency

Illustration: Aïda Amer/Axios

Facebook is finally sharing information about which posts on its platform get the most views, but researchers tell Axios the new quarterly reports lack the details or timeliness to be of much use.

Why it matters: While Facebook has offered its "Widely Viewed Content Report" as a transparency move, the company won't persuade critics that it's winning the fight against misinformation unless it gives outsiders more of an open window onto near-real-time data.

Catch up quick: Facebook's release last week of the new report marked its first ever attempt to quantify a kind of "greatest hits" of the News Feed for a particular time period.

  • The New York Times reported days later that Facebook had prepared, but not released, a similar report for the first quarter. Facebook later acknowledged and shared the list, which included a mainstream news article popular with the anti-vaccine crowd.

Between the lines: While the first quarter's release might have been less flattering to Facebook than the second one, researchers say the real issue is the reports don't reveal much.

  • Accountable Tech's Jesse Lehrich called them "performative and self-serving."
  • "They are constantly finding new ways to slice and dice data to advance their preferred narratives about their products," Lehrich told Axios. "But quarterly reports showing YouTube and Amazon were the most-viewed domains do nothing to help us understand the fast-changing threat landscapes on vaccine disinformation, political extremism, or anything else meaningful."

The big picture: Lehrich said that what researchers really need is true transparency — that is, broad access to real-time data.

  • While Facebook offers limited data to a few hundred researchers via special programs, far more data is publicly available from Twitter.
  • "There is a reason everyone focuses on Twitter when trying to identify bots," Media Matters for America president Angelo Carusone told Axios. "It's really easy to do that. You can't do that on Facebook." (In fairness, it is also easier to create bot accounts on Twitter, too.)

Carusone said his wish list for data from Facebook begins with information on how often specific articles and sites are being shown to users.

  • Facebook-owned CrowdTangle offers data on engagement, he noted, but the company has been reluctant to offer insight on how those metrics correlate to reach.

He said the company could also share data on how certain content or groups are getting their traffic — from recommendations, through Facebook's own peer-to-peer messaging systems or from activity off of Facebook.

  • That's what researchers need to untangle whether an unfolding debate is emerging organically or fanned by inauthentic behavior, he added.

The other side: Twitter is a (nearly) all-public platform, while Facebook offers users a promise of considerable privacy, even if it has a checkered record of keeping that promise. (Remember that $5 billion penalty from the Federal Trade Commission?)

The bottom line: Facebook can fairly limit public access to some kinds of data in the name of protecting users' privacy. But it could also provide tons more transparency into content metrics without putting anyone's personal data at risk.

Go deeper:

2. Substack scoops up social app Cocoon's team

Newsletter startup Substack has scooped up the team behind Cocoon, a subscription-based social network for connecting with close friends and family, Axios' Kia Kokalitcheva reports.

Why it matters: As newsletters grow along with the wider "creator economy," companies like Substack are looking for ways to help subscribers feel more like a community and interact with one another.

  • According to Cocoon co-founder and CEO Sachin Monga, his team will work on "all the different types of connections between readers and writers," going beyond the simple bond created by a reader signing up for an email newsletter.

Background: Founded in 2019 by Monga and Alex Cornell, two ex-Facebook employees, Cocoon is one of a number of startups that have tried to build a "smaller Facebook" focused more on close relationships than on other content.

  • While previous attempts — Path is the most famous — failed to challenge the biggest social networks, Cocoon's team bet on two new trends: consumers' willingness to pay for subscriptions and their dissatisfaction with Facebook and its big rivals.
  • "Back when Path launched, Facebook was still a pretty good place to be for staying connected to friends and family," said Monga.
  • As for his former employer's flagship app, Monga said he realized "how many things the app is simultaneously trying to solve for — if you're trying to do everything, you're not doing anything really well."

Cocoon also believed that subscriptions would have to be the business model for a social network where users have a small circle of friends.

  • "15 to 20 people are never going to create enough content to build, to aggregate enough attention to build an ads business," said Monga.
  • That's also why Cocoon's team was more interested in how often users opened the app per day and whether they remained users than in how much time they spent on the app.

Details: While Cocoon's team is joining Substack, its app is not. The startup is working on finding it a new home, although Monga declined to share more details.

Go deeper: Small social networks are having a moment

3. Twitch streamers revolt

Photo: Lionel Bonaventure/AFP via Getty Images

Streamers plan to hold a one-day blackout on Twitch to raise awareness of hateful practices on the Amazon-owned video platform, Axios' Megan Farokhmanesh reports.

Why it matters: The movement, #ADayOffTwitch, is an attempt to draw attention and support for streamers facing — and fighting against — hate raids.

  • During such raids, malicious users will swarm marginalized streamers with racist and hateful language and symbols.
  • A group of streamers, including ShineyPen, RekItRaven and LuciaEverblack, are encouraging others to not stream on Sept. 1.
  • "I think it's important to come together in a display of solidarity with those who have been affected by these hate raids," Raven told Axios.
4. Meet the AI ad man

Marketing and advertising companies are increasingly using AI models to track trends and generate slogans, Axios Future's Bryan Walsh reports.

The big picture: Marketers and advertisers focus on two things: identifying and predicting trends that indicate what consumers want, and shaping messages that will appeal to them.

  • AI — and especially new language models that can be tuned to produce messages far quicker than even the slickest Mad man copywriter — are ideal for both.

By the numbers: The global market for AI in advertising and marketing is valued at more than $12 billion, and is projected to reach $107 billion by 2028, according to a recent report.

  • Research by Gartner predicts that by 2022, more than 30% of digital content will be created with the help of AI.

How it works: AI models are particularly good at drawing in vast amounts of data and identifying connections and correlations, which makes them excellent at instant trendspotting, says Daniel Anstandig, CEO of the enterprise tech platform Futuri.

  • Futuri's AI application Topic Calls takes real-time data from Facebook, Twitter, Instagram and 100,000 new sources, and looks at early signals to determine what will be popular in a particular market or demographic up to 24 hours from the present.
  • A newsroom or a brand content marketer "could look at a specific audience and determine where their investment of time and energy in terms of content generation is going to be best used," says Anstandig.

Between the lines: AI can increasingly help with generating that content as well.

  • Neuroflash, a Hamburg-based marketing company, uses natural language processing AI to generate scores of suggested slogans, email subject lines and social media language.
  • Its AI is then able to rank each suggestion on a zero–100 scale, based off historical data about customer preferences and calculating how different messages will resonate with different consumers.
  • Big language models "are very good at generating possibilities," says Henrik Roth, a co-founder of neuroflash. "And then humans can bring an editorial quality to pick the best one," taking into account how the AI itself ranks suggestions.
5. Take Note

On Tap

Trading Places

  • US Digital Response, which helps infuse government agencies with needed tech know-how, named Jessica Cole as interim CEO, with founding CEO Raylene Yung headed over to the Biden administration to work on tech modernization efforts at the General Services Administration starting next month.


6. After you Login

OK, this kid might actually be louder.