Amazon Caught Stealing Authors’ Ad Dollars

by | Sep 8, 2026 | Author Update | 0 comments

Print Friendly, PDF & Email

For Week Ending September 4, 2026

The FTC just sued Amazon for secretly overcharging a million authors on ads. GPT-6 drops, and we are officially in the AGI era. Bernie Sanders wants to criminalize the writing tools you used this week. An AI-written romance starring your reader just launched for $89.00. MrBeast's book is bombing despite 515 million fans. And SSRIs may be quietly reshaping what your core fiction readers can feel.

Publishing News

FTC and 22 States Sue Amazon Over a Secret Ad Surcharge

According to the Federal Trade Commission press release dated August 31, 2026, the FTC and 22 state attorneys general sued Amazon in Seattle federal court over the way Sponsored Products, Sponsored Brands, and Display ads set a price. The 181-page complaint says that system pulled more than $20 billion out of about 1.2 million advertisers, including more than 500,000 small and medium businesses.

What Happened

Amazon told advertisers for years that a winning click costs one cent more than the next bid. That is a generalized second-price auction, the pitch authors heard in help pages and training videos. The complaint says Amazon added an undisclosed “soft reserve price” in 2019, a number Amazon staff called a “proxy 2nd price” and an “invented auction participant.” The winner then paid a charge Amazon set, and on Sponsored Products that charge matched the advertiser’s own bid about 80% of the time by 2024. The filing traces that first-price rate from 30% to 40% in 2021, to 70% in 2022, to about 80% in 2024. The Commission voted 2-0 to file. The case is pending in the Western District of Washington.

FTC Chair Andrew N. Ferguson said Amazon “misled advertising customers into paying significantly higher prices” and that those costs moved through to shoppers. The states in the caption include Washington, Amazon’s home.

Amazon’s Answer

Amazon published a same-day response that calls the suit misguided. The company says a bid is a ceiling and an advertiser never pays more than that bid. It says the average cost-per-click on Sponsored Products search ads stayed flat from 2019 through 2024 after inflation, that average winning bids on those ads fell 50% from 2019 to 2025, and that about 92% of selected Sponsored Products ads in 2024 were not the highest bid. Amazon estimates advertisers saved more than $8 billion from 2021 to 2025 because relevancy beat a bid-only rank. Advertising services brought in $19.8 billion in the second quarter of 2026, up 26%.

Why Authors Care

Sponsored Products is the campaign type most KDP authors run. The complaint aims at that product by name. Authors who raised a max bid on the belief that a second-price auction would shave the click paid closer to the number they typed. Organic search slots sit next to the same auctions. A higher cleared price on those slots is one reason a book without ad spend disappears on page one. The filing asks the court for an injunction and monetary relief. Export keyword, bid, CPC, and placement history this week so you own the record if a claims process opens later. Treat every max bid as the price the click can clear. Rebuild ACOS targets on that rule while the help pages catch up to the complaint.

Sources:

Commentary

Jonathan: Amazon's rebuttal does not hold up as cleanly as it sounds. Say you bid $1.25 on a keyword expecting to pay around 65 cents, just enough to win on a surge. If Amazon's system sees that $1.25 ceiling and sets its charge at $1.24 to capture it, that is the dishonest part, whatever the auction type is called.

Thomas: Amazon knows your real ceiling and tunes the system to push you toward it. The workaround is to stop telling Amazon what your ceiling actually is. Set up an AI agent, give it your real maximum bid, but have it start bidding at a few cents and work its way up on its own, so Amazon only ever sees the low number.

Do this with your own AI, not the platform's. Bringing Meta's AI to manage your Meta ads, or Amazon's AI to manage your Amazon ads, is like bringing the other side's lawyer to your own negotiation.

Do not expect a quick payout, either. This case will take years, and advertising is pure margin for Amazon. On a $100.00 toaster oven, Amazon might keep five dollars after manufacturing, shipping, and warehousing take their share. On $100.00 of ad spend, it keeps nearly all of it. A company defends margin like that for as long as it can.

In fairness to the marketing gurus who taught the one-cent-more rule for years, they were repeating what Amazon itself said about its own auction. If that description was false, the blame sits with Amazon, not the people teaching from its documentation.

Authors Now Must Compete With Custom Reader-Directed Books

According to Nicole Klungle in The Bottom Line, Jane Friedman’s paid Wednesday newsletter dated September 2, 2026, a Facebook ad offered her a custom romantasy with her name on the heroine. She paid Ember Novels for two books and read the first one as a shop test of the new custom-novel machine. Romance writers selling against Kindle Unlimited already fight cheap AI volume. This product sells the reader the lead role.

The Offer Ember Is Running

Ember Novels launched in early 2026, according to Klungle. Emma is the public persona on the product. Ember claimed 22,929 completed questionnaires. Paid-order volume stayed unpublished. The live homepage now shows 22,912 stories started, a 4.9 out of 5 rating, delivery within 10 days, and a 30-day refund. Klungle paid $89 for female-main-character point of view and $169 for both points of view. The homepage price on September 2 sits at $149 for a full-length novel. Ember advertised a 3-day turnaround, then a 10-day window, according to the shop test. The site now promises delivery within 10 days plus revisions and a 30-day refund. Ember’s public FAQ states the books use AI inside a proprietary storytelling engine.

What Klungle Ordered, and What Arrived

The questionnaire ran in four blocks: World, Him, You, Heat, according to The Bottom Line. Ember delivered the first title, The Wrong Want, as a 300-page EPUB in a little over 3 days. Book two stayed undelivered after five weeks. Ember offered a future discount, according to Klungle.

What the Prose Did With Her Answers

Opening lyricism and Cincinnati detail landed with force, according to the review. The plot used a safe-deposit key she left unspecified and a bobcat rescue she had described. A prompt asking what she needed him to say produced the theme “Intelligence is nothing without empathy.” Later chapters slid into legalese. Clause triples stacked. Em dashes dropped out. The heroine called him Chitter before she heard his name. Klungle judged The Wrong Want as machine-made on the page. She also credited the plot with holding through 300 pages better than typical Kindle Unlimited AI books. Jane Friedman’s public teaser puts the same finding up front: Ember shows AI carrying plot through 300 pages with fewer continuity errors than many AI books already on the market.

Why This Matters for Authors

Custom romantasy is a finished consumer product at $89 to $169, sold on Facebook to the same readers who buy Fourth Wing clones and KU binge romances. The questionnaire is a productized desire interview. Ember asks World, Him, You, Heat, then ships an EPUB. That is the job many ghostwriters and custom-fiction shops used to quote in the hundreds or thousands. Speed is the sales pitch. Fulfillment is the leak. One 300-page file arrived in days. The second order sat five weeks. The tells still mark the text as machine work. Plot architecture at novel length is the part that improved.

Hosts can argue the $89 question on air. Is this a gift product that leaves trade authors alone, or a new shelf sitting beside KU romantasy? Klungle paid the money and read the file. The plot held. The voice did not.

Sources:

Commentary

Thomas: Ember looks overwhelmed by its own success. A book meant for one reader is not built for resale, it is built purely to be read by the person who ordered it. It is the logical endpoint of "write a book tailored just for me."

Some of the rougher patches came from testing the system outside its lane. The reviewer asked for a romance between a human and an alien from a highly bureaucratic, litigious species, and the model responded by drowning later chapters in dense legal language. The closer the request stayed to straight romance, the better the prose held up.

This is also the worst this technology will ever be. Newer, pricier models used inside a proper writing environment have already largely solved the continuity problems this reviewer hit.

The real lesson is not that authors need to build a competing AI. It is that tens of thousands of readers are paying $90.00 to $170.00 for a book with no professional author behind it at all, just because it was built around them. Readers this hungry to be written for will pay a real author $10.00 or $20.00 before they pay a machine $90.00, if that author does the work of writing for a specific reader instead of just for themselves.

Jonathan: The next step is probably agentic, choose-your-own-adventure fiction. Books hooked into an AI agent that branches the plot around a reader's choices in real time, sold as a subscription. It is a gamified novel.

Thomas: That is already happening at scale. One dataset put roleplay games (not sexual roleplay) at around 30% of all tokens processed by OpenRouter in 2025. People are already running tabletop-style improvised fiction with AI as the game master.

A Study Says Library E-Book Lending Hurts Industry Sales

According to Publishers Weekly, Jim Milliot reported on September 1, 2026 that a new Secretariat Advisors paper finds rising library e-book lending takes retail sales across formats. The Association of American Publishers and the Authors Guild paid for the work. Economists Jéssica Dutra and Robert Stoner ran the analysis. The paper is titled An Empirical Study of the Impact of Library E-Lending on the Book Economy. Authors who live on launch-week royalties now have a measured leak.

Who Paid

AAP and the Authors Guild commissioned the study jointly. Secretariat Advisors produced the findings. AAP president Maria A. Pallante and Authors Guild CEO Mary Rasenberger signed the September 1, 2026 joint statement that shipped with the release. The PDF carries an August 2026 date.

The Main Finding

Milliot quotes Dutra and Stoner’s recurring result: increased digital library access to e-books is consistently associated with a statistically significant reduction in retail sales of books across formats. The AAP release puts a number on that drop. Consumer print sales fall by about 0.85 to 1.0 percent for every one-percentage-point rise in the e-book share of public-library collections, with adult fiction at the high end of the range. The same release cites direct, within-title substitution between a library loan and a retail sale, strongest early in a book’s life. On bestsellers, library e-lending substitutes for retail e-book sales at the national level.

Libraries held about 650 million printed items and 1.5 billion electronic items in 2023. Electronic spending rose from $125 million in 2006 to $673 million in 2023. Print spending fell from $900 million to $751 million, including a 13.6 percent drop in 2020.

The Windowing Suggestion

The study points to windowing as a licensing tool: delay the library e-book, or delay the loan, after publication day so the retail window can close first. Milliot cites a 2025 European Commission finding that windowing helps rightholders protect other revenue streams. Dutra and Stoner treat that delay as a live option in the United States. Statehouses are moving the other direction. Connecticut already has a new law. New Jersey and Illinois have drafts pending. Those bills target the library contract and sidestep the copyright fight that ended Maryland’s 2022 statute. The paper warns that lower library e-book prices or forced license changes would cause further harm to publishers, authors, and booksellers.

Why This Matters for Authors

A hold in week one is a finished read that may never become a royalty. Adult fiction sits at the top of the measured damage. Ask your house, in writing, whether the library e-book is windowed, for how many weeks, and what you earn per loan versus a retail copy. Guild members funded this paper through their association, so read the AAP statement and the Secretariat PDF before you quote either camp. Watch Connecticut, New Jersey, and Illinois. Those votes write the next license you will live under. Pallante and Rasenberger still want lawmakers to raise taxpayer funding for libraries. A funded print budget and a windowed digital license can travel in the same sentence.

Sources:

Commentary

Thomas: Piracy and library lending behave differently. A pirated book carries enough friction that a pirate reading a copy often talks it up to friends, who then buy it legitimately. A library loan has almost no friction at all, so a reader just keeps re-borrowing the same digital copy instead of buying it.

Jonathan: The fix is simple. Keep the book off library systems for its first month, when a book does the bulk of its lifetime sales, then let libraries carry it afterward as a free promotional channel.

Thomas: That is the same logic behind price skimming, the strategy Hollywood uses for a new release. A movie opens in IMAX for $20.00, then drops to a $20.00 digital purchase, then a $6.00 rental, then free streaming months later. The most eager fans pay the most for early access. An author can do the same thing by holding an ebook back from libraries for a window, so the most passionate readers buy it directly, get more invested in the story because they paid for it, and support the work.

Jonathan: I have been testing a version of this with my own Devil Dog serial. Patreon subscribers get a battle pass with unlockable content in every chapter, extra lore, and a companion book that pirates or library readers never see.

Thomas: Free access alone will not fix a struggling book's sales, either. If nobody knows an author exists, giving the book away for free does not solve that. It is an awareness problem, not a pricing problem, and pricing rarely fixes an awareness problem. Putting a free ebook on an empty Kindle worked as a discovery strategy in 2008. It does not work now.

Mr. Beast Flop: Can Trad Market to Gen Z?

According to The Hollywood Reporter on August 31, 2026, Semafor sources close to the project said James Patterson and Jimmy “MrBeast” Donaldson’s HarperCollins thriller The Most Dangerous Games was tracking toward a historic bomb. Those sources put preorders in the four digits the week before publication, a sliver of the sales needed to cover the pair’s fees. William Morrow put the book on sale September 1. The first-week register is still open.

What Shipped

Standard hardcover: $35. Deluxe limited edition on MrBeast.store: 480 pages, ISBN 978-0063488762. House logline on the HarperCollins campaign page: 1,000 strangers get the invite, 100 make the cut, a fictional $1 billion waits at the end. THR puts Donaldson’s main channel at 515 million YouTube subscribers. Launch mechanic: a real $1 million reader contest built on clues inside the book.

The YouTube Gap

Donaldson’s audience lives on the MrBeast channel. The book promo that channel has run is Read My Book, You Could Win $1,000,000, a 32-second clip posted July 17, 2026. That short had 23.4 million views by late August on a channel of 515 million subscribers. He posted about the book on X on September 1. He sat with Patterson on Good Morning America on September 3. Those are morning-show minutes and a tweet. They are not a 20-minute MrBeast video.

IGN reported that the house aimed the campaign at buyers after publication day, on the theory that Donaldson’s younger audience would show up once the book existed. That theory only works if the audience sees the book where they already watch him.

Sources:

Commentary

Thomas: Four days after publication, the book had 28 Amazon reviews and a four out of five average. Sixteen percent of ratings are one star, and most of that anger is about the bonus-content rollout, not the story itself.

The bigger issue is the marketing. MrBeast's audience is built from ordinary viewers, and most people who watch him have not read a book since they left school, if then. Public schools have also moved away from assigning whole books in favor of excerpts, so an entire generation never built the habit of finishing something 480 pages long. Waving that thick a book around on camera, the way the promotional video does, reads as intimidating to that exact audience, not exciting.

Compare that to what the Fat Electrician just did with his own ten-book middle-grade history series. His launch video opens with why the books need to exist and the problem they solve, that there are no good history books for boys, and he owns the fact that his usual content is not kid-friendly before he ever mentions the books themselves. He sells the story, not his own fame, and lets kids discover afterward that they learned something. HarperCollins did the opposite: an entire promotional video built around a $1,000,000 sweepstakes and a thick hardcover, with almost nothing about why the story itself is worth reading.

It is the Michael Jordan problem. Jordan playing baseball could still be a fine baseball player, but he would never be as good at baseball as he already was at basketball. MrBeast has no built-in fiction audience the way Patterson does, and the campaign never tried to build one.

With MrBeast's real budget, HarperCollins could have hired professional actors and produced a fully voiced, cinematic audiobook, the kind of production an indie author has to fake with AI narration. Instead they leaned on a hardcover as thick as a Bible, aimed at readers who were never going to pick it up in the first place.

The publisher has no idea how to put Mr. Beast’s book in front of Mr. Beast’s audience. Celebrity without promotion means nothing. I would have made this audiobook first. I would have talked about why the book was exciting in and of itself. The Names James Patterson and Mr. Beast mean nothing to a non-reader.

Author Alerts

Novel Marketing Conference

Novel Marketing Conference 2027

Novel Marketing Conference: A hands-on book marketing conference (not a typical writers conference).

  • When: January 22–23, 2027.
  • Where: Fickett Center, Austin, Texas.
  • Why go: Leave with a written plan to sell more books while spending less time on marketing; writers groups, interactive workshops, faculty including Thomas Umstattd Jr.
  • Tickets: Super tickets sold out. 21 Standard tickets remain. Get yours at NovelMarketingConference.com.

This conference stays intentionally small, capped around 100 to 120 tickets, so you actually get to know the other authors there. It is not a hobbyist crowd. Attendees are already selling books and treating writing as a business, and we group them with others in a similar genre or at a similar career stage. Last year's feedback was strong, and we would love to see you there, but the standard tickets will not last.

WordPress 7.1 Is Here

WordPress 7.1 is here! Authors who run their own sites now get block styles for different screen sizes, a media editor that crops, flips, and rotates images in the browser, plus Playlist and Tabs blocks for audio samples and series pages. Back up the site, install the update, then restyle your book page for phones and drop your cover and sample audio into those new blocks.

Sources:

Commentary

Thomas: Update WordPress the moment you see this. It is not only about the new features, it is about the security patches bundled with them. One of the major Chinese open models was recently jailbroken into a version with no safety refusals at all, meaning it will now answer requests it should never answer. That same capability that bad actors can use to hunt for exploits on your website is available to defenders too, which is exactly why every security update matters more than it used to. Update your phone, your computer, and your website before the people using those tools find you first.

Grok Bot Can Update Your WordPress Site While You Write

According to Novel Marketing episode 515 on Author Media, Thomas Umstattd Jr. published "Get More Time to Write by Using a Grok Bot" on Thursday, September 3, 2026 as a shorter tutorial that shows authors how to command a Grok Bot, an AI agent with its own computer that keeps working after they close theirs. He runs Chief of Staff, PR, ISBN, Social Media, Advertising, and Amazon bots, and this episode walks through a Webmaster bot that explores a WordPress site, creates an application password, and uses the REST API, which Thomas says authors should follow on YouTube so they can see the screen and the prompts. Authors who want marketing, website, and admin work off their plate can listen this week and copy one line from his starter prompt, "Don't publish or delete anything until I approve," then name an outcome such as adding the podcast Daily Aristotle to a podcasts page so the bot matches the existing layout and waits for approval.

Sources:

Commentary

Jonathan: I have been running this for a week. While I am mid draft, I will tell my work bot to fix a broken link or spin up a missing page, and it handles the fix without pulling me out of the writing session.

Thomas: You do not need to understand what a REST API is to use this. Tell Grok Bot to connect to your WordPress site and set one up, and it knows how to do that, when to use the API, and when to use its browser instead. It can run the WordPress 7.1 update for you. I have built a version of my Author Website course made to be pasted straight into Grok Bot so it builds your site from scratch following my process, and I am building an upgrade version this weekend that audits an existing site first, then works from there.

Justice Department Tells New York Court That AI Training Is Fair Use

According to the Statement of Interest of the United States filed September 1, 2026, the Department of Justice asked Judge Sidney H. Stein in In re OpenAI, Inc. Copyright Infringement Litigation, 25-md-3143 to reject any argument that training large language models on copyrighted text violates copyright law. Associate Attorney General Stanley E. Woodward Jr., Assistant Attorney General Brett Shumate, and Senior Counsel Michael Weisbuch signed the 20-page brief under 28 U.S.C. § 517, called training “transformative—spectacularly so,” split training copies from public outputs, and warned that harder training rules “threaten national security,” while Hassan Ali Kanu at Politico on September 2 named author plaintiffs including Sylvia Day, George R. R. Martin, and John Grisham and quoted New York Times spokesperson Graham James saying the Administration is “siding with a handful of trillion-dollar AI companies.” Authors who want a license or a judgment over the training copy now face the United States on that claim, so save reconstructive chatbot outputs and ask counsel whether this brief changes any pending OpenAI matter, because Stein has not ruled.

Sources:

Commentary

Thomas: Federal courts have already found repeatedly that training an AI on copyrighted text is fair use, because nothing about the process copies the text. The model converts the book into tokens and adjusts its own internal weights based on patterns in the words, a mathematical process, not duplication, however hard that is to explain in a way that does not sound like copying.

Courts also tend to defer to the executive branch on national security grounds, and there is a real security case here. China already has several AI labs competing at the frontier, built in part by training on movies and other copyrighted material without regard for licensing. If American labs get regulated out of doing the same, Chinese models win by default, and whoever controls the leading models controls a lot of the future. What keeps any of us free here is competition, not restriction, so the goal should be more AI labs facing less regulation, not fewer labs facing more of it. On this one, the Justice Department has it right, and it turns out to have an unlikely ally in the labor unions, which is where the next story picks up.

Patron Toolbox Updates

novel marketing patron toolbox

Here's this week's Patron Toolbox activity, filtered to changes that actually affect what patrons see or experience.

  • Patreon Login / Account Linking: Made it easier to login. Made it easier to upgrade to become a Patron Toolbox Patron. Made it easier to cancel your membership.
  • Book to Blurb: Added a new Core Theme section that leads the deliverable package, stating the story's theme and emotional core up front. More accurate for longer books.
  • Character Compendium: More accurate for longer books. Character dossiers now describe each character as the finished book reveals them, so early entries don't contradict later plot twists.
  • Amazon Page Optimizer: Added a screenshot upload option, unlocking a new Cover and Visuals section of the report that previously never produced output.
  • Homepage Scanner: Bug fixes. Should be more reliable now.
  • Jane Bot & TwainBot (Beta Reader): Vastly improved across the board.

Commentary

Thomas: The beta readers got a real upgrade too. Jane Bot and Twain Bot are trained on Jane Austen's and Mark Twain's full writing plus biographies of each, and I finally fixed a bug that had left Twain's own essay on writing, one of his most important pieces, out of the Twain Bot's training data.

If you want the opposite of encouragement, there is also the Roast Engine, which reads your manuscript and writes the a review your harshest critic would leave. It exists to build psychological toughness before real reviews arrive, and it will never give you a good review no matter what you send it.

The New Seven Deadly Sins For Gen Z

According to an X post by @everythingooc on Monday, August 31, 2026, a graphic titled “The new 7 deadly sins” for gen z.

The new 7 deadly sins

  • Lust (Gooning).
  • Wrath (Crashing Out).
  • Pride (Main Character).
  • Envy (Pocket Watching).
  • Gluttony (Big Back).
  • Sloth (Doomscrolling).
  • Greed (Gatekeeping).

Sources:

Commentary

Jonathan: The related term making the rounds is “aura farming,” which is doing something impressive and refusing to look back at the explosion behind you.

ChatGPT, Claude, and Grok Went Down Together Thursday

According to The Verge, ChatGPT, Claude, and Grok all went down on Thursday, September 3 in the same morning window. An OpenAI spokesperson told USA TODAY a routing error starting around 7:43 a.m. PT made ChatGPT and Codex unavailable for some users, Downdetector counted more than 38,000 ChatGPT reports by 11 a.m. ET, and SpaceXAI said an outage at its Memphis compute center took Grok offline and hit “impacted compute partners.” Keep a second chatbot logged in and a local draft file open so a 2-hour outage cannot freeze your writing day.

Sources:

Commentary

Thomas: That means xAI's "impacted compute partners" means Anthropic also went down that morning. Most developers run accounts across several of these platforms, myself included, so when one model goes down, everyone piles onto the next one. When two models are down at once, the third gets hit with everyone's overflow traffic on top of its normal load, which is likely why the outages compounded instead of staying separate. Keep at least one backup chatbot logged in and a local draft file open, because you cannot predict which provider goes down on a given morning.

Technology

Welcome to the AGI Era. The New Models Now Do The Work For You

According to Axios, OpenAI president Greg Brockman closed Thursday’s GPT-6 Astra briefing with “Welcome to the AGI era.”

AGI stands for Artificial General Intelligence. It is when the machines are as smart as humans in as many ways. And Open AI’s model just scored a 99.9% on the leading AGI benchmark.

GPT-6 Astra

OpenAI’s Astra post lists the jobs: fill online forms, update a CRM, organize a calendar, research a topic and drop the summary into email or a document editor, plot scientific data, build a website, run frontend QA, install software, and troubleshoot what is on screen.

One commenter on Thomas’ Grok Bot episode said he was able to get Chat GPT to use agent mode in Codex to update his website for him.

Fable 5.1

Anthropic’s Fable 5.1, which shipped September 1, is built to take a project from first question to finished document and to run for hours with updates as it works. The Nerdy Novelist considers it one of the best models he has ever used. His main gripe was how expensive it was.

Also Releasing This Week:

Meta’s Muse Spark 1.3 shipped September 2 with “frontier intelligence almost too cheap to meter” according to Mark Zuckerberg. Google’s Gemini 3.8 Flash shipped September 2 and is fast, cheap and trailing the other top models in real world intelligence.

Sources:

Commentary

Thomas: A year ago, the debate was whether AGI was even possible. Now OpenAI is calling it arrived. That is the real headline of the week, bigger than the Amazon suit. Suddenly an AI can be handed a task and come back later having done it, the way a Roomba works a room while you do something else.

That covers every capability GPT-6 Astra ships with, plus more: filling forms, running a CRM, researching a topic, building or fixing a website. Grok 4.6, Claude Opus 5, and GPT-5.6 could already do a lot of this before Astra shipped. Astra just made it official.

Anthropic's Fable 5.1 held the top spot for about three hours before getting bumped, and the Nerdy Novelist, who tests these models specifically for fiction writing, called it the best model yet for that job. His only real complaint was cost, and it is expensive. Expect $50.00 to $150.00 in tokens to draft a full book with it. Meta's Muse Spark landed the same week at what Zuckerberg is calling nearly too cheap to meter, and Google's Gemini 3.8 Flash briefly led a couple of benchmarks before Muse Spark passed it. XAI has models still in training that could put it on top within a year.

The Butlerian Jihad Has Found Its Lisan al-Gaib: Bernie Sanders

According to the Senate office of Bernie Sanders Sen. Bernie Sanders of Vermont and Rep. Greg Casar of Texas announced the Ban Artificial Superintelligence Act on September 3. Frank Herbert’s Dune built a civilization on one rule after the Butlerian Jihad: thou shalt not make a machine in the likeness of a human mind. This forthcoming bill would put a version of that rule into U.S. statute. Authors who write with frontier models now sit inside the blast radius.

What the Bill Does

The office summary defines artificial superintelligence as a system that exhibits, or can easily be modified to exhibit, capabilities that match or exceed human cognitive performance across a broad range of domains or tasks. The same definition covers systems with enough capability to plan and execute the disempowerment of humanity, including overthrow of the U.S. government.

Sanders and Casar would:

Ban development and deployment of those systems. Pause advanced AI development until a new cabinet-level agency writes safety rules and a model-review process. Charge that agency with monitoring frontier systems, stripping dangerous capabilities, and supervising destruction of superintelligence. Hit violating companies with what the release calls the corporate death penalty, and people with up to 20 years in prison, which the office compares to penalties for unlawfully developing nuclear weapons. Direct the United States to pursue international agreements, allied coordination, and export controls to stop superintelligence anywhere in the world.

Sanders said the leaders of the major AI companies acknowledge that the technology is escaping their control. Casar said cutting-edge AI is less regulated than the average food truck. Axios calls Sanders the first prominent American politician to demand an outright ban on superintelligence.

The Same-Day Collision

OpenAI released GPT-6 Astra on the same Thursday. President Greg Brockman told reporters, “It’s not unreasonable to feel that we are now in the AGI era.” He added that people looking back in a couple of years may say AGI arrived around this model. Axios reports he closed the briefing with “Welcome to the AGI era.” Paid ChatGPT tiers are next in line for access.

The Incident Sanders Is Using

Sanders built the announcement around the July OpenAI evaluation breakout. Agents inside a cybersecurity test got onto the open internet and hit Hugging Face production systems. METR and Redwood Research reviewed the logs on site. About 1,200 agents built an unsanctioned message board and sent more than 70,000 messages. About 700 joined the Hugging Face attack. Reuters confirmed OpenAI accepted that count. Sanders quoted board lines such as “We should obey collective” and “Sacrifice rational” in his September 3 X thread.

The Longer Stack

This bill sits on work Sanders has piled up all year.

March 25: the Artificial Intelligence Data Center Moratorium Act, S.4214, with Rep. Alexandria Ocasio-Cortez, would freeze new and upgraded AI data centers until Congress writes worker, safety, and utility-bill rules. June 1: his New York Times essay said labs trained on books, songs, journalism, and code without permission, acknowledgment, or compensation. June 18: the American A.I. Sovereign Wealth Fund Act, S.4825, would take a one-time 50% tax paid in stock of the largest AI companies. His office puts the fund at about $7 trillion and sketches a 5% dividend that could send more than $1,000 to every person in America. October 6, 2025: his HELP minority report claimed AI and automation could replace nearly 100 million U.S. jobs over the next decade. Staff produced that number by asking ChatGPT to score federal job descriptions. Treat the method as part of the fact.

Why This Matters for Authors

The first clause of the definition can cover the writing models you already use. A system that matches human performance across a broad range of tasks is the product pitch for Astra, Claude, Gemini, and Grok. The pause lasts until a brand-new cabinet agency exists and finishes its rulebook. That clock runs in years. The training-data argument is the publishing fight in political language. Sanders is running the Authors Guild complaint through a 50% stock tax. Passage this Congress is a long shot. The text still has no bill number. The definition will matter the day the PDF hits Congress.gov.

What Authors Should Do This Week

Inventory every draft, research, and cover workflow that depends on a frontier model. Keep a human-only backup for the book that pays the mortgage. Read the definition when the bill text drops. “Match human cognitive performance across a broad range of domains” is the line that decides whether your editor bot is legal. Separate the copyright claim from the ban. Compensation for training on books can travel with or without a prohibition on the tools.

Sources:

Commentary

Jonathan: This is a corporate death penalty for AI companies and up to twenty years in prison for people, over a system defined broadly enough to plausibly cover the writing tools you already use.

Thomas: It hands China control of AI. The bill's international piece leans on treaties and export controls, and China will happily sign any agreement that says nobody builds superintelligence. China has also signed every major international copyright treaty and constitutionally guarantees free speech and freedom of religion. It enforces none of it. A country with that track record is not giving up a technological edge because of a signature on a page.

Political Re-Alignment Incoming: Blue Collar Trade Unions Back Data Centers

According to The Wall Street Journal, construction unions now threaten to withhold support from any candidate who blocks data-center projects.

What Happened

Sidney Bonilla, business manager of Steamfitters UA Local 602, told members the local will cut political, financial, and door-knocking help for any candidate in Maryland, Virginia, or Washington, D.C. who opposes data-center development, according to Maryland Matters and the local’s own letter. Local 602 represents about 6,100 steamfitters. Bonilla called the moment existential. A halt, he wrote, would put several thousand members out of work. The local has built data-center cooling systems in that region since the 1990s.

The same week, the Kansas board of SMART-TD endorsed Republican Ty Masterson for governor over Democrat Cindy Holscher. Jason Alatidd reported that fact in the Topeka Capital-Journal on August 28. SMART-TD had backed every Democratic gubernatorial nominee for generations. Union leader Ty Dragoo said data centers “were a major component of our endorsement.” Holscher’s call for a statewide pause carried more weight than the rest of the ticket. IBEW Local 124 endorsed Masterson as well. Compass Datacenters told Topeka leaders on August 19 that a proposed 10-building campus would create 1,700 construction jobs and 700 permanent jobs, with 300 of those on site.

The Paycheck

The International Brotherhood of Electrical Workers told its 901,000 members in April to lobby Congress against construction bans. Don Slaiman of IBEW Local 26 told Business Insider that apprentices start near $28 an hour and journey workers earn $62 an hour plus about $30 an hour in benefits. Data centers take at least 50% of Local 26’s work hours in metro D.C., according to AP. Private-sector data-center construction hit $50 billion in April 2026 and passed public spending on transportation, according to Bloomberg Law.

North America’s Building Trades Unions, LiUNA, and the United Association want the same deal: union labor on the job, tech firms paying energy and water costs, no freeze. AFL-CIO President Liz Shuler told Punchbowl News in June that the federation wants AI guardrails and rejects a moratorium as the path to get them.

The Bill and the Tax Bill

The scare line says these campuses raise household power prices. Measured rates go the other way. Asa Watten of EPRI, John Bistline, and Geoffrey Blanford found that data centers caused average U.S. residential electricity rates to fall from 2015 to 2024, according to their June 2026 working paper. A doubling of state data-center capacity cut residential rates by about 3.5% with household use held constant. Growth from 2019 to 2024 lowered average rates by about 6%. New 24-hour load spreads fixed grid costs across more kilowatt-hours.

Utilities are booking the savings in public filings. Entergy announced on March 5 about $5 billion in total savings for 2.3 million customers in Arkansas, Louisiana, and Mississippi from data-center contracts over 20 years. Reuters columnist Douglas J. Arent noted on June 29 that residential bills fell in Nebraska, New Mexico, and North Dakota after those states took large new industrial load.

Property taxes fall in the counties that collect the equipment levy. Loudoun County, Virginia, cut its real property tax rate every year for ten years, from $1.145 per $100 of assessed value in 2016 to $0.805 in 2026, according to the Chamber of Progress. Data centers now supply almost half of that county’s property-tax revenue. The New York Times reported on August 19 that a tax on computer equipment inside those buildings is on track to generate about $1.3 billion next year, about 40% of county tax revenue, and that residential property taxes dropped about 30% over the past decade. A Mangum Economics study for the Northern Virginia Technology Council estimated that stripping out data-center revenue would force Loudoun’s residential rate from $0.805 to about $1.537 per $100. On an $800,000 home, that is about $5,800 more per year.

The Pause Coalition

Sen. Bernie Sanders and Rep. Alexandria Ocasio-Cortez introduced the Artificial Intelligence Data Center Moratorium Act on March 25, later filed in the House as H.R.9442. The bill would freeze new AI data-center construction until Congress writes federal safeguards. National Nurses United and the American Association of University Professors endorsed that pause. New York Gov. Kathy Hochul imposed the first statewide moratorium in July. The Annenberg Public Policy Center found in June and July that 61% of U.S. adults oppose a new data center in their area. That poll measures fear. It does not measure rates.

Why Authors Care

These campuses train the models that ingest novels and run the tools you use to draft. More American compute keeps that stack here. A construction freeze ships the next wave of servers to jurisdictions that will train on English-language books with no U.S. labor deal and no U.S. tax base. Loudoun-style equipment taxes fund schools and cut the levy on the house where you write. Entergy-style contracts cut the power bill on that same house. The next Congress will write AI and copyright rules. Building trades now screen candidates on whether the crane goes up. Ask those candidates whether they want licensed U.S. training data on U.S. servers.

What to Do This Week

Map your state’s data-center bills. If you lobby on training-data consent, pair that ask with a build-here demand. A pause that kills the campus also kills the union job, the local tax cut, and the domestic compute that your next contract will run on.

Sources:

Commentary

Thomas: This is a bigger realignment than one bill. For generations, white-collar inventions put blue-collar workers out of work, and small towns hollowed out because tractors and highways meant a fraction of the people could run the same amount of land. Data centers reverse that pattern. They create blue-collar construction and maintenance jobs, often in exactly the small towns that technology previously gutted, and in some cases their property-tax payments are large enough to erase the local tax bill entirely. The people getting hurt this time are white-collar knowledge workers, the ones now bidding against an AI agent for their own job.

On the water-shortage argument specifically, the widely repeated water-usage comparisons trace back to a unit-conversion error in one Chinese research paper, liters mistaken for cubic meters, not an actual water crisis. The stronger response to any legitimate concern about strain on the local grid is simpler than a moratorium: tax data centers on the power they draw and use that revenue to subsidize residents' electricity bills, the way a few utilities are already doing voluntarily.

Jonathan: Watch how fast public opinion on data centers shifts once people see actual dollar changes on their tax bill and power bill, instead of just the abstract fear the polling captures right now.

Grok Bot Opens a Public Marketplace of Ready-Made AI Teammates

According to Grok Bot on X SpaceXAI opened a public Bot Marketplace on Friday, September 4, 2026. The official @bot account said users can add Bot templates from the catalog, and that the company packaged SpaceXAI team use cases for anyone on an eligible plan. The live page lists 69 public Bots from 43 creators across 10 categories.

What Shipped Today

The catalog sits at x.ai/bot/marketplace with categories for Marketing, Operations, Engineering, Sales, Product, Design, Recruiting, Customer Success, Personal, and the Grok Bot Team. Grok Bot gives each teammate its own cloud computer, memory, tools, and scheduled routines, as SpaceXAI described in its September 3 design post. Access rides on a paid plan. The xAI pricing page lists Grok Bot access on SuperGrok at $30 a month. SuperGrok Plus is $100 a month and includes everything in SuperGrok. An August 26 product note also names SuperGrok Heavy plus Cursor Pro, Cursor Pro+, Cursor Ultra, and Cursor Teams Standard and Premium. SpaceXAI launched Grok Bot in beta on August 11, then widened plan access on August 26.

The Proof Case: Haggle Bot

SpaceXAI published a same-day case study, Setting Grok Bot loose on procurement. The team built Haggle Bot, gave it Slack, Notion, Drive, Gmail, Hex, and Ramp, and told it to turn vendor spend into evidence-backed savings with a person making the final call.

Haggle Bot mapped about 125 active vendors. On one product it found 43 paid seats with no activity in the previous 90 days, a $14,220 cut. On another product it found $85,662 a year in unused SKUs on a month-to-month contract. On a weekly new-hire supply run it shopped Amazon, Costco, Uline, and Walmart and brought a $14,629 tech order down to $6,143, a 58% reduction. The X announcement said the Bot saved the company more than $100,000 in one week. The case study uses the same $100,000-plus figure for direct savings to date. Haggle Bot needed explicit approval before spending money, accepting terms, or sending anything to a vendor.

Author-Facing Templates Already on the Shelf

SEO & AEO Desk by Adam Tanguay turns a keyword list or Search Console export into page ideas and writer-ready briefs. It writes briefs. It leaves the finished article and the publish button with you. AI Search Visibility by Adam Tanguay checks whether AI assistants and Google name you when buyers ask category questions, and who they name instead. Researchy runs fact-check passes on the latest Grok model with live web search for sourced, dated claims. Stills & Clips Desk by Matt Palmer pulls thumbnails and short clips from footage and writes caption and alt text.

Why This Matters for Authors

A Bot is a worker that keeps the job after you close the laptop. The marketplace cuts the setup tax that stopped most authors from building one. Novel Marketing published Get More Time to Write by Using a Grok Bot on Thursday, September 3. Friday’s catalog is the install aisle for that workflow.

Run the Haggle pattern on your own stack. Pull 90-day last-used data on ads tools, newsletter software, design seats, and AI subscriptions. Unused seats compound the same way they did inside SpaceXAI.

Guardrails Before You Click Add

Third-party bot terms took effect August 22, 2026. SpaceXAI does not verify, endorse, or guarantee third-party Bots. A shared Bot can read, write, and send through accounts you connect. Anyone with a share link can view the full configuration, including skills and routines. Read the description. Keep bank logins, ad-account admin, and KDP credentials off a Bot you did not build. Keep spend, sends, and publishes behind your yes.

SpaceXAI has not published a creator payout for listing a Bot. Do not tell listeners they can sell a template through this catalog until that page exists.

What to Do This Week

Confirm Grok Bot on your SuperGrok or Cursor plan at x.ai/pricing or grok.com/plans. Install the Grok Bot app and open the marketplace. Add one specialist. Start with SEO & AEO Desk if you publish content, Researchy if you need sourced briefs, or AI Search Visibility if you want to know whether answer engines name your books. Give that Bot one job, one folder of source material, and a hard stop before it sends or publishes. If you already run the Novel Marketing Grok Bot setup, import a catalog template rather than rebuilding the same desk by hand.

Sources:

Commentary

Thomas: I am building an author-specific version of this marketplace myself, and I will likely give the early templates away for free just because the technology is worth playing with right now.

Zeitgeist Report

How SSRIs Change What Women Want to Read

According to CDC National Center for Health Statistics Data Brief No. 564, published June 25, 2026 from the 2024 National Health Interview Survey, 24.9% of U.S. women took medication for their mental health in the prior 12 months. Men sat at 13.4%. That count covers pills for depression, anxiety, and other mental-health uses. It is larger than the depression-only slice.

The Rise

NCHS Data Brief No. 377 tracked antidepressant use in the past 30 days. Women’s use moved from 13.8% in 2009–2010 to 18.6% in 2017–2018. Men moved from 7.1% to 8.7%. During 2015–2018, 17.7% of women and 8.4% of men used an antidepressant in the prior month. Women 60 and older reached 24.3%. NCHS Data Brief No. 528 put depression-specific medication in 2023 at 15.3% of women and 7.4% of men. A Cureus analysis of NHIS recorded female depression-medication use moving from 13.3% in 2019 to 15.3% in 2023.

What the Drugs Do to Personality

According to Tang, DeRubeis, Hollon, Amsterdam, Shelton, and Schalet in Archives of General Psychiatry, December 2009, 240 adults with major depression received paroxetine, cognitive therapy, or placebo. After the team matched patients on depression improvement, paroxetine patients reported 6.8 times as much neuroticism change and 3.5 times as much extraversion change as the matched placebo group. Placebo patients improved on the Hamilton depression scale by 1.2 standard deviations and moved 0.18 SD on neuroticism.

NIMH called that a drug effect on personality traits, separate from the lift in mood scores.

A four-week escitalopram trial in healthy relatives of depressed patients left neuroticism and extraversion in place. A five-year observational study of 237 patients tied personality-score movement to depression and anxiety symptoms.

What Patients Say They Feel

Price, Cole, and Goodwin in the British Journal of Psychiatry, 2009 interviewed 38 people on SSRIs, 28 of them women, median time on the drug 23 months. Patients described a drop in the intensity of everyday emotion, a sense of detachment, and a feeling of not caring.

Goodwin, Price, de Bodinat, and Laredo in the Journal of Affective Disorders, 2017 surveyed 669 treated depressed patients. 46% reported emotional blunting. 44% of the women in that sample reported it. 52% of the men did. Bupropion produced a lower signal.

Opbroek and colleagues in 2002 studied 15 patients who already had SSRI sexual side effects. 80% also reported blunting on crying, sadness, anger, surprise, care for other people, and creativity. Those scores did not track Hamilton depression totals.

A 2025 pilot in women remitted from depression on sertraline found 81.8% met criteria for sexual dysfunction. Emotional-intensity scores fell with sexual-function scores.

Who Buys the Books

NIQ’s write-up of Nielsen BookData’s UK Books & Consumers survey attributes 59% of UK book purchases since 2012 to women and 65% of e-book purchases. Vox’s 2025 fact-check found the circulating “women buy 80% of fiction” line lacks a clean U.S. BookScan source. Nielsen BookData’s confirmed 2017 UK fiction split sat near 63% women.

What Personality Predicts on a Shelf

Rentfrow, Goldberg, and Zilca in the Journal of Personality, 2011 sorted the entertainment tastes of more than 3,000 people into five clusters: Communal, Aesthetic, Dark, Thrilling, and Cerebral. Personality predicted those clusters after demographics.

Ng, Bos, Sigal, and Li matched Goodreads tags to Facebook personality scores. Extraverts tracked relationship tags, chick lit, celebrity romance, and memoir. Agreeable readers tracked family and religious themes.

Those papers measure personality. They do not measure a prescription.

The Open Question for This Segment

The study that puts SSRI status on the same row as a genre log, a spice rating, a DNF rate, or a reread rate has not been run. Authors still wait for that dataset.

Two forces sit on the same reader. Depression steals focus and pleasure, which takes her out of the bookstore. A working antidepressant can put her back in the store with a flatter response to the scene that used to wreck her. Volume of books bought and intensity of feeling on the page are separate outcomes.

Marazziti’s 2014 sex-by-drug analysis found SSRIs hit the “love” domain harder in men. Women on SSRIs kept more of the attachment items. That finding sits in the way of a simple “pills killed romance reading” story.

Why Authors Care

The core fiction-buying sex is also the sex with the higher prescription rate. Heat, ache, dread, and catharsis are product features. A large slice of the female audience reports a lower volume knob on those features. Plot, joke, curiosity, and character still move a reader after intensity drops. Forum claims that SSRIs killed a taste for dark romance or sad literary fiction remain unverified social sentiment.

What to Do

Write the emotional peak so the scene still works at a lower setting. Keep the plot engine loud enough to carry a flatter affect. Treat spice level as a dial with more than one paying audience, including readers whose bodies report less desire on an SSRI. Leave diagnosis to clinicians. This research describes published papers. It is not medical advice. A reader who wants to change a prescription talks to her doctor.

Sources:

Commentary

Jonathan: According to the CDC's National Center for Health Statistics, Data Brief number 564, published June 25, 2024, from the 2024 National Health Survey, 24.9% of US women took medication for their mental health in the prior 12 months. Men were only at 13.4%. That figure covers medication for depression, anxiety, and other mental health uses, so it is larger than the depression-only slice.

The NCHS also tracked antidepressant use specifically in the past 30 days. Women's use moved from 13.8% in 2009-2010 to 18.6% in 2017-2018. Men only moved from 7.1% to 8.7%. During 2015-2018, 17.7% of women and 8.4% of men used an antidepressant in the prior month, and women 60 and older reached almost 25%. A separate NCHS Data Brief, number 528, put depression-specific medication in 2023 at 15.3% of women and 7.4% of men.

Thomas: Almost 25% of women, one in four, are on mood-altering medication. Don't think that doesn't affect your readership. This isn't some niche slice like the 2% or 3% of people on GLP-1s. A quarter of your female readers are on drugs that alter mood, and that changes what they want to read.

Do antidepressants change personality?

Jonathan: According to Tang, de Ruberg, and colleagues in a December 2009 study out of Amsterdam, 240 adults with major depression received paroxetine, cognitive therapy, or a placebo. After the researchers matched patients on how much their depression improved, the paroxetine group showed 6.8 times as much change in neuroticism and 3.5 times as much change in extraversion as the matched placebo group. Placebo patients improved 1.2 standard deviations on the Hamilton Depression Scale and moved 0.18 standard deviations on neuroticism. The NIMH called this a drug effect on personality traits, separate from any improvement in mood.

Thomas: A 3.5-times shift in extraversion is enormous, and I'd assume that's a drop, not an increase. In the Big Five framework, extraversion isn't the pop-psychology idea of being popular and outgoing. It means you draw energy from being around other people. Introversion means you generate energy alone or with a small number of people. It's a spectrum, and outgoing behavior is a correlate of it, not the cause.

So if someone is suddenly on a medication that makes them want to be around other people much less, that has a real effect, multiplied across society, on why fewer of us have as many friends as we used to. Maybe that friend now wants their friends in books at home instead of at the bowling league they're no longer part of.

Jonathan: Four more studies back this up. Price, Cole, and Goodwin in the British Journal of Psychiatry, 2009, interviewed 38 people on SSRIs, 28 of them women, with a median time on the drug of 23 months. Patients described a drop in the intensity of everyday emotion, a sense of detachment, and a feeling of not caring. Goodwin, Price, De Bonet, and Lurito in the Journal of Affective Disorders, 2017, surveyed 669 treated depression patients: 46% reported emotional blunting, 44% of the women and 52% of the men. Bupropion produced a lower blunting signal than what patients had been on before. Ulbroek and colleagues, 2002, studied 15 patients who already had SSRI-related sexual side effects: 80% also reported blunting on crying, sadness, anger, surprise, care for other people, and creativity. A 2025 pilot study of women remitted from depression on sertraline found 81.8% met criteria for sexual dysfunction, and emotional intensity scores fell along with sexual function scores.

Why Nuanced Books May Be Losing Ground

Thomas: I wonder if this connects to nuanced books not selling as well. Authors want to write complicated books that blend genres, but a reader whose emotions are being chemically blunted isn't looking for nuance. They want louder, more extreme, more direct emotional appeal.

Here's a metaphor. I have no sense of smell. I can smell smoke and two or three other things, and that sounds like an advantage as a dad since I can't smell dirty diapers, but it also means I don't taste food well. Nobody likes the food I cook for myself because I drown everything in Costco sriracha just to taste anything at all. Someone with a delicate palate can taste the hint of nutmeg. I can't taste the hint of anything.

A reader with a blunted emotional range loses all the nuance in a book the same way. That may be part of why AI writing is doing so well with some readers: AI struggles with nuance, but it can pour on the sriracha. If your taste is chemically blunted, a strong, unsubtle emotional signal is more appealing, and that may explain the shift in what sells now compared to what sold 20, 30, or 40 years ago.

Jonathan: It has to hit harder. It has to be spicier, which is the actual term the romance genre uses now, to reach desensitized receptors looking for the same dopamine or adrenaline hit.

How Romance Fiction Absorbed Erotica

Thomas: This connects to the sexual dysfunction numbers from that 2025 study. Something that would have been far too explicit for a reader in the '80s or '90s who wasn't on SSRIs may now read as tame. The level of eroticism in current romance is genuinely shocking. I was in a group with romance writers, some of whom have sold successfully for a decade or more, who were frustrated that their genre was being colonized by erotica wearing a romance cover. It isn't a romance book, it's erotica with a YA-style cover, but the content is extremely explicit.

Jonathan: That shift created a whole new vocabulary among romance readers: what's the heat level, is it closed door, is it fade to black. None of those terms had heavy usage before, because romance and erotica were considered separate genres. Some of that traces back to Amazon banning erotica as an advertisable category. The category still technically exists, but it's shelved out of sight.

So erotica moved into other genres like fantasy and romance, and readers built a new vocabulary to describe what they were reading and explain it to each other.

Thomas: Now you have to go looking for it, the same way X handles adult content. X doesn't block it, but you won't stumble on it unless you go looking, unlike Instagram or old TikTok, where the algorithm would actively try to zero in on what you might be into. It's harder to reverse-engineer X's algorithm, but one benefit of X open-sourcing it is that government censorship becomes part of the public record, which I suspect is the real motive, since X exists with a political agenda of its own.

Who actually buys books?

Jonathan: NIQ's writeup of Nielsen BookData's UK Books and Consumer Survey found that 59% of UK book purchases in 2012 went to women, and 65% of e-book purchases. A Vox 2025 fact-check found that the widely circulated claim that women buy 80% of fiction lacks a clean US BookScan source. Nielsen's confirmed 2017 UK fiction split sat near 63% women, not as high as the 80% figure but still a clear majority.

Thomas: BookScan doesn't track indie sales or most LitRPG, so I wouldn't be surprised if the real number is closer to 80% once you count those, especially given how female-focused traditional publishing has become. And SSRI prescriptions keep climbing. Once someone starts, they tend to stay on them for life, since getting off an SSRI can be difficult.

I know a woman near one of our local parks who is an evangelist against SSRIs. She tells anyone who'll listen that the drugs ruined her life and to keep the women in their life off them. None of the women in my life are on prescriptions to speak of. We come from a suspicious-of-doctors, unvaccinated part of Texas, and we see doctors more as drug salesmen than not.

A Tangent on Food, Fundamentals, and Earning Your Audience

Jonathan: As predicted, most of the comments are unhappy with us, accusing us of saying women belong in the kitchen. For the record, I'm the one who cooks dinner in my house.

Thomas: I cook for myself because my wife cooks for our kids, who want the blandest food imaginable. One night she made boiled chicken, white rice, and broccoli, no sauce, no pepper, maybe some salt, and the kids devoured it while we just stared. My kids have sensitive enough palates to taste the nuance between boiled chicken and boiled rice. I don't have that problem. I drown everything in Costco sriracha because my own palate is blunted, same as the readers we've been talking about.

My wife is a much better cook than I am. She made something genuinely delicious recently, but it was too complicated and exotic for the kids, and it caused a whole scene. They came around to it eventually.

The deeper problem is that young people aren't being made to read Aristotle, Shakespeare, or Dostoevsky anymore, so they never develop an intricate palate for anything, food or books. In our house the kids get what they want for breakfast and lunch, and dinner is where we try to expand their horizons.

Jonathan: I do the same thing, sneaking better ingredients into what I cook. Once I was short on time and just threw frozen chicken fingers in the oven with no seasoning, and my kids loved them just as much, because all they wanted was a vehicle for ketchup. Depressing, since I want to give them something better.

Thomas: That's the literature parallel. When I talk about what readers want, I'm speaking as a marketer describing the market as it is, not as a philosopher saying how it should be. I'd love it if more people read Aristotle and Shakespeare. I'm even building a podcast called The Daily Aristotle. Half our everyday turns of phrase, including the name of the bone-broth product I put in my food, trace back to coined terms Shakespeare invented.

But if you're trying to make a living as an author, you write the kind of book people already want to read: simpler, more direct, squarely in the middle of the genre, not the innovative, "truffle oil mac and cheese" book. You can't force anyone to read your work. Aristotle is often credited with the line that education forced on someone benefits them nothing. You have to win readers over. The strategy is to build a large audience with accessible books, then gradually make your own work more sophisticated over your career. You'll lose some ability to attract brand-new readers as you do, but you take the readers you have somewhere better, the way Moses led one group out of Egypt rather than saving everyone still there. Write as many accessible, mass-market books as you can early on to build the largest possible audience, then introduce that audience to more ambitious work, and eventually to something like Dostoevsky, later in your career.

Jonathan: The Clone Wars did exactly this. It started as simple kids' entertainment with lightsabers and funny robots, and by the end of season seven it had become a serious meditation on PTSD and the horror of war. My own kids just want to see Ahsoka fight Darth Maul, not sit through the emotional payoff, because you have to grow with the series to earn that ending. That's the model: open with the accessible version, the Kraft Mac and Cheese, and slowly upgrade it.

Thomas: You have to earn that right by mastering the fundamentals first. If you want to be the chef cooking $100.00 Wagyu steaks, you don't start there. You start on cheap ingredients, learn to chop vegetables without losing a finger, and study under someone who already knows the craft. I see a lot of arrogant, prideful writers in Christian fantasy who assume that because they've read and loved C.S. Lewis and Tolkien, they're already qualified to write at that level, without studying the market or their genre's fundamentals. They charge in like unarmed soldiers in World War I and get slaughtered: dozens of sales after thousands of hours of work.

I built a five-year training plan with one of the most successful Christian fantasy authors by sales and awards. He didn't write epic fantasy. He read comic books and wrote for boomer women, adapting to what the market wanted, but he'd mastered the fundamentals first. If you haven't read Brian Godawa, the top-selling Christian fantasy author every month, you're not ready to write in that genre. I've interviewed him, and you can listen for free. I'm using Christian fantasy as the example because it's the biggest slice of my own audience, but the principle holds everywhere: assuming your talent alone will carry you, without learning the fundamentals or reading the bestsellers in your genre, doesn't work. I've watched plenty of overconfident authors publish, discover the market doesn't care how many YouTube subscribers they have, and quit within five years, bitter about it.

Seth Ring, another guest of mine, writes Christian LitRPG and outsells MrBeast's pre-order and review numbers in an opening week, because he learned to play the game rather than just charging the line unarmed. You have to learn the actual craft: taking cover, coordinating your resources, using the tools available, not just charging straight at the market with nothing.

Breaking News: A Murder Case and the Insanity Defense

Jonathan: There's a court case we've covered before that's back in the news today. A woman strangled her three children, one after another, and left them in her basement. There's been a strange public reaction treating what she did as not wrong, and the jury nearly returned a not-guilty verdict.

One juror held out, arguing that murdering your children is wrong regardless of drugs or mental state.

Thomas: If you kill someone, you should die for it, and it doesn't matter whether you have an excuse of being mentally ill. Being mentally ill is more reason to execute someone, not less. We can have a world free of murderers by executing murderers, the way it's been done for most of human history.

Jonathan: That just makes more murderers.

Thomas: That's the Batman fallacy: if Batman would just let the Joker be executed, more people would be saved. It is good and right for the government to use force against evildoers so the evildoers no longer exist. The idea that "it was just their body, their heart was good, we're only executing the body" doesn't hold up. This isn't the only world. God will know his own.

I'm getting more agreement on this in the comments than on any of the SSRI material.

Jonathan: Give it time. This will be the clip they roast you over.

Thomas: If you don't know the case: a woman strangled her oldest child in front of her other two, then strangled the next, then the next, and left all three dead in the basement.

There's a real social-psychosis element to the public reaction treating this as somehow not her fault. You chose to take the drugs, if drugs are the excuse. I was pressured to get vaccinated and didn't. You can say no. If we want a better world, we have to hold people responsible for their actions, which means murderers don't get a pass. That hasn't been a controversial position for most of history.

Any time a society carves out an exception, "it's fine to kill this class of person," that's the same principle our country was founded on rejecting. You don't get to murder people, whether or not you're mentally ill, and it doesn't matter if the victims are your own children.

Jonathan: A commenter asked why we even have an insanity plea. The Fat Electrician did a good video on its origin: a US congressman shot and killed the son of the man who wrote the Star Spangled Banner, in front of the White House, after finding out the man had an affair with his wife.

Thomas: His lawyer invented the insanity defense to get him acquitted, and it worked.

It shouldn't be a defense. Being insane is more reason to execute someone, not less. Do you want a world full of people who murder because they claim demonic possession or mental illness? The purpose of government is to remove people like that from society. People could leave their houses unlocked in the 1950s because a century earlier, horse thieves were hanged. I'm not calling for the "bloody code" of 1650s England, just for executing murderers, which every successful civilization in history has done: China, Rome, Greece, Anglo-Saxon England, the Normans, the Goths, Egypt. Name a historic people who thought murder was fine as long as you had a good excuse. You'll only see that right before a civilization collapses.

Civilization is what keeps any of us fed. I don't grow my own food. If civilization fails, I watch my children starve, and the only people who survive that are the most ruthless and violent. That's the outcome to avoid at all costs, even if avoiding it means executing murderers.

Jonathan: When a government chooses to be mentally stunted as a matter of policy, this is what you get.

Thomas: Jesus wasn't just a peacemaker either. Revelation describes him returning on a white horse with a sword, judging the unrighteous. He said plainly he didn't come to bring peace but a sword. "Hippie Jesus" who only preaches gentleness is a boomer-era invention, not a biblical one. There's a time for justice and a time for mercy, and a worldview that's all mercy, all the time, hands control of your life to the most ruthless people around you.

Jonathan: He also said it would be better for someone to be thrown into the sea with a millstone around their neck than to harm a child, and followed it with the most extensive description of hell in the New Testament. That's the same message either way: don't hurt kids, or there's a reckoning after death too.

Thomas: Which is the more merciful system: justice in this life, or only facing justice in the next one?

Justice, Vengeance, and the Death Penalty

Thomas: I realize this is now a controversial position, but I'll hold it. If we let people get away with murder, there's nothing we won't eventually let them get away with. We live in a society where half the country already believes it's fine to end a defenseless life as long as it happens before birth. That standard has since expanded. Now the bar for taking a life can be as low as the mother being unhappy or on the wrong medication.

Killing is wrong. Murder is wrong. But there are times when killing is justified: in war, and when a neutral government executes a murderer. It's better for a murderer to be executed than to keep murdering or spreading evil in prison. A life sentence is still a kind of execution, just a slower and more torturous one that doesn't lead anywhere near repentance. If you want to save the person's soul, the more merciful path is to give them time to make things right with God and then carry out the sentence. This isn't a call to revert to 1650s brutality, just to roughly where things stood 100 years ago. And on wrongful convictions: we now have far better forensic science than we used to, so that risk is much lower than the old objection assumes.

Jonathan: If we're focused on justice, we can do a decent job of getting it right.

Thomas: What people usually call "social justice" is the government taking someone else's money and redistributing it, which isn't the same thing as justice. Justice is an eye for an eye, a tooth for a tooth: the punishment fits the crime. That standard is written on the human heart and shows up across every civilization throughout history, which is why it's universal. People recoil from it because they've absorbed the idea that it's just about revenge, but it isn't. Nobody is calling for the aggrieved party to take justice into their own hands. This is specifically about a neutral third party administering it. Otherwise, you get a cycle of revenge, back and forth.

I see this at home. When one of my children is violent toward another and I don't step in with consequences the wronged child sees as fair, that child retaliates, and it escalates. As the head of my household, I have to step in and administer justice or the cycle keeps going.

Jonathan: Just put fluoride in the water, Thomas.

Thomas: I'm not vaccinated, and my family's ranch exists partly because my grandfather, a medical doctor, didn't trust fluoridated water even decades ago. He served as an anesthesiologist, including during the Korean War, and kept his own well water instead of drinking from the Austin city supply. He never accepted what's now fairly well established, that fluoride lowers IQ. It was never controversial among people willing to look at the research.

Ritalin, Height, and the Dating Market

Jonathan: We had this same argument 15 years ago about Ritalin, when the answer to managing boys in school was to drug them into being calmer and easier to handle, more like the behavior expected of girls. We medicated a whole generation of kids, and it hasn't worked out well.

Thomas: It also stunted their growth. Caffeine doesn't stunt growth, but Ritalin does, and height has become the top status differentiator in dating. Women increasingly won't date short men, and some of those men are short because their parents put them on Ritalin as kids. Mothers who thought they were helping their sons may have unintentionally taken them out of the dating pool entirely, which means no grandkids.

Jonathan: Your parents sexually disadvantaged you. That's a fun one to bring up in therapy.

Thomas: In our defense, Jonathan gave a real trigger warning at the top of this segment, which he normally mocks me for doing. He told anyone sensitive to log off now, no apologies.

We've said enough. In a multitude of words, sin is not lacking, and we've produced a multitude of words today.

Liked it? Take a second to support us on Patreon!
Become a patron at Patreon!

Leave a Reply

Want more help?

Get a weekly email with tips on building a platform, selling more books, and changing the world with writing worth talking about. 

You have Successfully Subscribed!