(SEMI-)CROSSPOST/HOISTED: MATTHEW YGLESIAS (2022): What I Learned Co-Founding Vox
Yglesias's postmortem of Vox shows how distribution economics shapes what journalism can be. Paying salaries required income. Ad-funding required scale. Scale required kowtowing to FaceBook’s and Google’s algorithms. That made for enormous pressures to back off from true data and explainer journalism back toward clickbait—in Vox’s case, a very mild form of clickbait, but still it was what the eyeballs that it could harvest and sell to advertisers wanted to see. It was not what the eyeballs it attracted needed to see…
Back in the Second Age of Weblogging, Melissa Bell, Ezra Klein, and Matt Yglesias decided to try to change the world and make some money for themselves and funders by disrupting journalism.
They decided to try starting an organization that would do explainer journalism for real, and focus on informing its readers rather than the myriad of missions existing journalistic organizations undertook, many of which were at best orthogonal to the idea that your customers were your readers and viewers and that your job was to help them learn true things about the world.
By and large, it went moderately well. <http://vox.com/> is still around, still paying its bills, and if it is now much more simply standard journalism for the decade of the 2020s, this is 3/4 the system wearing down its idealistic and public-service commitments to informative explainer journalism, and 1/4 its dragging others along with it to its point of view.
I wanted to hoist/crosspost and then comment on this from Matt Yglesias nearly four years ago, but then I found that the entire thing is (still?) behind the paywall:
<https://www.slowboring.com/p/what-i-learned-co-founding-vox>
What to do, what to do? Well, I feel no compunction about an extensive quote giving away the stance Matt takes here:
Matt Yglesias (2022): What I Learned Co-Founding Vox <https://www.slowboring.com/p/what-i-learned-co-founding-vox>: ‘It’s a subject that’s been rattling around my head for a long time…. I recently gained some clarity when I mentioned offhandedly to a younger journalist that I thought I’d failed with Vox and she was surprised because Vox is still there, running articles and putting out podcasts and employing lots of people…. Vox… are producing a lot of great stuff like Rachel Cohen on abortion rights referenda and education in the midterms and Li Zhou breaking down the actual content of the congressional marriage equality bill. Quitting a good job that I liked at Slate to launch a new website was a ton of hard work and a huge pain in the ass! I only did it because I had insanely high ambitions, and I think Ezra and Melissa felt the same way — the current state of the site is a failure only relative to these ambitions….
The folks who run Vox Media, Jim Bankoff and Pam Wasserstein, are smarter about business than I’ll ever be, and they have what I think is a reasonable plan….So when I say that I think that we failed, I’m not dumping on the current Vox management team or editorial staff, who are doing a good job executing a reasonable strategy — it’s just not the strategy we set off with. There’s no innovative editorial project along the lines of our long-abandoned cardstacks. They don’t really do explainers anymore. You’d be hard-pressed to describe what differentiates a Vox article from an Atlantic article or a Slate article… an incredible flattening of the content landscape, and Vox has become part of that flattening rather than an alternative to it.
This post is my effort to unpack how we fell short of those aspirations…
As to the rest of it, let me attempt to not infringe on Matt’s IP—of the 17 articles on <[slowboring.com](http://slowboring.com)> so far this October, only three are not paywalled, and the paywalls are typically ferocious. But I will summarize, extensively:
The principal lesson Matt says he learned is this: A good idea plus a great team is not enough. A business also needs people it can actually hire, bosses who can effectively coördinate, an awareness of how the competition is changing, an ability to find a lane, and a revenue model that fits the product. The secondary lesson Matt says he learned is this: Running a business stacks many medium-difficulty problems on top of each other, and intellectuals tend to massively underrate the skills of the coördinating and organizing people who solve them. The tertiary lesson Matt says he learned is this: While hiring is hard, training is even harder, for there was no recipe, no secret sauce, to turn new hires into Ezra Klein–style writers.
The principal argument Matt makes is that Vox failed relative to its founders’ aspirations because:
The kind of talent they needed didn’t scale and was hard to find.
Their business depended on ad-driven platforms.
The New York Times adapted rather than let itself be disrupted.
Vox was driving hard to achieve serious scale and support itself via a web-advertising model. But advertisers demand huge scale since each viewer’s eyeballs are worth very little. In the mid-2010s, huge scale meant traffic from FaceBook and later Google
Thus in order to stay in business, Vox had to recognize that its readers were not its customers, its advertisers filtered through FaceBook and Google were.
That meant bending to their technical, editorial, and above all algorithmic business preferences. For example, Vox’s “cardstack” explainer format couldn’t work on Facebook. And what went viral on FaceBook was a mix of identity politics and denunciations of the 1%, and that then shaped what got written. Vox therefore flattened into the same sameness as everyone else, as people noted that saying things were not simple did not get eyeballs.
Product innovation therefore became impossible.
The scarce talent that could have set Vox apart was hired away by the Times, which had a subscription-revenue base, went all-in on it, and thus found a business model in which each reader was potentially worth more.
Yes, Matt may be self-serving here: Talent shortage, platforms, and NYT competence are all external or structural causes, and the founders’ own strategic choices get little scrutiny.
The inability to create a pipeline to spread Ezra Klein’s expertise is also worth much more scrutiny: Many young journalists in the mid-2010s wanted to acquire the expertise of Ezra Klein and were willing to work very hard to figure out how to do that. Hell, many young and old journalists today want to acquire the expertise of Ezra Klein and are willing to work very hard to figure out how to do that. We have a society that is superb at turning knowledge of how to do things into recipes that can be taught, into tekhne, and, yes, on top of tekhne there rides what Michael Polanyi calls “tacit knowledge”, what James Scott calls metis, what Sean Illing calls phronesis—not just doing the thing, but doing the thing in the time, place, and way that makes it useful. And pretty much the only way to learn that is extensive apprenticeship, working and assisting at the elbow of someone who has it.
But I think that the fundamental point is that there was an audience for explainer journalism, but not an audience that could be harvested for ads at scale in the malign social-media internet of the day, with its ragebait, its clickbait, and its shouting class.
OK. So, then, where are we now? What does this mean for me, as I try to make my fall push to figure out what to do with this SubStack thing? Especially in the onrushing world of “AI”, which can combine and remix ideas both assisting me and replacing me in a way that was impossible two years ago, and in which a good deal of what I have done has been playing the glass-bead game in a way that machines are now learning to do, in some ways better than I can?
Readers-versus-customers is easy, thanks to SubStack. SubStack really does solve the alignment problem. <http://vox.com/>’s finding that to pay its bills it had to align with the FaceBook and Google algorithms was the biggest blockage to Vox’s success. I only have to write what the low six figures of people find worth their scarce attention and, for one in forty, their money. That is a much lower bar to clear in terms of scale, and a much higher bar to clear in terms of quality. It is also, for me, I think, the right bar.
But note that the platform-dependency problem has not vanished. Google Zero—Google driving zero traffic to anyone else’s website—is coming. The Information Superhighway’s on-ramps are being turned into toll booths that answer the question themselves, by remixing what the rest of us wrote. Being discoverable is no longer the asset. Being trusted is. Whatever value I add does not come from producing grammatical paragraphs about the Smithian division of labor, the Polanyian double movement, or the post-1870 long century. Machines now produce those by the bushel. My value lies in knowing which of a thousand such paragraphs is actually true and relevant this week.
And this brings me to the Hermann Hesse novel, The Glass-Bead Game. In Hesse’s novel, the Castalians spend their lives creating ever-more-elegant syntheses. A theme from Bach is linked to a theorem of Euclid, which is linked to a passage from the I Ching. All of it is gorgeous, all of it is self-referential, strung together like glass beads on a string, and all of it is useful and interesting by virtue of the juxtaposition and the remix. And now there is a machine that has read everything, and that strings beads via juxtaposition and remixing faster than I can, and sometimes more prettily—although I think it has useful and relevant taste only by accident. Not to mention the deep question of whether the glass-bead game is a game worth playing in the first place
So what does this mean for my fall push? Five things, I think:
Fewer beads, more vouching. Less of the “here are fifteen interesting things” firehose. More of “here is the one thing this week that I think is true and important, here is why, and here is what would make me change my mind.”
Show the work. That will be a large part of what readers will increasingly pay for: domain-knowledge expertise visibly hacking through increasing amounts of AI-slop.
Be the apprenticeship. Model phronesis, what of it I have, so that others can apprentice. This newsletter’s highest use may be opening the gate so that readers can watch judgment being exercised, in real time, on live questions.
Choose the questions. Machiavelli wrote that in the evening he put on formal robes and entered the courts of the ancients to ask them questions. I should not simply react to the current thing. I should draw people over to the important questions.
Small is fine. Vox failed by its founders’ aspirations because it needed gargantuan scale. I do not.
Will this work? Is a fall push worth making? I genuinely do not know. I will find out.