Publishing is fucked. The ATmosphere could save it.

Publishing is fucked.

But then, really, you have to ask yourself, what isn’t, these days? As an author staring down the barrel of increasingly enshittified ecosystems, all degrading at the same time, it’s hard to keep positive. I’ve been self-publishing my novels online since May 2016, and I can tell you that what once felt like the front line of an exciting new frontier of democratised publishing, well, it doesn’t feel that way anymore.

Where is the publishing landscape right now? Well, for all the talk of disruption, you still have two main avenues, as a writer, to get your work out into the world — the traditional route and self-publishing.

If you go down the first route, well, good luck to you. Traditional publishing is harder and harder to get into, year by year. Agents take on fewer clients, then compete for fewer and fewer spots amongst a dwindling ecosystem, one that is increasingly risk-averse, and focused primarily on the few remaining book genres that are proven sellers, like books about celebrities purported to be written by celebrities but written by people who might have spent an hour on the phone with them once.

If you do manage to get yourself onto that ladder, you’ll find a world of dwindling advances, contracts that require you to sign away every single aspect of your work that may ever be invented, and the increasing risk that you’re doing a deal with a company that may disappear overnight, taking all of your royalties with it. Oh, and did I mention that as a new author you’ll likely be expected to run your own publicity machine?

Well, that doesn’t sound much fun, so let’s look at self-publishing. Well, Amazon currently controls roughly 70% of all eBook sales in the United States, and that’s of all books. The share of self-published books is way, way higher. at around 95%. That dominance doesn't rest primarily on having the best technology, of course. The Kindle is a fairly rubbish e-reader, increasingly bloated with other, more profitable software. But that’s what most eBook readers have, and they’ll never be able to move away without losing access to their entire library, because, well, enshittification. It’s not just readers, either.

Obviously, as an author, you want to be where the readers are. But you also know that a huge chunk of those readers pay for a Kindle Unlimited subscription, and you really want to reach those readers, too. Reaching them means publishing your eBook exclusively through Amazon, even though you get far less per book through that scheme than you would with a direct sale. Oh, and did I mention that Amazon are making it harder to get visibility all the time, especially for those who don’t choose to use their advertising platform?

Still, it’ll be more money than you can easily make through Kobo, Barnes and Noble, and the like, because reaching readers on each of those platforms means diversifying your approach to suit each one. Advertising, promotions, and mailing list building that’s site-specific, all being done by one person, usually. Who just wants to write.

Goddamnit, I just want to write.

Of course, all of this is then compounded by two things. The first being the creeping enshittification of the entire social media ecosystem that we all use to, you know, sell our books. The second is the proliferation of AI slop, clogging up everything. That, as a writer, is particularly enraging — because when I used that em dash right there, a chunk of people will have assumed that this came out of SuperClippy, and not from the head of an actual human being.

I want you to get mad….

So, things are bad. As the man once said, ‘I don't have to tell you things are bad. Everybody knows things are bad.’ Day after day we stew in a never-ending morass of bad news, bad people, bad vibes. It's easy to get disheartened, and I will admit that that's where my head has been for a few years now, when it come to publishing. But we all know the adage of letting the bastards grind you down. You're not supposed to, right?

So, what could the future look like? One where things aren’t so fucking abject?

For me, the answer could well lie in something called the AT Protocol. I know, catchy title, right? It’s also known as the Atmosphere (or ATmosphere, sometimes), which is also a bit of a silly name, but at least it doesn’t sound like one of the droids in Star Wars that you’re just getting attached to when it gets killed off. This is the underlying protocol beneath Bluesky’s infrastructure and a growing ecosystem of other apps, like, well, the one you’re reading this on right now.

How does it work? Well, bearing in mind that I have all the technical savvy of a bowl of leek and potato soup, this is my understanding of it. I then took my understanding, realised it didn't have much understanding in it, and I ran it past some much smarter people. So you can be sure that it's at least partly right.

When you’re looking at the internet as we’re used to, be it a website through a browser or an app, you’re looking at a document-delivery system. When you visit a webpage, a server sends your browser a package of HTML, CSS, and JavaScript, all rendering instructions. The meaning of the content (this is a book, this is a review, this is a price, this is a social follow, this is a video of a dancing cat, oh no, this is AI slop, oh look, a fascist) is locked inside that rendering code. Different websites can't interoperate because they each encode meaning privately, in their own way. When you interact with those sites or apps as a user, your user data sits on that system and belongs to them.

With AT Protocol, that relationship is fundamentally shifted. The AT Protocol is made up of three layers. Your PDS is the foundational layer of the stack, where your identity is stored, called a decentralised identifier, or DID. The second layer is centralized server infrastructure (known as the relay), which aggregates data across every PDS accessible on the internet for applications and services on the AT Protocol to use. And the final, top layer that wraps around that is apps like Bluesky, Offprint, etc. Now, it’s worth noting here that this currently doesn’t exist by default. If you approach the Atmosphere armed with little more than a Bluesky account, then that PDS lives with Bluesky. But you can move it to be self-hosted, or to another PDS, like Blacksky or Eurosky, which aim to give you greater ownership of your data. So, this benefits people who put in the effort to have digital sovereignty of their data.

But I digress. Where was I? Ah, yes, your PDS. Whether hosted by Bluesky, yourself, or some other provider, these are windows into your data. The revolutionary thing is: if any app disappears, or its CEO turns out to be lizard-brained, your data can stay with you, because it lives in your layer, not theirs. That means that whatever website or app you interact with, your data stays yours.

Let’s look at an example. I’ve been wanting to get off Goodreads as a user for ages, because it’s owned by Amazon, and it’s genuinely one of the worst places on the internet. Absolute hot garbage. If I move to something like NeoDB that is built on AT Protocol, I can rate books, build lists, log films, TV, music, games, and podcasts, sharing that activity to Bluesky (and the wider Fediverse, but let’s not get into that because my head is already beginning to overheat like a Cybertruck in a car park) through a system that’s built on the same protocol.

When you interact with a piece of content on AT Protocol (a book review, say, or a podcast episode), the site asks for information from a centralised server (known as the relay), and confirms whether or not every record matches the Lexicon’s schema. So, even if that content was created by an app you've never used, the protocol doesn't just shrug and give up. Instead, it performs a quick lookup, a bit like asking a Babel fish, ‘What is this thing, and what are the rules for working with it?’ The answer comes back from the Lexicon, and suddenly your app knows exactly what it's looking at and what to do with it.

What is the Lexicon, I hear you ask? No, it’s not the baddie from one of the more recent Mission Impossible films; it’s a global schema blueprint that the AT Protocol is built on, like a shared dictionary that any app built on the protocol can read. Using it, any developer can define records in repositories of data, essentially telling it what a 'thing' is — a book, a review, a recipe, an event RSVP — and every other app across the ecosystem can interpret and understand it if willing to do so. So 'book' or 'reading event' or 'edition' could be defined once, and then any app, built by anyone, would know what to do with it.

Then, if NeoDB turns out to have directly funded a genocide for some reason, migrating your data to another platform is feasible as long as it also uses the AT Protocol. I find another app that’s doing the same thing and, as long as it supports everything that NeoDB supports and the app developer implements a way to view them, I just have to log in through the AT Protocol. Boom. Everything’s there. All my followers, all my reviews, all my lists, everything.

This is why the 'new Twitter' framing for Bluesky slightly misses the point. AT Protocol isn't just asking what replaces That Fucking Place, it's asking what happens if the semantic layer of the entire internet were open and interoperable.

Publishing, untethered from the big river.

Okay, I’m going to stop there in trying to explain what AT Protocol is, because, quite frankly, I don’t really know what I’m talking about on a technical level. There are plenty of really good pieces already out there on AT Protocol, and I’ll link to some of the key ones at the end.

The real question for me, as a writer and publisher, is this: if AT Protocol becomes the foundation of a new internet, what does publishing on that new internet look like?

Well, this is where we deviate from what is possible now and start to look at what might be possible.

Imagine a Lexicon for books. If you define records for: an edition (title, author, ISBN, format), a reading event (started, finished, abandoned), a review (linked to an edition, with rating and text), a purchase (proof of ownership, tied to your DID), and a lending event (temporarily sharing your ownership right with another DID), you have a working lexicon for books.

Once that Lexicon exists, any reading app (on any device, made by anyone) can render your library, because ownership is stored in your PDS, not in, say, Amazon's database. As a reader, if you bought an eBook, for example, from an author directly, or through a publishing platform, or through a store of any kind, you could access it through a reader built on AT Protocol that could be completely separate from any bookstore, or tied to it. And if you want to change the eReader, you can, because the ownership of that book remains with you. But for the author, that sale remains, but so too does any data that goes with it. The reviews for that book on one system are visible on another.

As a publisher, you make the book available, and any bookstore built on the Lexicon can request to sell it within their ecosystem. Or maybe a reading app comes along with a subscription model, a la Kindle Unlimited. They can agree that relationship with the authors, and that author gets a whole new audience. Or they can say, no thanks. Or, a decentralised bookstore becomes possible. Imagine a cooperative or non-profit marketplace where the catalogue, reviews, and purchase records are all on AT Protocol, and multiple storefronts can compete to sell the same books without any single one owning the customer relationship. We could see authors contributing to, for instance, a decentralised instance to raise funds for a charity.

For the reader, proof of ownership for an eBook lives in your PDS as a cryptographically signed record. You bought that book, and that fact is in your data repository, tied to your DID. That’s wildly different to Kindle, for example, where you only license the right to access a copy, a license that can be revoked at any time. Any reading app that understands the Lexicon can verify your ownership and render the book. Now, I must stress that right now, this is only an idea of what could be possible. As things stand, currently, records can't verify purchases or ownership, beyond ‘user is the owner of that record’, so in the case of buying an e-Book, we'd need the following two things to happen:

• A private state needs to be implemented within the AT Protocol so that authors can feasibly store their books without concerns about piracy (especially from AI companies).

• A lexicon schema would need to be established that becomes standard in the industry.

I’m sure they’re both dead easy, though, right?

Let’s get radical

If we look beyond selling books, at the wider creator community, AT Protocol could provide a similar radical transformation, no matter the creative discipline.

Say you have a podcast (I mean, you’re reading this as a human being on the internet, so there’s a good chance you have a podcast) that you currently publish through Spotify or Acast. You monetise that through Patreon. One of the benefits you offer to listeners is Discord access, and the other is a dedicated ad-free podcast feed, maybe with bonus episodes. And you run a Substack alongside it, just for good measure.

At present, that's three or four separate ecosystems that you're working in. Four ecosystems run by absolute bastards, by the way. Just terrible human beings. In the future, if you move to ecosystems that are built on AT Protocol, all of the friction goes away because your listener has an AT Protocol handle that stores all the information natively. So they might have a podcast player built on AT Protocol that logs their use of your main feed, but when they join through an AT Protocol Patreon equivalent, it automatically switches the RSS feed to the paid one by changing the setting on your PDS, which is what the podcast player pulls from. They might then automatically receive an invite in, say, Roomy (a Discord alternative), to join the dedicated server. Less work for the creator, less friction for the user, and everyone in turn keeps all their own data, so they could move, say, podcast player, and the protocol handles all the migration.

On top of that, for the creator, the people who enjoy your work will be able to shout about it much louder, with clean embeds for sharing across Bluesky, or whatever other AT Protocol-led social media is designed, because it's all built on the same basic foundation. A fan of your podcast could share an episode on Bluesky, which in turn turns up as a review on the podcast app, and their audience might listen and subscribe without ever leaving Bluesky.

The same goes for newsletters. Currently, I read a lot of articles through an RSS reader. If I have an article come through that’s pay-walled, I have to leave the RSS reader to log in and read it, which diminishes the reading experience for me, adds a layer of friction, and makes me not want to subscribe. But if the reader and the publisher are on AT Protocol, that can change natively. And sharing that article across the other social media on the protocol becomes easier. Hell, you can publish on Offprint or Leaflet or Pckt, and readers can follow those blogs inside your own Bluesky feed, right now.

Across the board, the pattern is the same: less friction, more ownership, and a web that works for people using it rather than solely the companies monetizing them.

That's the vision, anyway. The reality, as ever, is a little more complicated.

The small print…

So that’s what’s possible. The truth is, we’re still a ways from this right now. As I mentioned earlier, when it comes to publishing, the missing piece that doesn't yet exist is the commerce and ownership layer — a Lexicon for book purchases and eBook delivery. That's unbuilt territory.

As well as that, AT Protocol is still trying to grapple, as I understand it, with private and paid content. Private data is still in development, potentially using end-to-end encryption similar to what's being developed through the IETF's MLS standard. Please don’t ask me what that is, though, because I absolutely couldn’t tell you, because I'm cosplaying tech literacy again. I read it somewhere, and that’ll have to do for now.

Until that's mature, I, as a publisher, can't put a paid eBook behind an AT Protocol ownership record in a fully decentralised way. You'd need a centralised component for the actual file delivery. It’s also worth noting here that all of this is reliant on Bluesky's willingness to complete an implementation of the private state. For all the talk of decentralisation, Bluesky is the one driving this, and if they decide they don’t want to, well, then I guess we’ll all continue to be shit out of luck. But it does seem to be something they want to implement.

Then there’s the question of the app ecosystem. Now here, things are moving fast, but we’re a long way from seeing a fully mature ecosystem that’s likely to bring in people in the kind of numbers that are needed to achieve traction online.

When Discord appeared to shit the bed in the most spectacular way possible, I saw a huge amount of talk about finding ‘a new Discord.’ But save for a dent in Discord’s Nitro income, I don't see any wholesale movement from any of the servers I am in. Is that a lack of options, server owners not wanting to make the move, or just that the experience of being on Discord, in those servers, has not radically changed? I suspect a mix of all three. There’s a really interesting Discord alternative called Roomy being developed, but it’s not there yet.

Finally, there’s the switch. What makes a social space comes from both the environment — its ecosystem, its moderation, its features, and its aesthetic — and the people who are there, and the things they create.

Take Letterboxd, for example. This has been, for me, my most pleasant social media experience in the last few years. When I watch a film now, I log it, leave a silly little review, and read reviews by others. Over the last few years, I’ve found a few people to follow, and I follow a few people I know from Discord servers, etc. It’s fun. I trust Letterboxd averages now more than most other metrics when looking at where I want to watch a film.

Now, there already exist alternatives to this on AT Protocol, but they don’t have those friends or their reviews. They don’t, in fact, have a huge amount of reviews, because there aren’t many people there. So why would I switch away? So far as I know, Letterboxd isn’t owned by someone who likes to bathe in the blood of virgins, and if they are, please don’t tell me.

Where I believe AT Protocol has a chance to shine, I think, will be in its ability to provide easier switching as it matures. The more AT Protocol apps a person is on, the easier it becomes to develop that ecosystem. And as more and more of our corners of the internet get enshittified, the more attractive these protocols will become, and the ecosystem they exist in. The more that happens, the more attractive it will be for developers to work with the protocol, and the whole thing becomes a self-driving car, except not one of the ones that drives you off an overpass, or bursts into flames. We hope.

And yet…

As an author and a publisher, I am going to keep my eye on The Atmosphere, because I think it might just be the future, not just of the social web, but of publishing. A publishing that frees up authors, publishers, and readers. A publishing ecosystem that lets authors just write, readers just read, and nobody's data belongs to a billionaire along the way.

The infrastructure for a better internet is being built right now, quietly, by people who actually give a shit, and that can only be a good thing.

Right?

Further reading:

Joe Debasser (@joebasser.com )'s two blog posts about AT Protocol were what first piqued my interest, and I would recommend them as a great jumping-off point.

Muni blog has a great post by @zicklag.dev about what AT Protocol is, and isn’t

And they have a fantastic post by Erlend Sogge Heggen (@erlend.sh ) about On-Protocol Organizing, which is a fascinating look at building communities on the protocol, which touches some of the same points as the above, but from a different angle.

And lastly, this essay is published on @offprint.app , which is already an incredibly well developed alternative to Substack, built right on the protocol, and they have a great post that goes into what it can do.

Offprint is built on (and is a main architect of) standard.site the protocol underlying a lot of the way blogging and web development is moving on At Protocol, and their site, too, has a lot you can dig into to learn more.

About the Author

Paul Stephenson writes pulp fiction for the digital age. His first novel series – the indie-published apocalyptic Blood on the Motorway trilogy – has been an Amazon bestseller on both sides of the Atlantic. A former journalist, he holds a diploma in Creative Writing at Oxford University.

His stories have been featured on the chart-topping horror podcasts, The Other Stories and The Night’s End. His newest project, the eBook serial The Sunset Chronicles, is a dystopian sci-fi thriller that will delight and terrify fans of science fiction and horror alike. He is also the creator of the podcasts, Bleakwood, tales of terror from a mysterious English town, and Darkness Come Alive, a modern vampire tale set in York.

He lives in England with his wife, two children, and one hellhound.

Check out all his books here: