Welcome to Galaxy Brain — a newsletter from Charlie Warzel about technology and culture. You can read what this is all about here. If you like what you see, consider forwarding it to a friend or two. You can also click the button below to subscribe. And if you’ve been reading, consider going to the paid version.
Reminder: Galaxy Brain has audio versions of the posts for you to enjoy. I’ve partnered with the audio company, Curio, to have all newsletters available in audio form — and they are free for subscribers.
Back in 2017, I got an email from somebody asking if I wanted a file containing the entire archives of 4chan’s /pol/ message board, part of the toxic troll swamp of the internet. After vetting the offer to make sure it wasn’t some horrible phishing scheme, a colleague scraped tens of thousands of URLs and organized them by how frequently they appeared across the board. I remember being very excited to open the spreadsheet. I expected the worst of the internet, neatly indexed. I thought it would tell me some story of how right wing media wound its way into young, nihilistic audiences, one shitpost at a time.
What we found was surprising — but for different reasons. The most popular domain, by far, was YouTube. I don’t have the file anymore, but I remember that about 60-70% of all the links scraped were YouTube clips. It was hard to find what was being shared, as many of the videos were disabled or taken down. The rest of the top ten was filled with mainstream media domains. CNN was in the top five, as was the New York Times. I know the AP and the Washington Post and Drudge were up there as well. 4chan’s media diet, at the domain level, was not that dissimilar from…mine?
The non-YouTube links were frequently anodyne, straight-news stories or clips from a press conference. A domain level view didn’t matter nearly as much as the post that accompanied it. The 4chan post itself (meaning the anonymous commentary) was where you could find the (often vile) context. Even then, it wasn’t always clear what was going on. I’d often have to run down through the entire thread and see how the content fit into the conversation among the anons. Sometimes it was difficult to see how a piece of news was being weaponized, even in the most feverish of fever swamps.
I was thinking about the /pol/ spreadsheet today as Facebook released it’s “Widely Shared Content” report — a quarterly snapshot of what the social network claims are the most-viewed pages, posts, and domains shared across the platform between April and June. At the domain level, YouTube is #1…just like 4chan!
Jokes aside, the report is very weird. Will Oremus at the Washington Post summed it up nicely, describing Facebook’s list most viewed links as resembling “the news feed as a junk-mail folder”:
It shows, for instance, that the most-viewed link on Facebook in a recent three-month period was to the website of a Wisconsin firm that offers to connect Green Bay Packers fans to former players. The second-most-viewed link was to the online storefront of Pure Hemp, which sells CBD products. The third-most-viewed post in the same period, by Facebook’s reckoning, no longer exists at all. A preview of the post that appears in search engines suggests it was a viral meme encouraging users to disclose personal information to discover their “porn name.”
And then there’s this:
And this, which makes a lot of sense to me:
Why is this list so weird?! UMass Amherst’s Ethan Zuckerman’s offered some ideas, suggesting that the Green Bay Packers fan/player connection group might be buying tons of Facebook ads. Maybe someone’s juicing the posts. Or, he argues, the links pop up in a ‘suggested for you’ box in news feeds of people from Wisconsin (or some similarly huge bucket). Ultimately, Zuckerman argues, this information inspires more questions than answers. And he’s right.
Kevin Roose’s Facebook Top 10 Twitter account and Crowdtangle reporting on the platform’s top-performing U.S. Facebook pages likely prompted this entire report — and he might have also found the answer. Apparently former Green Bay Packer place kicker Chris Jacke, whose Facebook page has 121,008 followers, links to the meet-and-greet site in all of his posts — which tend to see high interaction numbers on the platform.
Roose found the same dynamic at play with the CBD e-commerce links, which seem to have attracted lots of ‘views’ (views very much in quotes here) because former Family Matters star Jaleel White spams them from his page. According to Facebook, it’s possible that the most viewed URLs on the largest platform in the world for a period of time were the result of the spamming tendencies of a nostalgia-powered Facebook page run by Steve Urkel.
This weirdness is enough to make people (myself very much included) wonder if this content list isn’t just obfuscation from Facebook. As Gizmodo’s Shoshana Wodinsky notes, Facebook’s list, “paints a very different picture than the many, many reports from independent researchers and reporters who have found that hyper-partisan content garners the most clicks on Facebook.”
It doesn’t seem outside the realm of possibility that Facebook can slice the data —which only it can see! — in a particularly flattering manner to satisfy its transparency requirements. Most users will be satisfied, if they notice at all. Box, checked.
Of course, as plenty have pointed out today, it’s telling that Facebook chose to highlight its "most viewed" content over "most engaged." Engagement, after all, might give us better understanding of emotional salience, intent to share, etc. You might recall that back in 2018, we all found out Facebook was inflating its video view metrics. One cause of the inflation: the arbitrary way the platform defined a video “view” (three seconds or longer). In the case of Facebook’s “most viewed,” it doesn't tell us anything about where people lingered or clicked. Content with links could represent <13% of views but 60% of time spent in-app (within the in-app browser).
The taxonomy of all these terms is critically important in order to make any sense of this stuff. Without it, the data can only be…interesting. Here’s one example from the report:
Facebook says that “the vast majority of content viewed in News Feed during Q2 2021 (87.1%) did not include a link to a source outside of Facebook. Only about 12.9% of News Feed content views in the US during Q2 2021 were on posts that contain links.” That makes sense, but what if your friend shares a Facebook post from a user and that post has an outside link? Does that count as “outside of Facebook” or not?
Does any of this matter? Well, maybe! As I noted with the 4chan links, context matters a great deal, when it comes to the way that content travels and is consumed. Often, it’s not the outside content that matters as much as what’s surrounding it and who is sharing it. Through this lens, view numbers and percentages might not be all that helpful. As Alexis Madrigal points out, little things can matter on social networks, especially if they come from the right messenger. This is also a reason is why the interested parties should be able to look and see what's happening on the platform in real-time. The quarterly snapshots distort timescales and obscure the context, which is crucial for understanding what on earth is really happening with a piece of content.
Back to the weirdness, though. In one sense, I actually appreciate the peculiarity of the posts in this report, because I think they might actually represent a slice of Facebook that I’ve seen in my reporting. For both BuzzFeed and the New York Times, I’ve had regular people grant me access to their Facebook feeds for extended periods. Each time, I’ve been astounded by the profoundly weirdness of the content. It’s not a reflection on the people who gave me their accounts so much as the prevelance of hyper-viral, non-political content. It’s just…strange as hell! A lot of it is pixelated meme garbage that’s played a game of ‘share/screenshot’ telephone across 40 different platforms. Other bits look like chain email FWDs. I immediately recognized the stuff in this tweet the same sort of Feed Filler garbage inside those FB accounts I perused.
What we are seeing in this report is, in a roundabout way, indicative of a real user experience. People are constantly scrolling past or passively interacting with Feed Filler and content chum that is forgotten as soon as the next pixel hits the user’s retina. But it’s still a marginally meaningful part of the Facebook experience to somebody, somewhere. How could it not be? There are billions of people scrolling constantly. Me, you, the guy down the street are also likely seeing stuff like raccoontv.net links pass by for the weirdest possible reasons — maybe because somebody has an app that’s accidentally inflating the numbers of a piece of content and tricking Facebook into thinking people want to see it. Or maybe it’s because tens of thousands of greedy content jockeys across the world are racing to the bottom to try and game the algorithms and make quick money before the company bans them.
The content chaos that occurs when billions of people and hundreds of billions of pieces of content intersect with recommendation algorithms and real live posts is kind of unfathomable. I’m not saying it’s nefarious (though it may be at times)…just fucking bonkers.
Which is why I found this quote from Facebook spokesman Ryan Peters, speaking with Will Oremus, absolutely hilarious. When Oremus asked him about the strange prominance of the Green Bay Packers meet-and-greet website, Peters replied, “When content from lesser known creators goes viral it isn’t necessarily a bad thing. It shows that anyone, not just established superstars, can reach a wide audience on the platform so long as their content is compelling.”
Peters, FYI, has no clue what the hell is happening at a granular content level on Facebook. Neither do I. Neither do you. Some people at Facebook surely know a lot more about it than any of us (and more than they’re saying publicly…) but they also don’t have any good bird’s eye view of it. When I asked a number of sources inside platform companies about this report today, they all said some variation on this same theme: Facebook is obfuscating. It’s choosing how and what it is presenting in its transparency attempt. But also: many of its classification decisions could be arbitrary and made by a team of engineers on the fly — many of them might have been made before Facebook was in the public eye. As such, there might not be a coherent overarching ideology in terms of how the company is classifying different kinds of content; different metrics were defined one way without much thought about how they’ll influence a report like this one.
In other words: Facebook knows more about its platform than what it’s showing, but also, this weird slice of Facebook we see today might also be representative of how internally disorganized the platform is.
Understanding Facebook is incredibly complicated and likely impossible. This isn’t an attempt to let Facebook off the hook. It’s to say that Facebook is far, far too big.
What you glean about the platform is heavily dependent on the slice of data you’re looking at. Is Facebook a right-wing echo-chamber where Ben Shapiro and Dan Bongino are effectively our news content overlords? Yes?…Probably! Does that matter…probably, yes! Is Facebook full of inexplicable weirdness? Uh huh. Does that matter? I have no idea.
This is not an argument against transparency. I agree with Ethan Zuckerman that “it’s time for us to figure out how — respecting user privacy and research ethics — to get the data we need to understand what’s going on with these platform.” What happens on these platforms matters. Understanding them matters, and it’s a useful endeavor to pursue. But Facebook likely isn’t going to be much help.
What would happen if we did get all the information about how content moves around Facebook. I’m curious what we’d learn. Probably some really scary things, and a lot of profoundly sad things about how people spend their time. There’d likely be some poignant stuff too. Who knows. Would we know what to do with it all? Would we be able to make sense of it? Smarter folks than I will make sense of some of it; others will likely manipulate those findings. Whatever we do learn will be one sliver of the larger picture.
When I asked about the vague, unhelpful nature of the Facebook report, one of the platform employees I spoke with today had a great line. If a similar report had been written about our universe, they said, it would just say: “it’s all empty space.”
I love that. It’s cliché, at this point, to throw out analogies for Facebook. It’s just too big and unruly for any of them to work. But when I think of the process of trying to understand Facebook, it feels a lot like those Cosmos clips where Carl Sagan just keeps zooming out from Earth to highlight the size of the universe and you start to get existential vertigo. At a certain point your brain wants to give up.
That’s how Facebook makes me feel. It’s all empty space and it contains more than I can fathom.
Ok! That’s it for today. If you read this newsletter and value it, consider going to the paid version, and come hang out with us on Sidechannel, the Discord you’ll get access to if you switch over to paid.
If you are a contingent worker or un- or under-employed, just email and I’ll give you a free subscription, no questions asked. If you’d like to underwrite one of those subscriptions, you can donate one here.
If you’re reading this in your inbox, you can find a shareable version online here. You can follow me on Twitter here, and Instagram here. Feel free to comment below — and you can always reach me at firstname.lastname@example.org.