Facebook, for the first time, shares a report on the most widely viewed domains, links, posts, and Pages on Facebook in the US during Q2

A video of two cats walking, tails intertwined, looking like the best of friends. A resource site for Green Bay Packers alumni. Source: Facebook Transparency Center .

Protocol 2021-08-19 Issie Lapowsky

Context & Ripple Effects

Facebook’s release extends an earlier debate over what news links gain reach on the platform, after the company disclosed more data on top-performing U.S. link posts in its earlier link-performance data release. Publishing rankings for domains, links, posts and Pages makes the distribution layer more inspectable for publishers and public-interest critics.

The report also sits in a widening transparency arc: Facebook’s subsequently released Q1 report acknowledged broad reach for content questioning COVID vaccines, while a later Q3 report focused on whether highly viewed posts violated platform rules.

First-order effects

Facebook gives U.S. publishers, Page operators and domain owners a public benchmark for which content and sources reached the broadest audiences in Q2.
Facebook’s Transparency Center becomes a channel for scrutinizing the outcomes of News Feed distribution rather than only the company’s policies.

Second-order effects

Publishers and researchers can compare the newly disclosed rankings with Facebook’s moderation claims, increasing pressure on the company to explain why particular sources and posts receive reach.
The recurring reports create a clearer way to track whether Facebook’s anti-spam and rule-enforcement efforts change the composition of its most-viewed content, as reflected in the later Q3 widely viewed-content report.

Third-order effects

If Facebook continues releasing comparable rankings, content transparency shifts from one-off disclosures toward an accountability mechanism centered on measurable distribution outcomes.
That reporting model raises the strategic importance of ranking systems: platform governance will increasingly be judged by which sources and formats reach users, not solely by what Facebook removes.

The trend: Facebook is moving toward distribution transparency in which public rankings make the practical effects of feed governance easier to evaluate.

Discussion

@andymstone Andy Stone on x
What do people actually see on Facebook? Find out here in our first Widely Viewed Content Report, which will be updated quarterly in our Transparency Center: https://transparency.fb.com/ ... https://twitter.com/...
@alexweprin Alex Weprin on x
According to Facebook the second most popular link on the platform in Q2 2021 was to “https://purehempshop.com/” Yahoo was number 6. https://twitter.com/...
@giladedelman Gilad Edelman on x
How did a humble website offering paid interactions with former Packers players rack up 80 million views on Facebook in three months? Memes. https://www.wired.com/...
@caseynewton Casey Newton on x
There's *some* value in having a quarterly view of the most-viewed content on Facebook, but what we really need is a real-time view. https://transparency.fb.com/ ...
@alicetiara Alice E. Marwick on x
Whenever FB gets called out for suppressing independent academic research, like in the NYU case that just happened last week, they're like BUT WE HAVE CROWDTANGLE
@ognyanova Katherine Ognyanova on x
Transparency is good & added bit of information helps. That said, the level of detail they offer makes the implementation less than ideal. So, what could be done to make this report more useful? (see also @shiringhaffary's related story https://www.vox.com/...) (Thread: 2 of 10)
@alexhern Alex Hern on x
It's hard to view the company's new “Widely Viewed Content Report” as a sincere attempt to help people understand the site with that sort of basic contextualisation missing https://transparency.fb.com/ ...
@profcarroll David Carroll on x
Until there are enforceable regulations there are only public relations. https://twitter.com/...
@daveyalba Davey Alba on x
I asked Facebook if it was planning to add “reach” to the CrowdTangle tool as a metric—letting researchers & reporters investigate FB themselves—instead of pulling out a report in the aggregate every quarter. Here's what I heard from @guyro in response. https://transparency.fb.co…
@digiphile Alex Howard on x
“The cynical side of me reads this report as transparency theatre - a chance for @Facebook to tell critics that they're moving in the direction of transparency without releasing any of the data a researcher would need to answer a question”- @ethanz https://ethanzuckerman.com/...
@swodinsky Shoshana Wodinsky on x
can't believe facebook thinks we're dumb enough to buy facebook's reports about facebook https://gizmodo.com/...
@drewharwell Drew Harwell on x
Finally, Facebook has released what we've all been waiting for: A vague once-a-quarter summation of the 20 most viral posts from four months ago https://transparency.fb.com/ ... https://twitter.com/...
@emilybell Emily Bell on x
Excellent read on ‘transparency theater’ of Facebook's content report by ⁦@EthanZ⁩ - a document that raises more questions than answers https://ethanzuckerman.com/...
@anthony Anthony DeRosa on x
Pretty interesting that Twitter is the 5th most viewed domain on Facebook, with 116.1M content viewers in Q2 2021 https://twitter.com/... https://twitter.com/...
@yaeleisenstat Yael Eisenstat on x
Facebook likes to paint people like me as disgruntled and self-promoting, to try to discredit my critiques. What about a former Facebook VP who spent 11 years at the company? Are his thoughts valid enough for FB and others to pay attention? 👇🏼 https://twitter.com/...
@wired @wired on x
The single most-viewed URL on Facebook over the past three months, with 87.2 million views, belongs to an obscure site devoted to charging people to hang out with former Packers players. https://www.wired.com/...
@alicetiara Alice E. Marwick on x
It also ignores the VERY OBVIOUS FACT that “organic” content (posted by friends and family) can include disinformation.
@timjhogan Tim Hogan on x
Facebook's Transparency Center report opens with this pie chart and it's perfectly representative of how useless the whole thing is. https://transparency.fb.com/ ... https://twitter.com/...
@sarafischer Sara Fischer on x
100%. Quarterly view masks when viral misinfo spikes in reach in real-time. It helps tell the story FB wants to show, which is broadly, this type of bad stuff is tiny compared to everything else, but it hides fact that bad stuff sometimes does have high reach & that's problematic…
@alicetiara Alice E. Marwick on x
So what this shows is that a) FB has no intent of making any data available to researchers that's actually useful and b) they can always argue against researchers by saying we're not using accurate data. LOL THANKS GUYS
@issielapowsky Issie Lapowsky on x
It's hard to see how Facebook argues views “more accurately” depict people's experience of using Facebook. I think most of us forget a lot of what we idly scroll past, but not the stuff we like or comment on. https://www.protocol.com/...
@ethanz Ethan Zuckerman on x
Is a speakers bureau for former Green Bay Packers players really the most popular URL on Facebook in Q2 2021? Or is something really strange going on with FB's transparency efforts: https://ethanzuckerman.com/...
@alicetiara Alice E. Marwick on x
Also, the ill-fated Social Science One dataset, “condor” (my team is working with it right now - it's a HOT MESS), which was another big FB PR effort to make data available, is ALSO just a set of URLs and links!
@ethanz Ethan Zuckerman on x
Stop fucking waiting for the platforms to give us data. They're not going to give us data. We need to get our own data. https://ethanzuckerman.com/...
@jason_kint Jason Kint on x
This is really good. “I am genuinely glad that Facebook is releasing more data about what's popular on their platform. I am also genuinely astonished that this is what Facebook produced for their first effort.” https://twitter.com/...
@bostonjoan Joan Donovan on x
Someone explain to me how “https://www.playeralumniresources dot com” is the most popular site in Facebook newsfeeds? It's a green bay packers web page and somehow manages to her tens of millions of views? Someone over there is a genius! THE ROOSTER BREAKS IT DOWN. https://twitte…
@brandyzadrozny Brandy Zadrozny on x
This Facebook “transparency” report is clearly a stunt meant to clap back at a single reporter who made a list that mildly embarrassed some Facebook execs. It's incredible stuff. Whoever is in charge of this self-dunking effort, just phenomenal job. https://twitter.com/...
@ceciliakang Cecilia Kang on x
great 🧵 reach v. engagement is an important distinction https://twitter.com/...
@guyro Guy Rosen on x
Today we released two transparency reports covering Q2 2021. First, the 10th edition of our Community Standards Enforcement Report. Prevalence of hate speech has decreased for three quarters in a row, primarily driven by progress on AI. https://about.fb.com/... (1/6)
@kevinroose Kevin Roose on x
For example, as @jamisonfoser points out, knowing that https://youtube.com/ is the most widely-viewed domain on FB doesn't tell you what people are actually watching. Could be cat videos, could be QAnon! https://twitter.com/...
@kevinroose Kevin Roose on x
Again, more data is good! And I'm sure people smarter than me will find interesting things in this report. All I can say is that sending a bunch of very smart data scientists on a fishing expedition to disprove a Twitter account doesn't feel like the best use of anyone's time!
@kevinroose Kevin Roose on x
Reach data isn't totally useless. It can help us understand what's being recommended vs. actively sought out. But the way this report is structured (domain-level data, quarterly aggregates, lumping news/entertainment together) makes it hard to know what's actually going on.
@kevinroose Kevin Roose on x
The most basic thing is: despite FB's objections to @FacebooksTop10, there is no inherent reason that reach (how many people see a post in their feeds) is a better measure of popularity than engagement (how many people like/share/comment on it).
@devinkofsky Devin Kurant on x
@Techmeme @issielapowsky Oh now they don't like engagement when every damn metric is based on.. wait for it.. engagement!
@spandi_s Spandi Singh on x
NEW IN THE ALGO & CONTENT TRANSPARENCY SPACE: Today @Facebook launched the 1st edition of its Widely Viewed Content Report (WVCR), with info on the top 20 public domains, links, pages, & posts viewed by U.S. users in the News Feed in the last quarter. https://transparency.fb.com/…
@brendannyhan Brendan Nyhan on x
My August 2020 tweet below on what's potentially misleading about focusing on highest-engagement public page posts on Crowdtangle. FB data shows... -all most-viewed content is .1% of all views (top pages/posts even less) -most views aren't of pages at all https://transparency.fb.…
@willoremus Will Oremus on x
Facebook is launching its own answer to @kevinroose's popular @FacebooksTop10 account. https://twitter.com/...
@benedictevans Benedict Evans on x
Interesting. People keep posting the crowdtangle charts showing ‘bad’ stuff ranked highly, and FB people kept saying 'yes, but that's not what people actually see', but never gave numbers to back that up. https://twitter.com/...
@willoremus Will Oremus on x
@kevinroose Whereas @FacebooksTop10 highlights the U.S> Facebook *pages* whose *link posts* get the most *engagement,* Facebook's first “widely viewed content” report slices the data in many different ways, such as the *domains* with the most *content views,* etc. https://twitter…
@evelyndouek Evelyn Douek on x
My only take so far is it should be “'Widely Viewed Content' Report” not “Widely Viewed Content Report” https://twitter.com/...
@engadget @engadget on x
Facebook's first ‘widely viewed content’ report argues political content isn't actually popular https://www.engadget.com/... https://twitter.com/...
@shiraovide Shira Ovide on x
Welcome to the inaugural “@kevinroose made-Facebook-do-it” report: https://about.fb.com/...
@issielapowsky Issie Lapowsky on x
My girl @birnbaum_e more or less asked as much on their press call. In reply, Guy Rosen said: “There's a few gaps in the data that's been used to date, and the narrative that's emerged is quite simply wrong.” https://www.protocol.com/...
@kevinroose Kevin Roose on x
I'm all for more data, but this new FB “widely viewed content” report (which I think represents the most labor-intensive effort ever made to dunk on me personally?) is just a tremendously weird document. https://about.fb.com/...
@brianboland Brian Boland on x
This is anything but transparancy. A quarterly report on content isn't very meaningful or helpful for the things most concerning to me. I'll spend some time with this today and share more thoughts. First glance is very discouraging. https://about.fb.com/...
@minimaxir Max Woolf on x
According to Facebook, this GIF from Tumblr is one of the most viewed links on all of Facebook in the past quarter, with 49.1M content views. https://transparency.fb.com/ ... https://twitter.com/...
@drewharwell Drew Harwell on x
The no. 3 most-viewed Facebook post for the quarter, marked as “no longer available,” appears to be this viral meme post that asked people to basically share what they'd put in their password-recovery “secret answer” box https://transparency.fb.com/ ... https://twitter.com/...
@kevinroose Kevin Roose on x
For example, #1 on this list of widely viewed links is a site where you can book meet-and-greets with former members of the Green Bay Packers. If you reverse-search the URL, it's all from the account of former place-kicker Chris Jacke, who appears to link to it on every post. htt…
@orientaljanedoe Jane Chung on x
FB really delivering on its crucial mission of making the world more open and connected (these are some of the top posts in the US this quarter, as reported by FB itself) https://transparency.fb.com/ ... https://twitter.com/...
@ariezrawaldman Ari Ezra Waldman on x
This report is NOT Facebook being transparent. It's Facebook cynically *performing* transparency in a way that actually obscures the truth. This isn't an accident. The co is purposely skewing the narrative while making it look like it's just doing what we asked. https://twitter.c…
@caseynewton Casey Newton on x
The No. 9 most common domain you'll find in Facebook's News Feed is ... a speaking agency of former Green Bay Packers players?? (via @EthanZ) https://ethanzuckerman.com/...
@ethanz Ethan Zuckerman on x
I read Facebook's Widely Viewed Content report. It's really strange. https://ethanzuckerman.com/... cc: @kevinroose @BrandyZadrozny @BostonJoan
@minimaxir Max Woolf on x
Despite getting nearly 50M views, it only received 840 notes/likes on Tumblr itself. https://lunamiangel.tumblr.com/ ... https://twitter.com/...
@alexismadrigal Alexis C. Madrigal on x
Also appearing on the list of the most viewed content on Facebook is https://raccoontv.net/ which has not published a new story since June 30 https://twitter.com/...
@rebekahktromble Dr. Rebekah Tromble on x
First of all, the report only covers public posts. Why? 🤷🏻‍♀ ️ FB tells external researchers that it can't give us private post data because, well, user privacy, FTC consent decree, GDPR, etc, etc. But this was internal analysis. So why not look at everything? 3/
@rebekahktromble Dr. Rebekah Tromble on x
Next, notice that the report only covers public posts viewed via users' News Feeds. It doesn't include ads. Nor promoted posts. Nor posts a user views on a Page. Nor in a Group. Nor in the Groups tab. Nor in the News tab. Nor via notifications. Nor via a friend's timeline. 5/
@rebekahktromble Dr. Rebekah Tromble on x
Note that the report doesn't offer any justifications for the types of content FB included in the analysis, nor why they excluded other types of content. And a LOT was excluded. 2/
@rebekahktromble Dr. Rebekah Tromble on x
In other words, this report leaves us w/o any real sense of what it's *actually reporting on*. Despite the title, we know it's not a complete picture of content viewed on FB. But we don't even have the basic descriptive stats needed to assess just *how* (in)complete it is. 7/
@rebekahktromble Dr. Rebekah Tromble on x
I also found Facebook's “Widely Viewed Content” report strange. But for slightly different reasons. Setting aside the weird spam in those top 20 lists for a min, I want to talk about *how* FB reached its conclusions—specifically, what they looked at & what they ignored. 🧵 1/ http…
@rebekahktromble Dr. Rebekah Tromble on x
My guess is that including non-public posts (ie, shared with just friends or within a private group) would change some of the conclusions fairly significantly. But that's just my conjecture. Because FB doesn't even indicate how much of the overall content is in fact public. 4/
@alexismadrigal Alexis C. Madrigal on x
My other thought about it: this is a perfect example of Platform Blindness. It's all about percentages and gives not one thought to the idea that a small Facebook phenomenon could have large implications in the actual, functioning world of American social and political life.
@brianboland Brian Boland on x
Released Quarterly - Unlike data Facebook provides via CrowdTangle this report will be released only once a quarter containing data from the prior quarter. If we look at this first release that data is coming out six weeks after the end of that quarter. /3
@alexismadrigal Alexis C. Madrigal on x
To be a little unkind, Facebook's argument with this report seems to be: Actually, the biggest stuff on our platform is *total* garbage, not right-wing garbage.
@drewharwell Drew Harwell on x
“Instead of presenting Facebook as a hotbed of right-leaning politics, the company's inaugural report presents a far weirder, messier, and spammier picture: the news feed as junk-mail folder” https://www.washingtonpost.com/ ... @WillOremus
@alicetiara Alice E. Marwick on x
OK everyone, strap in, this thing is nutso. It claims that the /sixth most visited link on Facebook in Q2/ was........ yahoo.com
@jesselehrich Jesse Lehrich on x
“It's like ExxonMobil releasing their own study on climate change,” said a former Facebook employee, who spoke on the condition of anonymity due to a nondisparagement clause. “It's something to counter the independent research and media coverage that tells a different story.” htt…
@kevinroose Kevin Roose on x
As @EthanZ points out, something very weird is happening with this FB “widely viewed content” report. https://ethanzuckerman.com/...
@markseibel Mark Seibel on x
Oh, man. I'm worried. The No. 10 most-viewed page on Facebook is the subscription page to the Epoch Times with 44 million views. https://www.washingtonpost.com/ ...
@alicetiara Alice E. Marwick on x
Like half of the report is FB sternly disclaiming that most of FB content isn't links and that most people don't see linked content. Uh, OK. Then what is the point of @crowdtangle? Is @crowdtangle just weird niche data that is mostly irrelevant?
@brianboland Brian Boland on x
Summarized data - Being quarterly the report will summarize the key metrics over the course of 3 months. This is likely to cause more persistent content to rise to the top of the list rather than things that may spike at any given time. /4
@kevinroose Kevin Roose on x
Nice @WillOremus analysis of the oddness of today's big FB content report, featuring a comms quote that I may never recover from. https://www.washingtonpost.com/ ... https://twitter.com/...
@blackamazon @blackamazon on x
...... we gotta start talking to more lay content creators and SEOs https://twitter.com/...
@ethanz Ethan Zuckerman on x
@podehaye @benwagne_r It has a ton to do with what sort of access is envisioned. It's genuinely hard to give access to some data sets, and should be easier to give access to others. Thus far, fines haven't done too much to curb bad behavior from very wealthy platforms.
@markseibel Mark Seibel on x
An obscure page involving the Green Bay Packers is most viewed link? Call me skeptical. https://www.washingtonpost.com/ ...
@brianboland Brian Boland on x
Subset of data - There is so much content on Facebook the top 20 of a list like this only produces a tiny sliver of a very defined set of content. With the scale of Facebook there is a ton more content that will reach people and impact them that will never see this report. /7
@brianboland Brian Boland on x
Aggregated links — As Ethan Zuckerman points out in this article most of the links we are provided are from large internet services that host a broad array of content. The biggest websites are also the biggest domains... of course. /5 https://ethanzuckerman.com/...
@bostonjoan Joan Donovan on x
When EZ starts ranting, we all sit back and listen. (If someone could please make a browser extension that changes the word “data” into random nouns, I'd be very happy.) https://twitter.com/... https://twitter.com/...
@kevinroose Kevin Roose on x
#2 on the list, with 72 million views in Q2, is an affiliate marketing link to a CBD e-commerce site that appears to have been shared exclusively by the actor Jaleel White. https://twitter.com/...
@brianboland Brian Boland on x
Building on my conversations last month with @brianstelter and @kevinroose I put together some thoughts on the new Widely Viewed Content Report from Facebook - a report that fails at it's promised transparency. /1 https://medium.com/...
@geoffreyfowler Geoffrey A. Fowler on x
“It's like ExxonMobil releasing their own study on climate change” Thanks, Facebook! https://www.washingtonpost.com/ ...
@kevinroose Kevin Roose on x
And so on down the list. I have a theory about why these links were deemed “widely viewed,” but it's possible that Facebook accidentally shipped a list of its most prolific affiliate marketing spammers instead of its most viewed domains.
@brianboland Brian Boland on x
I was hopeful that the company would take a step to provide more meaningful transparency to the public but that isn't what is happening here. There is new data being released and that's the good thing. The problem is that the data released is generally useless. Some thoughts: /2
@jason_kint Jason Kint on x
Not sure I've ever seen an anonymous quote from a former Facebook employee labeled with a clear explanation for why they can't speak on the record. Bravo for making this clear, @WillOremus. https://twitter.com/... https://twitter.com/...
@brianboland Brian Boland on x
US Only - For a service that has the vast majority of its users outside of the US releasing a report focused on the US only is highly problematic. While the bottom of the press release says that they will add more countries in the future we really need to know where and when. /6
@willoremus Will Oremus on x
You thought Facebook was all Ben Shapiro and Dan Bongino posts? Wrong, says a new Facebook report. We're actually full of.... whatever this stuff is! (Holds up giant bag of sundry link spam and CBD products) My story on the company's strange new data: https://www.washingtonpost.c…
@podehaye Paul-Olivier Dehaye on x
@EthanZ @benwagne_r What do you think about GDPR-demanding “access” and “portability” of the data to researchers, and fining them to no end if they don't comply?
@brianboland Brian Boland on x
Facebook could commit to making this public data available in a searchable tool like CrowdTangle. Without those commitments and no effort to be more transparent the solution likely needs to be a regulatory or legislative one. We should move in that direction and do so quickly.
@digiphile Alex Howard on x
In the wake of (damning) reporting about internal conflicts regarding transparency, the @Facebook has published an index of its most widely viewed content for the 2nd Quarter of 2021: https://transparency.fb.com/ ... Progress is progress: credit due. Would be good to see this wee…
@swodinsky Shoshana Wodinsky on x
lmao “sorry about the crowdtangle stuff—but this is just as good, right” https://about.fb.com/...
@martinsfp Martin Sfp Bryant on x
Oh interesting. Facebook releases “an overview of the most widely viewed content in News Feed” “The top-viewed news domains account for only 0.31% of all content views in News Feed. Of those outlets, mainstream media dominate by views.” https://about.fb.com/...

Chronicles