What follows is an unedited guest post by the author of Applied Divinity Studies about Substack. While I don't necessarily endorse everything said in the text that follows, I thought it was an interesting post and the author asked me to guest-post it here.

Some thoughts of my own as a prologue:

Looking at things I read in my RSS I notice first that indeed I look at my RSS reader (Feedly), and not at my mailbox. RSS is a civilized way of following updates from disparate sites you like, aggregating them in one central place that's separate from the mailbox where they can be saved to be read later, organized by source if need be. Newsletters —stuff that hits your inbox— strike me as barbaric. We solved the problem of centralization of consumption of diverse sources and then at some point there was some collective forgetting?? Maybe for a business that doesn't have a corporate blog a newsletter (With news about the sender) makes more sense, but blogs-via-RSS do not find a good replacement in the form of spammy inbox periodicals. One argument for newsletters is that it allows senders to track metrics about who reads their work, how much they read, etc. In Nintil for example while I have some Google Analytics metrics, I have zero information if you read it via RSS and not open the page; but I don't care, I don't even know off the top of my head how many people read Nintil every month; the reason I have Analytics is to tell me from where my blog is being cited, so I can follow the discussion about what I have written elsewhere. Nintil is not written for an audience I carefully cultivate. If you like Nintil you are free enjoy it (no paywalls here) and if you don't you can read something else. Because of that, I don't have a pressure to keep writing blogposts all the time, or feel the urge to squeeze a post out of the random daily occurrence just to be able to push the daily post.

But not every publication is a vehicle of self-expression. Monetized newsletters are there to make money. And that's fine. But paid content seems to fall into one of a few buckets: either politics (hooliganism ensures you'll pay) or businessy/financey pieces (Because reading may give you an edge somehow, or so you think). Who would pay (regularly!) for Nintil? If you like Nintil, you can donate via Patreon. Nintil is also the production of public goods (Like my Longevity FAQ; there I offer you months of research, for free, open to everyone), which is also incompatible with the paid model.

Do I read any substacks? I use kill the newsletter to convert newsletters into well behaved RSS feeds, and perusing my stash of feeds I can see Nadia Eghbal's and Anton Howes'. This will be no surprise given what I said above: My interests do not lie in operating businesses or following current events ¹ , and that seems to be the bulk of what the Substacks are about. Not that I have tried particularly hard to scan their repertoire for something that I may like more.

So by no means I can be a qualified critic of Substack; first because I am not that well aware of the various 'stacks, second because they seem to target an audience I am not part of. It would be as if I wrote beer (All beers taste equally awful to me) reviews; the problem would be me, not the various beers.

This all said, there is something to be said for being paid to write and it's good that there exist platforms that allow people interested in publishing under that model. I am not making a Sandelian critique here, I very much am a proponent of markets in everything, but I also want this blog to be fully unpaywalled (I'm not selling you a product, after all), and I wish the same were true of other blogs (or newsletters); moreover I don't think paywalling content will lead to better quality unless the author is working full time on it. Is there a moral case for supererogatory open blogging? Probably.

Now, the guest post. Enjoy.

How Substack Became Milquetoast

The style will often be strange, incorrect, overburdened, and loose, and almost always strong and bold. Writers will be more anxious to work quickly than to perfect details. Short works will be commoner than long books, wit than erudition, imagination than depth. There will be a rude and untutored vigor of thought with great variety and singular fecundity. Authors will strive to astonish more than to please, and to stir passions rather than to charm taste.

Alexis de Tocqueville, Democracy in America (1835)

The mathematics of churn are the best thing for niche artistic interests since the invention of patronage.

Byrne Hobart, Understanding Netflix (2017)

De Tocqueville’s indictment of American style is equal parts exoneration. If things have always been bad, we can hardly blame new media for the decline of society.

And yet, each time a new platform claims to be the exception to the rule, it bears repeating, as many times as it takes, that this is not a good state of affairs.

I’ve rewritten this post three times, each more dogmatic than the last. So I’ve decided to forsake all the flashy rhetoric and just soberly explain the structural mechanics behind Substack’s relentlessly milquetoast taste.

Homogeneity Within Newsletters

Top substack writers all have a clear focus. The Dispatch writes about politics, Matt Stoller writes about monopolies, Bill Bishop covers China.

Isn’t this normal? After all, no writer can be expected to cover every topic. In an age of disaggregation, we should read from experts instead of casual polymaths.

And yet, the apparent normality obscures its historical peculiarity.

Consider SlateStarCodex, one of the most popular blogs in a pre-Substack era. It had posts ranging from political science fiction, to pharmacology, to artificial intelligence, to media economics. That’s not a selective assortment, it’s his last few posts in chronological order.

(Or consider Nintil, which covers everything from longevity, to meta-science, to education and decision theory.)

Heterogeneity works for blogs since you grow through an occasional viral hit, and gain loyal readership through the occasional post that strikes a reader just right. As SlateStarCodex’s author describes his success:

Granting the premises, maybe because I write a lot of stuff? There are about 500 SSC articles, one every couple of days. Even if I'm no better than anyone else, there are a lot more opportunities for people to stumble across the one that they like, be impressed, and start following me.

The problem is not merely homogeneity of topic, but homogeneity of substance. If you have to publish a newsletter every week, you don’t have the room or incentive to take risks.

In financial terms, blog posts have asymmetric returns with capped downside but unlimited upside. If you write a bad post it won’t get shared and no one will see it. If you write a great post and it goes viral, everyone on the internet thinks you’re a genius. Since content is shared organically, your best work gets way more exposure than your worst. The incentive in these situations is to ramp up variance and do the most interesting writing you can muster.

That dynamic is also good for readers. If a piece is bold but wrong, you can just close the tab. If it’s bold but right, you’ve hit a gold mine. Since readers can be massively diversified across authors, they can benefit from increased variance in the underlying assets while insulated from associated risk. In the Blogosphere, Twitterverse and even on Medium, readers could enjoy long tail upside without commensurate cost.

In contrast, every single edition of a newsletter is delivered to every single reader, and since a lot of it is paywalled, there’s little potential virality. This is aggravated by the one way valve on subscribers: one someone churns out, they’re unlikely to give you a second try. So the ensuing incentive is not to take any bold risks, avoid alienating readers, and write whatever will appeal to your current audience.

This might sound true for other newsletters, but not for yours. After all, you’re an interesting person, and you have interesting thoughts. More importantly, your readers are smart people who demand insight!

The problem is that like pop-music, pop-science and all other pop-culture, pop-writing can only appeal to the largest common denominator. Even if the author and readers are interesting people in isolation, aggregation forces this blob of individuals into a milquetoast morass. As David Foster Wallace once described it:

It is of course undeniable that television is an example of Low Art, the sort of art that has to please people in order to get their money ...this is a logical function of its need to attract and please Audience. And I'm not saying television is vulgar and dumb because the people who compose Audience are vulgar and dumb. Television is the way it is simply because people tend to be extremely similar in their vulgar and prurient and dumb interests, and wildly different in their refined and aesthetic and noble interests. It's all about syncretic diversity: neither medium nor Audience is faultable for quality.

Homogeneity Across Newsletters

Still, homogeneity within newsletters might not sound like such a crime. If each person is an expert in their domain, a reader still has the option of reading broadly and getting a taste of each hermetic utopia.

Except that structural forces also ensure a homogeneity across newsletters, ensuring that you never read anything too original.

Global Ranking

Subcultures allow for an escape from status as a zero-sum game. If you’re the king of your own world, you don’t feel bad for being at the bottom of someone else’s social ladder (1, 2). Bloggers know they aren’t universally popular, but it doesn’t matter as long as you have your corner of the internet.

In contrast, Substack totally violates Social Fog of War by instituting a global leader board. The result is, if you’re an up and coming writer, you can see exactly what’s popular, and since humans are apes, you’ll be inclined to imitate it. Sure, it’s just a number, and there’s no real reason you have to make it go up, but as Byrne Hobart (#17 Paid) once said, “the global leaderboard is very addicting”.

This might be fine if the leaderboard was interesting and diverse, but as I’m writing this (September 6) , the top 3 posts are about Donald Trump, Donald Trump and Donald Trump. As I’m doing final edits a month later (October 4), the top 3 posts are again about Donald Trump, Donald Trump and Donald Trump.

The obvious objection is that Twitter also has public follower counts but still manages to host vibrant subcultures.

The difference is that status on Twiter comes from who follows, likes and retweets you, not just how many of them there are. On an unpaid platform, no one cares about pumping their follower count with random strangers and bots, but everyone cares about getting retweeted by the people they personally admire.

Homogeneity of Style

A Substack newsletter can have an About page, but that’s it. There will never be anything like Nadia’s notes which are a regularly updated half-baked stream of consciousness, or Guzey’s list of Tweets or even Nintil’s categories. And they certainly won’t have Gwern’s link preview on hover, or sidenotes.

These aren’t just good features, they’re forms of self-expression, and even more critically, variations on the medium that creates the message, and in the extreme case, media for thinking the unthinkable.

This might not sound like a big deal, who cares if there’s only one font?

But as Nadia once wrote (before she joined Substack):

Maybe it wasn’t likes (or “claps”), but the WSIWYG editor, that destroyed writers on Medium. If you all think the same, you write the same... In other words, did the pressure to santitize our words come from the homogeneity of demand (writing for others’ approval), or the homogeneity of production (writing for the text editor’s approval)?

Limited Sampling

As a reader, you used to be able to sample an endless buffet of diverse and wacky creations. You might read a single post from a hundred different blogs, never stopping long enough to get bogged down in one author’s pattern of thought.

Today, once you subscribe to an author, you’re in for life, or until you sever the relationship entirely. There’s no more intellectual fling or extended flirting, it’s a binary choice.

Organic sharing, growth and virality exposes readers to a wide variety of authors, serving up the best of each one’s writing. On Substack, instead of getting the best 1% of posts from 100 authors, you get 100% from each one. Instead of getting the cream of the crop, you’re left with low fat milk, mostly water.

Homogeneity Across Time

It’s better for authors to think persistently and write occasionally than the other way around.

But on Substack, you’re paid monthly, creating pressure to churn out regular updates. Since it’s impossible to have interesting novel thoughts twice a week every week, this also means writers skew heavily towards summarizing the news, pumping out quick takes, or riffing on whatever they read on Twitter.

This is bad for intellectual biodiversity, but it’s also just bad for quality. After seeing a draft of this post, a blogger known for their longform content wrote:

If you're writing a substack, you can't go on a creative vacation! You can't spend 3 months writing something epic! you have to churn out content week after week after week preferably many times per week.

Since there’s always something to read, increasing output is not intrinsically good. What matters is quality-density, which we can naively model as temporal leverage. In the past, you might have spent 10 hours reading a book that took 4 years to research and write, a 3500x multiple on time! Today, a newsletter that publishes M-F and takes 30 minutes to read only provides a 67x multiple.

Incentivization Against Archives

The biggest risk to Substack isn’t that Gmail changes its algorithm or that readers set up automatic forwarding and share accounts. It’s that years from now, each author will have built up so much content that a reader can pay a 1 month subscription, download the archive, and be set on reading material.

It’s really hard to prevent text scraping, so imagine this as “for $9, you can download every Netflix show ever”, and then imagine the effect on churn-rate.

The only reliable way to avoid the backlog effect is to ensure that posts are time-sensitive, where “time-sensitive” is a nice way of saying they don’t reach for any kind of fundamental truths or deep abstractions.

On a more personal note, this is antithetical to the theses of some of my favorite writers across domains, ranging from Gwern’s Long-Content to Andy Matuschak’s Evergreen Notes and Byrne Hobart’s Ideas that Don’t Mean-Revert.

Gwern, quoting The Clock of the Long Now:

What has been done, thought, written, or spoken is not culture; culture is only that fraction which is remembered.

Andy:

Much of the day-to-day thinking involved in creative work is simply lost, like sand castles in the tide. Ephemerality can actually be useful in low-fidelity thought, but it’s simply an accidental property in many cases. We should do our serious thinking in the form of Evergreen notes so that the thinking accumulates.

Byrne:

The world is full of phenomena that move in cycles — countries rise and fall, economies speed up and slow down, ideas come in and out of fashion, people dig new trenches in the culture wars. And then there are times that the world changes, and it’s never the same again….

There are a bunch of topics that I avoid writing about. My rough rule is that I’d like to write stuff that will still be worth reading in five years, and ideally stuff that will be more relevant a year from now. Because of the way news site algorithms currently work, that’s the opposite of what everybody who writes for a living does.

Substack as Platform, Substack as Aggregator

Despite my misgivings, it’s pretty clear that Substack isn’t going away. The internet had the last 20 years to figure out sustainable monetization for blogs, and largely failed. Like most digital and creative work, blogging has zero cost of distribution, which means that revenue follows a power law. Outside of a few tremendously popular writers, it’s not possible to fund yourself through ads, grants and patronage.

Instead, I can only hope Substack will improve, and I’m hopeful they will.

Moral and aesthetic arguments aside, the problems I’ve laid out are a financial imperative.

I said earlier that the biggest risk to Substack is archives, but that’s only true in the existential sense. With regards to future growth, the biggest risk is that the mainstream population continues to read mainstream journals, and Substack never crosses the chasm.

The second biggest risk is that at some point, all the biggest authors will leave. It costs maybe $50k to pay a freelancer to duct tape Twilio SendGrid to Stripe Checkout, and Substack takes 10%, so at around $500k in revenue, authors will start migrating away (actually much earlier since it’s a function of all future expected revenue). Since popularity in creative work has a power law, revenue skimmed off the top represents the bulk of all possible revenue. This hasn’t happened yet because migrations are always a massive pain, but $500k is still the right ballpark.

In the language of Ben Thompson, Substack has two choices. It can become a full fledged aggregator, build network effects and community, personalized content and so forth, but risk the moral purity of being one of the last ad-free algorithm-free corners of the internet, or it can become a platform, provide valuable infrastructure and flexible pricing for writers at scale, and operate as a cheap service for small writers while benefiting from the upside of the huge successes.

These are both great options, the problem is that they’re mutually exclusive.

The aggregator world where writers participate in a shared interlinked reality is the world of Twitter, and thus a violation of Substack’s founding principles.

But the platform world they currently live in consists of milquetoast hot takes churned out day after day after day. It’s not good content. There’s no editorial accountability. It will not produce the next great American novel, or any serious investigative reporting, or even a memorable essay.

As Jonah Goldberg (The Dispatch, #1 Paid) describes it: I’m in a mood, and in a “news”letter such as this, written on the fly, mood is often the only muse you’ve got.

Epilogue: The Dogmatic Bits

Okay, so here’s what I wrote initially, and what I would have led with if I wasn’t a coward.

Substack may be printing money, but it’s totally out of ideas.

They tell the founding mythology as: “Chris, who was taking time off after leaving Kik, had built a prototype in the spare room of his Kitchener, Ontario, apartment, by lashing together Stripe, an enterprise email delivery tool, and some quickly-built web publishing software.”

It would be a prototypically cool and scrappy founding story, except that 2 (now 3) years later, that’s still all Substack is.

They’ve ventured into fellowships and legal support, but it’s difficult to get through the claim “to accelerate the advent of this new media economy, we’ll build increasingly powerful tools to help readers” without feeling cynical. It’s been a year. They have ~~44~~17 employees. [Edit 29Nov 2020: I(Divinity Studies) was confused by their Linkedin] They raised $15,000,000. Name literally one “powerful tool” built for readers.

(The candidates are Import from Twitter, and a poorly designed share modal.)

Or in the worlds of Packy (#15 Free) in response to an earlier draft, “their product velocity is dog shit… don’t do anything for discovery… it crashes all the time… It absolutely blows my mind that they’ve raised as much as they have and have improved the product as little as they have.”

Now obviously, there is nothing about Substack that poisons words as soon as they’re written in its editor. Unlike TV, there is not a barrier to entry, only a barrier to fame. So if you need a convenient way to publish your writing, then by all means use Substack, but if you’re looking for good writing, or for the savior of good writing and deliverer of financial sustainability for good writers, keep looking.

The most damning thing about Substack is not any of these theoretical structural mechanics, it’s the easier more intuitive understanding that nothing great will be written here. Each piece we read and publish is a bite sized dose of momentary stimulation. It follows an unwritten contract between each party–I will not try too hard to writing anything serious, you will not try too hard to understand my writing, and both of us will be happier for it–that feels less like patronage, and more like a cheap imitation of actual craft, something that fulfils its surface level goals but goes no further.

At the risk of veering into the extended metaphor of insight as porn (at least the latter does actually have infinite variety), or the worse and more general crank-dom lamenting the death of my culture and its replacement with the horrid youth culture, it is still worth speaking the obvious truth that having someone who is not your spouse feed their thoughts to you 5-days-a-week, thoughts that they themselves have only had a day to work on, thoughts which would likely go refined or unexpressed in a publication with longer-time horizons, is probably not good for your brain. That’s not a “weekly update”, it’s a parasocial relationship, it’s the same poison you get from every other one-sided social media platform.

Now take a step back, and imagine you’re a book written in 1920. You know that some books become classics while others have a mere flash-in-the-pan success. Naturally, you’d like to know where you fit on the spectrum. But note that survival in this sense is closer to power-law than normal, only a tiny minority of books can endure, least the canon bloat endlessly.

So with each passing day, rather approaching an impending expiration date, you approach the possibility of longevity. Conditional on being alive, it becomes increasingly likely that you are part of the privileged cohort of immortal classics. Paradoxically, the longer you live, the longer you can expect to continue living. This is the Lindy Effect, a kind of near-magical aging-in-reverse enabled by survivorship bias.

This is an unusual, but valid, application of the Doomsday argument. Given no other knowledge, each observed instance is assumed to be sampled from the median, so any writing you come across is around the halfway point of its lifespan.

Back in your body now, you’re a reader once more. Your favorite author publishes a post at 9:15am that arrives in your inbox minutes later. You skim it over lunch at 1pm, delete the email, and then at 4:45pm that same day, it breathes its last breath.

If this all sounds overly theoretical and not at all applicable to your lived experience, then fine. But how often have you gone back to read an old edition of your favorite newsletter? Why bother when you’ll have a new one tomorrow? Has anything you’ve read recently stuck with you? And if so, has it stuck in your heart like your first kiss, or merely to your tongue, not poor or foul, spicy or sweet, but bland, boring, milquetoast.