Week Ending April 17, 2026 | Author Update

Draft2Digital levies a tax against the “poors,” Amazon kills millions of old Kindles and the Windows 10 Kindle app, and Barnes and Noble Press drops three policy bombs that could wipe your books from sale. Plus, an AI artist just hit number one on iTunes without a single human voice, Ukraine’s robots won their first solo battle with zero casualties, and fresh Pew Research data maps exactly where your readers draw their moral lines. Most importantly, it’s our first birthday! You have our permission to eat a piece of cake while we update.

Author Update Celebrates Its First Birthday

We launched Author Update on April 16, 2025. In December we moved the live show from the Novel Marketing channel over to its own channel which now has over 2,100 subscribers and more than 111,000 total views across the episodes and clips.

Do you know what we want for our birthday? New subscribers. Tell a friend about the show and make sure you have subscribed and clicked the bell icon to get notified when we go live. 

Thomas: Also special thanks to those of you who have become paid channel members and thank you for your YouTube Super Chats! Your support helps keep this show on the air. Our goal in year two is to get to break even on direct costs for the show. 

Draft2Digital Introduces Account Fees for New Sign-Ups and Low-Earning Authors

Jonathan: I am angry about this, not at the news itself, but at how the author community is reacting to it.

I got this story from Kristen McTiernan, who runs the YouTube channel Nonsense Free Kristen. She noticed that Draft2Digital published a blog post on April 14th, 2026, introducing account activation and maintenance fees for the first time in the platform’s history. If you sign up for a new account, there’s a $20 fee. If you make less than $100 in gross sales per year, you pay a $12 maintenance fee.

The planet flipped out. Everybody in the author community said it’s a tax on the poor. “If somebody’s already not making $100, why would you charge them? Why would you charge new people?”

This is a basic defensive strategy taken by literally every platform, including Author Media Social, to keep out bots. Do you have any idea how much Draft2Digital is paying to host book pages? Do you have any idea how many of those are AI slop? People were using the volume approach, flooding the platform with AI-generated content, and Draft2Digital had to host all of it.

They’re just cutting out the AI slop. Relax. If you’re not making $100 a year on your books, you need to pay more attention to the Novel Marketing podcast or something.

Thomas: If this $20 fee is hard for you, you don’t need to be writing books. You need to get a job.

Jonathan: Get married rich or something.

Why isn’t $20 a real barrier?

Thomas: Writing books is not a good source of small money. If you’re in a place where you can’t find $20, there are hundreds of thousands of open job positions. They may not be glamorous.

A lot of authors are seeking status and affirmation, but if you’re willing to give up status, there are a lot of low-status jobs that pay really well and can solve your money problems. I know authors making good money cleaning houses. They listen to podcasts and brainstorm their book while working, then go home and write. But a lot of people are status seekers even though they don’t have the money for it.

People who know what they’re doing with AI can spin up 100 or 500 different pen names, and this new Draft2Digital strategy is going to crush that approach, because now the only AI authors who survive are the ones who found a real readership.

Part of how running an AI puppet network works is that you’re using the machine learning algorithm on Amazon to find readers. You throw a whole bunch of books and authors at the algorithm, playing a numbers game. The hope is that a few take off. Since it doesn’t cost much time to write a book, just a few hours of labor and $10 or $20 worth of tokens, you’re carpet bombing and hoping to get lucky. This fee is going to inhibit that strategy.

I think it’s good for the industry overall. If you want to get rid of slop, this is how you do it. It’s not just going to eliminate AI slop.

It’s also going to clear out slop from authors who haven’t done the work to figure out how to write the kind of book people want to read. It’s an anti-slop strategy across the board. $20 is not onerous. If it is, get a job in the real world. It’ll help you as an author. You’ll interact with people of other ages and backgrounds. You’ll meet characters at your job that might make for good characters in your book.

Does this hurt new authors’ access to libraries?

Jonathan: I’ve seen this talking point a lot: “This limits new authors’ access to libraries because they pay the $20 setup fee and then $12 a year.” You’re telling me the only thing stopping them from getting into libraries is $32? That has not been my experience. If I only had to pay $32, that would not be the problem. Getting into libraries is hard because people have to request your books, there are gatekeepers, and the process is complicated. It’s not a $32 barrier. If that were all it was, it would be super easy.

Thomas: This $32 barrier will actually make it easier for you, the human author, to get into libraries because there will be less AI slop for you to compete with. It’ll be easier for you to stand out. This is really good news if you look at it from a game theory perspective.

Is the outrage even real?

Thomas: I don’t know why people are freaking out so much. I don’t know if they’re all human.

Part of the reason I didn’t want to cover this as our opening story or make it our headline is that I think most of the outrage is coming from a handful of puppet master humans getting all of their bots to do the outrage.

I don’t know a single human who’s actually bothered by this $32 fee, because real humans spend real money on their real books. If you spend 500 hours or even 100 hours writing your book, spending $32 to get it listed is no problem. You don’t have to be on Draft2Digital. You could just do Amazon KDP, which is what I recommend for beginning authors anyway.

Everybody calm down.

Sources:
Understanding D2D’s Activation and Maintenance Fees
Draft2Digital FAQ on Fees and Publishing Requirements
ALLI Coverage: Draft2Digital Introduces New Fees

Amazon Retires Early Kindles

Thomas: Amazon is in the process of bricking all of the original Kindles. They’re not bricked yet, but starting on May 20th, you won’t be able to buy new eBooks from the old devices.

Jonathan: Amazon sent targeted emails around April 7th to customers still using devices released in 2012 or earlier. The company is ending Kindle Store access for those models on May 20th. Owners will keep any books they’ve already downloaded as long as they stay logged in, but they lose the ability to buy, borrow, or download new titles directly on the old hardware after that date.

This will probably hit middle grade and children’s authors the hardest. The Kindle already had low penetration into the middle grade market, and what Kindles did exist in the hands of children were usually older models. Parents were looking for deals to put eBooks into their kids’ hands, or they considered the older devices safer. Honestly, as a parent, I kind of like it better if my kid has no access to the Kindle Store from the device.

Thomas: You also can’t load in new books, though.

There’s already very low penetration of Kindles into the children’s and middle grade markets. I often get questions from middle grade authors asking what kind of reader magnets they can use. The challenge is that the middle grade readers they’re writing for don’t own Kindles, so all the eBook-based reader magnets don’t work well. Nonfiction eBooks targeting parents can work, and there are other reader magnets like coloring pages and checklists.

What reader magnets work for children’s books?

Thomas: I’ve actually developed a series of checklists for my own kids because I realized I’m raising a bunch of large language models, and if I use better prompting, we can get out the door with less friction.

My wife will vague-prompt, which works on our older children. She’ll say, “It’s time to get ready to go,” and my seven-year-old knows what that means. My four-year-old is befuddled. He doesn’t realize that also means the potty needs to happen and all the shoes and socks need to get found.

I created a checklist with little emojis for each step. Find the shoes, find the socks, wear the right clothes, very basic stuff. Now my four-year-old will go point to each emoji on the checklist. We have a go-to-bed checklist, a getting-ready checklist, all of them. Something like that would make a great reader magnet for a children’s book author. An eBook wouldn’t.

Could the Kindle transition benefit authors?

Thomas: eBooks will not be good reader magnets for children as a lot of kids’ old Kindles stop functioning.

On the other hand, Amazon is trying to get these original Kindle users to upgrade. Part of the deal is 20% off a new Kindle, which isn’t that much, but they’re also giving them a $20 eBook credit.

That $20 could go to you, dear author. These Kindle upgraders with their $20 credit are going to want to spend it before it expires. There may be a little extra money on the table for authors who are promoting around this transition.

It’s hard to know if your target audience even has one of these original Kindles, although two million is a lot. We don’t know how many are still in operation. I bought an original Kindle and have no idea where it is. It got put in a drawer somewhere and slowly stopped working.

Why is Amazon doing this now?

Thomas: To be fair, supporting these old Kindles creates real friction. The original Kindles were connected through a contract with Sprint, a cell phone company that got purchased by T-Mobile. They had this Whisper Sync technology traveling over Sprint’s network. There is real tech debt there. I understand why Amazon wants to do it, especially if it means selling more devices.

But it does make me sad that none of our gear lasts. Everything you buy breaks, and you’re basically always renting your computer equipment. The old equipment will just cease to function.

Even if you really love Windows 10 as I do, that doesn’t mean you get to keep using it forever. You may get forced to upgrade to Windows 11, which is the absolute worst. I will say I only like Windows 10 for gaming. I don’t trust it with any work stuff.

Sources:

Amazon Official Help Page: Kindle E-Readers Released in 2012 or Earlier

BBC: User anger as Amazon ends support for some older Kindles

The Guardian: Amazon upsets ebook lovers by ending support for old Kindles TechCrunch: Amazon to end support for older Kindle devices

Amazon Drops Kindle Desktop Support for Windows 10

Thomas: Besides dropping the old Kindle hardware, they’re also dropping Kindle desktop support for Windows 10.

Jonathan: Good E-Reader reports that Amazon will shut down the current Kindle for PC app. The company confirmed it is building a replacement that will only run on Windows 11 and will arrive exclusively through the Microsoft Store. The original app launched in 2019, and Amazon delivered a few updates over the years. Readers downloaded books straight to their hard drives, and some people were stripping the DRM from those files. The switch to a native store app blocks those steps far more effectively. Authors will gain better protection against unauthorized copies spreading online.

Windows 10 users and anyone on older systems are going to lose their native desktop option. The operating system still runs roughly 27 to 31% of desktop computers as of early 2026, which surprised me. You can still reach your full library through the web browser, and the Cloud Reader works on any computer with no installation required. But if you’re going to strip DRM, you’re probably going to find another way to do it anyway.

How does this help authors fight piracy?

Thomas: The app they’re getting rid of was a channel for your book to get pirated. Even if you kept DRM on your book, savvy people were able to use this app to remove the Digital Rights Management protection, then upload the file to a piracy website. Those piracy sites then source content into large language models.

This is what Anthropic got caught doing, but all of the other AI companies are doing it too.

I was listening to a Stanford presentation on how to create a large language model in their computer science program. They talked about how they rank different content and how they rank books higher. The message was essentially that they all train on books, but they want to hide that fact. The presentation walked through the steps you use to obscure the fact that you’re training on books, even though you are.

Sources:
Amazon is discontinuing Kindle for PC on June 30th
Amazon Pulling the Plug on Kindle for PC on June 30th
Kindle for PC Goes Dark on June 30 – eReaders Forum (details Windows 11 requirement and market-share context)
MobileRead Forums – In-app pop-up text and Windows 11 confirmation

Barnes and Noble Press Sets $14.99 Minimum for Print Books and Caps Accounts at One Hundred Titles

Thomas: Barnes & Noble is setting a floor price of $14.99 on all paperback books in their store published by indie authors. This is yet another anti-AI-slop strategy. The best way we prove our humanity is with commercial activity. The bots don’t have dollars, so everyone is raising prices as a way of discriminating against the droids.

Jonathan: Barnes & Noble published an official help center article this month announcing three policy changes taking effect in the coming weeks. The company says it aims to protect independent authors and raise the overall quality of titles on its platform.

The first change hits printed books the hardest. Starting April 22nd, Barnes & Noble Press will block authors from creating any new print listings priced below $14.99. If you currently sell a title for less than that, the company will remove it from sale after May 14th unless you raise the price yourself. You can log into your dashboard right now and update any retail prices you need to keep live.

The second change limits how many books one account can keep on sale. Beginning May 14th, Barnes & Noble Press will enforce a 100-title cap per account. Each book counts as one title whether you offer it in print, digital, or both formats. Titles you upload but keep off sale don’t count toward the limit. The company states it may remove excess titles at its own discretion to bring accounts back under the cap. You can delete titles yourself through your dashboard if you want to stay compliant without waiting for them to act.

The third change ends the sale of public domain material entirely. Barnes & Noble Press built its service to publish and promote original work from independent authors. Starting April 22nd, the company will remove any public domain titles immediately and will no longer accept new uploads of public domain content.

Isn’t Barnes & Noble protecting its own public domain editions?

Thomas: If you’ve ever been to Barnes & Noble, they have those collectible editions.

Jonathan: They don’t want the competition. That’s 100% what this is about.

Thomas: I’m actually building out a collection of Preston Speed edition G.A. Henty books. My goal is for my backdrop to eventually have a complete collection of Henty books published by Preston Speed Press, a publisher that’s no longer in business.

I’ve been looking for Henty books, and the amount of slop on Amazon is overwhelming because all of his books are in the public domain and have been forever. It’s very easy to get an AI cover for a Henty book and upload the file from Project Gutenberg to Amazon, coasting off the value Henty created back in the Victorian era. It’s not a problem when one person does it, but when 30 people all have different versions that are basically the exact same book with different AI covers, it fills the market with noise.

Are these changes good for the industry?

Thomas: I’m not against any of these changes. These are the inevitable reactions to AI slop, and they also have the effect of reducing human-made slop. There’s a lot of corporate slop and a lot of low-effort slop from authors who haven’t developed their craft, who don’t know how to market, who rush their first book out there. Those authors need to listen to Novel Marketing and learn how to do it right.

The 100-title cap isn’t targeting humans who spend time handcrafting every one of their books. If you’re an AI author making enough money, you can create more accounts and manage it. But this reduces the quantity carpet-bombing, machine-learning strategy we talked about with Draft2Digital. It’s the exact same story, just different companies trying their own methods.

This is what we did at Author Media.social, my online social network for authors. I offered it for free for years, and it’s still free if you’re a patron or coming to one of my conferences. But I recently added a one-time $9 fee to keep the bots out, and it worked. I don’t think we’ve had a single bot enter once I put up that tiny paywall. The quality of the community has really increased because people appreciate a human-only, curated space.

Sources: April 2026 Barnes & Noble Press Policy Updates B&N Press April 2026 Newsletter Getting Started with B&N Press

Author Media Launches Noble Dark Mug Collection

Thomas: I’m excited to announce that Author Media has launched its Noble Dark Mug Collection. It’s a series of eight mugs available for a limited time. Just like the zeitgeist, they’re going to come and go.

One of the mugs is “Turtles All the Way Down.” I also have one called the Mug of America, and it has two sides. The side facing the camera says “It’s a free country,” and the side facing you says “You are an adult.” The reminder for you is to take responsibility for your actions. We left the safety nets in the old world. But it’s also a reminder to the people you’re on a call with that they can’t interfere with your freedom. It’s a warning to the king not to tax our tea unless he’s going to give us representation. It also has a very angry-looking eagle on it.

All of the mugs are compatible with coffee and tea, except perhaps the Mug of America. I think coffee will taste better in that one.

Jonathan: You can’t put tea in that one.

Where can you get them?

Jonathan: Visit merch.authormedia.com. It’s not a big revenue source for us. These are printed on demand, so the cost is kind of high and we get a couple bucks per mug.

One of them says “Love Thy Reader” on one side and “Love Thy Book” on the other. The greatest book marketing commandment is “Love thy reader as much as you love thy book.” Those two things go together.

Another mug with two sides says “Author” on the side facing the camera when you’re on a Zoom call, but on the side facing you, it says “Darling Slayer.” Other people see the glory, the success, your name on the book. Only you see the sacrifice required to make your book a success.

I’ve ordered mugs for me and for Jonathan. Once those show up, you’ll see us with some product placement. I’d love to see you all using them on patrons-only Q&As.

Jonathan: He asked for my opinion on what should go on these, and let me say they have been heavily censored and brought within guidelines. Maybe we’ll do an uncensored mug collection as a limited release later.

Thomas: I did have fun with the marketing copy, though. Check out the product descriptions because they’re pretty intense for some of them.

We may do other merch drops, and some of these will get retired, so they won’t be available forever.

Noble Dark would also be a great coffee name. We have grim, bright, noble, dark. Those could be the four zeitgeist coffees.

Jonathan: Tea could be cozy mystery.

Thomas: If we get a bigger audience, I may look into tea and coffee. For right now we’re just doing mugs. Check out the Mug of Determination. I can’t tell you anything more about it, you have to read about it yourself. But if you’re needing some inspiration and some stick-to-itiveness, some grit, the Mug of Determination is the one you should check out.

Sources:
Noble Dark Mug Collection – Author Media Merch Shop
The Mug of America Product Page
Darling Killer Product Page
The Mug of Focus Product Page
The Greatest Mug Product Page
Mug of Zeitgeist Product Page
Mug of Determination Product Page
Mug of Persistence Product Page
Turtles All The Way Down Product Page

Registration for the Dictation Bootcamp Now Open

Thomas: Sarah Elizabeth Sawyer does a dictation bootcamp from time to time. I’ve had her on as a guest on Novel Marketing talking about dictation. She’s a trained Cherokee oral storyteller, which puts her in a unique category for teaching dictation. The Cherokee are one of the few peoples who preserved oral storytelling from the olden times, and that training makes her approach far more useful. When you incorporate oral storytelling techniques into your dictation, it gets better.

I’m a big fan of getting trained broadly. If you’re an outliner, you should learn how to write by the seat of your pants. If you write by the seat of your pants, you should write at least a short story with an outline.

Learning from a Cherokee oral storyteller is beneficial for just about any author. You’re going to get something out of it. Check out Sarah Elizabeth Sawyer.

Sources:

Dictation Bootcamp for Fiction Authors — April 27 Register for Dictation Bootcamp Triple Your Writing Speed with Dictation

Spotify Rolls Out In-App Print Book Sales and Expands Audiobook Discovery Tools

Jonathan: Spotify has rolled out in-app print book sales and expanded its audiobook discovery tools. The company’s newsroom reports that it launched new audiobook features this week letting listeners buy physical books and switch formats with less friction.

We reported a few weeks ago that this was coming, and the feature is now live for users in the United States and the United Kingdom. Listeners can purchase print copies straight from the Spotify app through its partnership with Bookshop.org.

Android users gained access right away. iOS users will see the option next week. Listeners who find an audiobook they like simply tap to order the physical edition. The setup routes sales through Bookshop.org so that independent bookstores get support.

Is this good for authors?

Thomas: We reported on this before and framed it as Spotify declaring war against Audible and really against Amazon. These are the first shots of that war. Think of it like an old-fashioned 18th-century war where you send an ambassador with the official declaration, and he reads it in the court and hands it to the king with all this pomp and ceremony.

You may already be on Bookshop.org. I don’t think you have to take any action to get your print book in there.

I see this as really good for authors because more competition is good, especially with Audible giving indie authors a bad deal right now. If Spotify can take more market share and more authors start switching, and if more readers switch over, that’s the real sticking point. Audible is able to give indie authors a terrible deal because they have 85% market share. With that kind of dominance, you get to set your own terms and give everyone a take-it-or-leave-it offer. If they leave it, they’re fighting over scraps.

If Spotify can gobble away some market share, it’s going to be really good for all of us.

Jonathan: If you look at it as a hand-to-hand duel between two giants, this is not a huge step.

Amazon isn’t going to lose a ton of paperback sales through Spotify. But I think of it as a foot pivot. In a martial arts duel, when you swivel your foot, it means you’re opening up. Power generation can flow in a different direction.

My little brother once said he thought the other guy was about to die because he saw the foot turn on the ground. When a martial arts fighter’s foot turns, he’s lining up a hit. To me, this is a foot turn. Spotify is prepping for a much bigger struggle by creating a downhill ramp to slide more power into that fight.

Can Spotify be trusted long term?

Thomas: A viewer asked whether Spotify can be trusted long term, because they don’t have a great reputation.

Jonathan: None of them can be trusted long term.

Thomas: None of these companies are good. None of them love you. None of them care about you at all.

What’s good about this for us is that their self-interest causes them to keep each other accountable. Spotify doesn’t care about you, but it does care about Amazon. Amazon doesn’t care about you, but it does care about Spotify. If they can fight over who treats us better, that’s our best protection as consumers.

That’s why monopolies are so bad for the individual. Once a company gets dominant market share, like Audible, they have no selfish incentive to treat their customers well. You’re left hoping for noblesse oblige, which corporations are incapable of because they have a fiduciary duty to shareholders.

If you’re a publicly traded company, your biggest shareholders are effectively artificial intelligences run by BlackRock, Vanguard, and the other big investment firms. There is no noblesse oblige. A family-owned or founder-owned company might have that, but big corporations have a fiduciary duty to maximize shareholder value. They have to believe that treating you better is a way of maximizing shareholder value.

Sources:
Spotify Expands Audiobook Features, and Printed Book Sales Go Live in the US and UK
Audiobook Charts Launch in US and UK
Page Match and Bookshop.org Partnership Announcement

Novel Marketing Conference Ticket Sale Update

Thomas: The Novel Marketing Conference Super Tickets are officially sold out. There’s a chance one purchase was made by mistake, so I’m waiting to hear back. If that’s the case, one Super Ticket may reenter the market. Otherwise, they’re gone.

Standard tickets are selling quickly. Last I checked, there are only 54 left and about 18 gallery tickets remaining.

To learn more or secure your spot, visit NovelMarketingConference.com. We sold out last year, and tickets are selling even faster this year. I never even got a chance to announce it. I still haven’t done my episode on Novel Marketing about the conference, and by the time I announce it on the main channel, there won’t be any Super Tickets left.

I’m excited about the conference and would love to see you there. Grab your ticket, because once I do that announcement, there may not be many left.

AI-Generated Song Reaches Number One on iTunes Charts

Thomas: We’ve had AI songs top Spotify before, but that can be gamed with bots generating streams. To reach the top of the iTunes charts, people have to buy the song with actual dollars. This demonstrates very clearly that humans don’t care how a song was made. They only care if they like it, and a lot of people can’t tell if something is AI-generated.

Jonathan: Pop Vortex reported that “Celebrate Me” by Inga Rose sits at number one on both the US and global iTunes charts. Inga Rose does not exist as a human performer. Dallas Little is a content creator based in Greenville, South Carolina, who produced the entire project. He wrote the lyrics himself and fed prompts into Suno to generate the stems, arrangement, and vocals. Inga Rose released the single on March 31st under the Myers Music Label, and by April 17th the track claimed the top download spot worldwide.

How much human craft goes into an AI-generated hit?

Thomas: I have a friend who’s building out a musical using Suno as part of the workflow. There are basically two ways to use it. There’s the quick “make a happy birthday song for my friend” method, where it’s a quick prompt and a cheesy song meant to be listened to once or twice by one or two people. Then there’s the very sophisticated approach with detailed prompts, a lot of artistry going into them, and a lot of iteration. The end product is the kind of thing that can chart.

That’s what we’re seeing here.

This wasn’t somebody saying “write a chart-topping song for me” and hitting go. There was a lot more to it. But the performance is still AI-generated. The creation and writing of the song is human, but it’s not a human-performed song. The human performer has been eliminated, which is what we’re starting to see with AI puppet master authors. It’s the same carpet-bombing strategy on Spotify, putting out a dozen songs a day and hoping some pop through machine learning.

I will say, the lyrics of this song “Celebrate Me” are evil. I started listening and thought, this is a song that would appeal to a very narcissistic person.

Music has gotten very degenerate and disordered, but this song is evil in an entirely different way. Most songs are disordered because they have a distorted and unhealthy sexual ethic. That’s not the problem here. This song is all about how I am the greatest, how you must celebrate me, how I only need myself. If you were to embody these lyrics, your life would be very sad and very dark.

It makes me sad this song is on top of the charts. It makes me sad for all the human performers who could not compete with it, although a different song will probably be number one tomorrow. But it also makes me sad that these lyrics are resonating with so many people.

We need people who can write good, ordered lyrics that inspire people toward the good, the true, and the beautiful to use AI and compete.

What enabled AI music to win in the first place?

Jonathan: I saw a meme about Brazilian jiujitsu that said, “How do I get myself out of the problems I made for myself?” This is what came from artists using autotune instead of actually singing. AI can use autotune too.

Thomas: It’s also the result of using machine learning algorithms to pick our next song. A lot of people just listen to their Discover Weekly or Release Radar playlists on Spotify. That’s all machine learning curation.

As soon as you hand over curation to machine learning, you eventually hand over the winning to AI generation. If you want to fight AI-generated content, you have to first fight the algorithms selecting the content.

Nobody fought those algorithms. Nobody is fighting against Amazon’s machine learning recommendations. Everyone’s trying to adapt. Same thing on Spotify. To win in a machine learning context, quantity is so much more important than quality. You have to try a lot of stuff. The authors making more and more money are the ones using the AI puppet master strategy.

If you want to compete with this, you really have to play the quantity game.

What does AI worship music mean for the church?

Jonathan: Christian music has been about two years behind secular music for a while. We’re already seeing AI Christian music getting into church worship services. People are still pretending they’re having an emotional experience with some seven-eleven song, the same seven words eleven times.

Now they’re going to be pretending to do it with AI-written lyrics and AI-written music. It’s going to raise very serious questions. I think we’re going to experience another surge of people falling away from the faith. They’re going to find out the song was written by AI and ask, “Can this even be worship? What have we been doing?” Because it doesn’t look any different.

Thomas: I disagree, because I think the real question is whether it’s true, not whether it’s “real.”

Your emotional experience is real, but what matters is whether the content is true. Humans can create things that are untrue, and AI can create things that are true.

I was at a religious event where somebody was reading a reflection on scripture. It was very moving for them, they were weeping. All of the individual sentences were true, and I detected no doctrinal issues.

But I also detected many vectorized sentences. In vectorized sentences, the pattern is “it’s not just X, it’s also Y” or “it’s not X, it is Y.” AI understands knowledge relationally and expresses its views relationally, even at the sentence level. Once you know to look for vectorized sentences, you can identify AI-generated text even when a human is reading it aloud.

Jonathan: When something is written or generated for worship, it involves two points. There’s the origin point, the anchor, where the worship is transmitted from. That’s the worshiper like David writing the Psalms or a songwriter writing a hymn.

Then there’s the reception point, where it’s aimed. AI lacks the anchor point because it’s not offering anything up to God. It’s just generating what it thinks everyone else is talking about. When a person repeats that, it introduces error. It’s not an actual original point of transmission for worship.

Thomas: This is a fascinating discussion and a little off-topic for this show. I feel like this is an “even the rocks will cry out” moment. It’s more of an indictment on us as humans.

Jonathan: The rocks aren’t generating worship. They’re creating, not imitating.

Thomas: On the Christian Publishing Show, we need to have this debate in more depth.

Sources: Exclusive: iTunes Gamed Again with Another AI Singer at Number 1 — IngaRose Follows Eddie Dalton As Fake Hit Performer AI generated song takes #1 spot on iTunes global charts iTunes Takeover by Fake AI Singer “Eddie Dalton”

Zelenskyy Announces First All-Robot Capture of Enemy Position in Ukraine

Thomas: President Zelensky just declassified a major announcement. This is a historical moment. His army captured a position using nothing but robots. On one side were human Russian soldiers. On the other were three kinds of Ukrainian-made robots.

These weren’t Terminator-looking skeleton things. In some ways they were even scarier. One was four giant wheels with a little board between them and three giant bombs. It just wheeled into a bunker and blew up.

At one point, some of these weapons systems had cameras on them, and the Russians in one of the bunkers wrote “We surrender” and held the sign up to the camera of the robot.

Jonathan: Which doesn’t care.

Thomas: The robot doesn’t, no. But the humans piloting the robot accepted their surrender.

What we’re seeing now is all-robot armies winning battles. It’s no longer humans and robots fighting together. Ukrainians are playing Call of Duty while the Russians are still sitting in trenches.

While this was just declassified, it actually happened back in the summer of 2025. We’ve now had nine months of all-robot battles, and that explains a lot of the chatter I’m seeing about rethinking war.

The primary weapon of war is no longer the rifle, it’s the drone. We’re having to rethink everything, and it’s happening very quickly. These kinds of changes often happen first in a military context, then trickle out into all of society. We’re not going back to rifles. Drones are the future. AI is the future. There’s no fighting it, but there is surrendering to it.

How do you break the morale of a machine?

Jonathan: If you’re writing in this space, this changes how morale works. You can break a man’s morale and you can break a robot’s morale, but it happens differently because robots make decisions differently. A human has to believe in what they’re doing, that it has a purpose. Someone can offer their life as long as they think it makes sense.

With Islamic extremists, the ones I fought, they thought they were trading a horrible present existence for a better future one.

With an AI or operating system, you have to change why it’s making the decisions it’s making. Introduce corruption into the communications network. It’s fairly standard in written military doctrine that if you lose communications, you immediately fall back and reestablish them. We never actually do that, but it’s written down, which means it would go into a robot’s operating system.

Marines lose communications? We’re like, “Finally, that guy’s leaving me alone. Now I can get something done.”

Thomas: Plausible deniability.

Jonathan: It can’t be a war crime if you didn’t tell me to do it. Robots are going to treat war differently. You can create a very compelling story around trying to break the morale of a system that doesn’t have emotion.

Does cheap war mean more war?

Thomas: As we look at the cultural impacts, this is going to make us feel that war is cheap. “It’s no big deal going to war because it’s just a bunch of bots. Maybe their people are going to die, but only our drones will die, and nobody cares about the drones.” That calculus can be poisonous to a society. Anytime we think war is cheap, war tends to become more common.

We’ve had an era of the United States acting as the global police on land and absolutely on water. We haven’t allowed regional wars to kick off, and that era is coming to an end. We’re potentially leaving NATO, potentially making other countries defend their own shipping. The free ride of America protecting your boats on the ocean is ending.

I suspect World War III will look entirely different than World War II. Everyone thinks it’s going to be nukes and annihilation. What I predict is a million regional conflicts all kicking off simultaneously, all the pent-up border disputes the United States didn’t allow to happen when it was the global police. None of them will have anything to do with each other. There’s a war that would happen in Eritrea, but it hasn’t.

Part of me wonders if these border disputes haven’t been delayed by Trump insisting on being the police on the ground more while simultaneously threatening to be the police on the water less.

Trump is kind of the last of his breed of global-police-style presidents. Eventually we’re going to get tired of being the cops for the world and protecting a Europe that won’t protect itself. Very likely World War III will not involve the United States at all. That’s what will make it World War III. It’ll be everybody else fighting for once and not us, as we see the European wars kick off as well.

Haven’t we seen this before?

Jonathan: The same conversation was had when the longbow was invented. War was the province of the knights. There were spearmen, but they weren’t worth much. It was the knights who did the actual work. Several philosopher-knights lamented the change, saying war had remained controlled because nobles had the honor and tradition to keep it localized.

But when it became a matter of lining up enough weapons to take down knights, they said war would become cheap, and if war was cheap, everyone would do it.

Thomas: It also shifted the balance of power. France was the home of global chivalry. They had the Norman heavy cavalry, which had been basically undefeatable for hundreds of years. They had this chivalry code, and the Arthurian myths were very popular in France at the time.

Suddenly, a rude, unsophisticated peasant from poor, backwater England, where their own leadership didn’t even speak their language, these poor peasants were able to kill sophisticated knights at a distance and at scale. It was a complete changing of the world, and it took a hundred years. The Hundred Years’ War lasted for a long time.

Sources:
Every New Ukrainian Development Shortens the Distance to Peace – Address by the President on the Arms Makers’ Day
In an Address on Arms Makers’ Day, the President Showcased Over 50 Types of Ukrainian Weapons
Robots captured Russian army positions for first time in history, Zelenskyy says
First victory for the battle brigade run by robots alone
Zelenskyy announces ‘the future is here’ after war’s first all-robot capture

Is Anthropic Nerfing Claude?

Thomas: Yesterday Anthropic released a new Claude Opus 4.7 model, and a lot of people are saying it’s worse.

Jonathan: According to VentureBeat, Anthropic launched Opus 4.7 yesterday. The company positioned the update as a clear step forward in coding, agentic workflows, and instruction following. But independent tests and early user reports are telling a different story. Some developers have found the new model worse than Opus 4.6 on real-world tasks.

A detailed thread on X by a developer called The Smart Ape captured the buzz. The 4.7 release landed with stronger benchmarks in coding and vision but is delivering weaker real-world results whenever developers use the same loose prompting habits they relied on before.

Users who fed the new model the casual, ambiguous prompts that worked fine on 4.6 watched outputs degrade and token costs climb. Vague prompts trigger extra reasoning tokens as the model tries to reconstruct intent, burning through task budgets faster and producing rushed or incomplete results. This is the same thing that hit OpenAI a few months ago.

Why does vague prompting fail on newer models?

Thomas: There are a couple things to pull out here.

First, vague prompting was never good. This is the whole origin of my principle of paragraphs, not sentences. People who don’t know how to use AI will type a sentence or a few words and treat it like a Google search. They get awful results and conclude AI isn’t very good, when in reality they’re just vague prompting.

It’s like telling the kids “get ready to go” when the kids need to be told “find your shoes, put your shirt on, go to the potty, find your backpack.” AI needs that level of detailed instruction. It needs the context of who you are, what you want, and why you want it. The more context you give in your prompt, the better your results.

Second, some people are saying prompt engineering doesn’t matter, that it’s about long conversations. That’s garbage advice. Your best response is always going to come from your very first prompt.

The longer the conversation goes, the more the outputs degrade. All of the benchmarks and best practices around AI are built around single prompting. That’s part of why the Patron Toolbox works so well, because a lot of these best practices are baked into the tools.

Why are Anthropic’s models getting worse?

Thomas: There’s a deeper issue here. Anthropic’s models became the best about a month and a half ago. Opus 4.6 was the best when it came out. Most of the savvy people started migrating from OpenAI, which got crushed. They had too many users and not enough compute to keep their models performing well. People found that OpenAI just wasn’t as good as it used to be.

I still think the best OpenAI model was ChatGPT 4. It was really expensive for them to run, and they got rid of it quickly. Every model since, 4o, 5, 5.1, 5.2, they’re all worse. They could make them better, but they literally don’t have the servers.

Everyone says there’s an AI bubble, but the supply of compute is not enough for current demand, much less future demand. As people start using AI more, they use bigger and more expensive models. We have no idea how high the demand is, but it seems far greater than supply.

Anthropic is now facing the same challenges OpenAI faced. They don’t have enough CPU and GPU cycles to go around, so the new 4.7 is stingier. You can’t set reasoning to maximum anymore. It’s dynamic now. It’s token rationing, which I had to implement on the Patron Toolbox because some people were using $50 a month worth of tokens. Unlike Anthropic, which has basically unlimited venture money, I do not.

As the models get more expensive in terms of tokens, the rationing gets stricter.

I will say the  X AI models, Grok 4.20 and Grok 4.1, are very cost-effective right now. They’re much cheaper per token than the OpenAI and Anthropic models. I expect Grok to release another really expensive model at some point, but using the right model for the right task is going to be a key strategy moving forward.

Just be aware that new doesn’t always mean better. Progress is not a straight line. Sometimes things get worse. There’s nothing new under the sun. The cycles turn ever onward.

Sources: X Thread by The Smart Ape on Opus 4.7 Prompt Changes Anthropic Official Announcement: Claude Opus 4.7 VentureBeat: Anthropic Releases Claude Opus 4.7 AOL/Finance: Anthropic Faces Backlash Over Performance and Compute Speculation Business Insider: OpenAI CFO on Compute Crunch Trade-Offs (February 2026)

Sources: Is Anthropic ‘nerfing’ Claude? Users increasingly report performance degradation Claude Opus 4.6 Is Getting Worse, and Anthropic Isn’t Saying Why OpenAI CFO Says Compute Crunch Is Forcing Tough Trade-Offs Exponential View: The labs are rationing Anthropic Official Opus 4.7 Announcement

Zeitgeist

What Do Americans Consider Immoral?

Pew Research Center Survey Exposes Deep Divides in American Views on Morality

Thomas: A Pew Research study just came out about what Americans view as moral and immoral. This matters for authors because every novel has a moral system. Actions have consequences in your fiction, and what those consequences are is governed by your moral framework as the author. How the story ends is governed by your moral framework. Evaluating that framework is critical.

This is part of why people are fatigued by woke storytelling. They’re fatigued by the woke moral system. They do pattern recognition and say, “I don’t like stories with these elements.” It’s not that they dislike female protagonists. It’s that stories with female protagonists recently tend to carry a woke moral framework that audiences are rejecting.

Let’s go through this Pew Research study. The survey gave respondents three options for each topic: morally wrong, morally acceptable, or not a moral issue.

Married people having an affair

• 90% said morally wrong, the highest of any topic surveyed.
• This is the only issue where Americans have near-universal moral agreement.

Jonathan: That’s the only shared moral anchor point. No matter which side of the spectrum you’re on, loyalty is the one thing everyone agrees on.

Eating meat

• 4% said morally wrong, the lowest of any topic surveyed.
• 41% said morally acceptable.
• 54% said it’s not a moral issue.

Jonathan: You wouldn’t think that based on the amount of entertainment and propaganda covering this. You’d expect the “morally wrong” number to be much higher.

Thomas: It’s because vegans are so loud about being vegan. Every vegan will tell you they’re vegan. That 4% probably lines up close to the actual number of Americans who are vegan.

Viewing pornography

• 52% said morally wrong, the second highest after affairs.
• 15% said morally acceptable.
• 32% said it’s not a moral issue.

Thomas: The number of people who say something isn’t a moral issue at all is its own story here. “My morality doesn’t speak to this” suggests a very limited moral framework.

Jonathan: That usually means they want to keep something out of the moral sphere so it can’t be judged or spoken against. This tracks with the “you do you” epidemic from 2010 to 2020.

Thomas: Very Gen X worldview. I’m okay, you’re okay, I’m not going to judge you.

Jonathan: “As long as it doesn’t affect me, I don’t care.” Well, now it’s affecting you.

Using contraceptives

• 8% said morally wrong, the second lowest.
• 40% said morally acceptable.
• 51% said it’s not a moral issue.

Thomas: In the olden days, contraceptives were very controversial, then almost everyone accepted them. But we’re starting to see the impacts as the birth rate goes to unsustainable levels. Everyone using contraceptives means grandma starving because there’s not enough money in the pension. We’re not there yet, but European countries are going to see it as they start paying for their own defense again and making big cuts to pensions. I’m anticipating a movement as people realize that choosing not to have children affects the economy and affects everyone else too.

Having an abortion

• 47% said morally wrong.
• 21% said morally acceptable.
• 31% said it’s not a moral issue.

Jonathan: The 47% tracks with voting patterns and is expected. But 31% saying it’s not a moral issue is really interesting.

Homosexuality

• 39% said morally wrong.
• 23% said morally acceptable.
• 37% said it’s not a moral issue.

Jonathan: I would’ve expected the morally acceptable number to be higher.

Thomas: I wonder if we’ll see movement on this. There’s a viral post from just yesterday of two gay dads with a surrogate baby, and the baby is saying “mama.” They’re laughing and saying, “No, there’s no mama, there’s daddy or pop.” The baby starts crying. The moral outrage of “you have stolen this baby from its mother” is new. Five years ago it was “you do you, what happens in your bedroom, I don’t care.” Well, this baby cares. This baby doesn’t have a mother and never will.

A lot of these numbers aren’t broken out by generation, but I suspect the large “not a moral issue” responses are heavily weighted toward Gen X, who seem to have the narrowest morality.

Doctor-assisted suicide

• 35% said morally wrong.
• 34% said morally acceptable.
• 30% said it’s not a moral issue.

Jonathan: This has the closest margin of any topic. The morally acceptable number being higher than “not a moral issue” is really surprising.

Thomas: Other than affairs, this is the topic where the fewest people say it’s not a moral issue. Death seems to be a morally charged event. It’s hard to say somebody dying isn’t moral. The moral arguments are strong on both sides. One side says killing people is wrong unless they’ve forfeited their right to life through capital crime. The other says people have a right to die and forcing them to suffer is immoral. Both sides agree this is a moral issue. They just disagree on which direction.

This is similar to the old Democratic framing on abortion. The nineties argument was “safe, legal, and rare,” which acknowledged it wasn’t morally acceptable but argued it should be legal. Now you also have the “shout your abortion” faction that’s proud of it.

Death penalty

• 34% said morally wrong.
• 38% said morally acceptable.
• 26% said it’s not a moral issue.

Thomas: Other than affairs, this has the lowest “not a moral issue” score at 26%. Again, end-of-life issues tend to carry moral weight.

Gambling

• 29% said morally wrong.
• 20% said morally acceptable.
• 50% said it’s not a moral issue.

Jonathan: The 50% are the ones doing it online and telling themselves they’re in control. The 29% don’t do it. The 20% don’t do it either.

Thomas: If you have a Christian worldview, there are two matrices for these kinds of questions: evil versus not evil, and foolish versus not foolish. Some people would say gambling isn’t a sin, but it is a foolish act. It’s not morally charged, it’s wisdom-charged.

Spanking children

• 23% said morally wrong.
• 36% said morally acceptable.
• 39% said it’s not a moral issue.

Jonathan: That number was way higher 15 years ago. I think people started seeing kids who hadn’t been spanked and said, “Nope, let’s bring it back.”

Thomas: I think the anti-spanking crowd is loud, like the vegans. But there’s also truth to the fact that a lack of discipline harms a child too. People see screen-zombie kids who are so poorly self-disciplined they need a screen at all times to not embarrass their parents and wonder if that’s better.

Jonathan: Why would 39% think spanking isn’t a moral issue? Everyone should have an opinion on this.

Thomas: It’s like asking if war is morally acceptable. Either it is or it isn’t. But people don’t really choose war. Every president who’s run for office has run on an anti-war platform, and every single one has initiated conflict. Obama is droning people. Trump is droning people. They’re all droning people.

Using marijuana

• 23% said morally wrong.
• 24% said morally acceptable.
• 52% said it’s not a moral issue.

Jonathan: That’s my experience talking to people about marijuana.

Thomas: This also falls into the wisdom-folly matrix. There’s no prohibition against consuming marijuana in the Bible, which is where I get my moral framework. You can make a strong case that marijuana is included in the blanket of foods God declared clean. But something being religiously clean does not mean consuming it is wise.

Getting a divorce

• 23% said morally wrong.
• 31% said morally acceptable.
• 45% said it’s not a moral issue.

Jonathan: None of these numbers surprise me. The nineties preached hard in favor of divorce. It became a feminist thing for women to free themselves from abusive relationships, and then the definition of “abusive” loosened to include emotionally abusive. 23% morally wrong is exactly what I’d expect.

Thomas: There seems to be a big anti-divorce push in some circles, but I think the people morally opposed to divorce are just really loud about it, like the vegans.

Being extremely rich

• 18% said morally wrong.
• 18% said morally acceptable.
• 63% said it’s not a moral issue, the highest of any topic surveyed.

Jonathan: I track that. An even 18-18 split on either side, and 63% don’t care at all. Why would this be a moral issue? Everyone wants to be rich. Am I going to be a bad person when I get there?

Drinking alcohol

• 16% said morally wrong.
• 29% said morally acceptable.
• 55% said it’s not a moral issue.

Thomas: This is really interesting because of the split among Christians. If you’re Catholic, drinking alcohol is morally acceptable. Consuming wine at mass is a core religious practice. If you’re Baptist, you don’t drink alcohol for communion, you drink grape juice, and alcohol is forbidden.

Jonathan: You’re framing independent Baptists as what they were 25 years ago. The question says “drinking alcohol,” not “alcoholism.” Most of Christianity has grown past the tea-totaling era of the seventies and eighties. That 16% are probably holdovers from that period.

Thomas: One zeitgeist observation on alcohol: overall use in the United States has been declining since its peak in the 1840s. The decline has really accelerated in the last 10 years because Gen Z doesn’t drink alcohol at all. If you’re staying home on a screen all day, you’re not out drinking. The alcohol industry is hurting. If you’re making an 18-year bourbon, you put that bourbon in a barrel 18 years ago when millennials were still drinking, and now you’re breaking it open while Gen Z doesn’t care.

If you listen to Gen Z articulate why they don’t drink, they don’t frame it morally. They frame it on the health matrix. Alcohol is bad for you.

Part of me wonders if there’s been a change in our gut biome. I have a theory that people process alcohol more poorly than their ancestors did, partly because of all the antibiotics we consume. You read historical accounts of how much Winston Churchill drank. He was powered on cigars, hard alcohol, and hatred of Nazis. That was his diet for the whole war. I watched a YouTube video of guys trying to recreate Churchill’s daily drinking schedule. They had good accounts of how much he drank and when. They followed it exactly, and they were completely non-functional. They were not ready to run a country and take down the Nazis.

In vitro fertilization

• 9% said morally wrong.
• 42% said morally acceptable.
• 47% said it’s not a moral issue.

Thomas: I’m expecting this one to move quite a bit as people learn more about it.

Jonathan: Nobody knows anything about it right now.

Thomas: People don’t understand how IVF destroys a bunch of fertilized embryos in the process. Many pro-lifers would consider those babies. Even the babies that go to term often have long-term health ramifications. I think that’s going to move the numbers, but right now only 9% think it’s morally wrong.

Jonathan: That looks like propaganda data to me. Most of what’s being put out about IVF right now is positive. People are either for it or against it because they were told to be. I think the “not a moral issue” number is going to shrink and we’ll see more distribution on either side.

Thomas: My guess is most people don’t even know what IVF is. The safest answer is “it’s not a moral issue.”

What’s the big takeaway for authors?

Jonathan: When I look at these numbers, the takeaway is that a third to a half of American society doesn’t want anything to be a moral issue. That’s disturbing. They want to remove things from the moral binary.

Thomas: You need to know who you’re writing for when you’re structuring your novel and setting up consequences. Knowing your reader’s moral framework is really important.

Jonathan: You need to know how your reader is going to react to the moral issues you’re posing. If they don’t think pornography is a problem, you can’t make that the climax that breaks your family or your romance beat. Cheating always works.

Why does worldview in children’s books matter?

Thomas: There’s a children’s book my dad read as a kid. He read it to me and I’m reading it to my children. One day I realized this book from the 1950s has a terrible moral framework. It’s called Little Black, one of those pony stories. A boy loves his pony Little Black, but then he gets Big Red, a big horse that can do tricks Little Black can’t. Little Black gets sad, stops eating, and runs away.

The boy chases Little Black on Big Red across a frozen lake. The ice cracks, Big Red falls in, the boy falls in and starts drowning. He calls for Little Black, who comes back and rescues him because he’s light enough to walk on the ice. Everyone celebrates Little Black.

I looked at this and thought, this entire problem was created by Little Black’s envy. If Little Black had been a little more thankful, none of this would have happened. He helped solve the problem he himself created. That is not a thing of moral celebration. Big Red did nothing wrong. It’s not his fault he’s bigger and stronger. He’s framed as the villain because he’s competent, and Little Black is framed as virtuous because he’s sad. This is the moral framework of baby boomers. This book poisoned an entire generation into despising competence as morally bad.

Jonathan: The one that connects with it is Rainbow Fish, the book about the fish with shiny scales. Everyone wanted his scales, he said no, and everyone treated him like the bad guy. At the end he gave everyone one of his scales, and that’s what made him a good person. It was just socialism. You’re a bad fish for having pretty rainbow scales that you won’t share. Just because you have something someone else doesn’t have doesn’t mean you’re morally required to give it to them.

Thomas: Because the morality was embedded in the story itself, no sermon was ever given. No one detected the moral message, but they still internalized it. That’s the power. You can embed a moral framework into your stories. In fact, you are embedding one whether you mean to or not, and your readers are internalizing it.

You owe it to your readers to embed a good moral framework. I have an entire episode on the Christian Publishing Show about why the Chronicles of Narnia have been so enduringly popular. It’s because the moral framework works. The children are vicious, nasty, and evil, and they go through a repentance arc in every story. They’re not good and virtuous, but they are moral. The story is moral even though the children are bad.

What we see now are stories where the characters are good but the actual moral framework structured by the author is bad. The characters are too good to be believable and the consequences aren’t real enough. Real consequences for real actions can help your readers avoid real evil. If you look at the old fairytales, the consequences are brutal, and they were brutal for a purpose.

Sources:
What Do Americans Consider Immoral? | Pew Research Center
Pew Research Center X Post with Chart and Full Report Link

Watch on YouTube