Pay Attention Proves Its Own Point: David Pakman Wrote a Book the Way the Feed Taught Him To

Pay Attention is the most honest book about the algorithm I've read this year. Its own pages prove the thesis.

David Pakman spent years tuning YouTube thumbnails, then wrote a 288-page book that five of its 23 Goodreads reviewers call repetitive. The machine he describes got to the manuscript too.

Harper published it September 22, 2026. As of November 5, 2026 it sits at 4.12 stars on Goodreads off 96 ratings and 23 written reviews, with 118 people reading it right now and about 2,600 more who've marked it want to read. The hardcover runs 288 pages. It's on Amazon in hardcover and Kindle.

He hosts The David Pakman Show, progressive political talk that built most of its audience on YouTube. Pay Attention follows The Echo Machine, which the jacket calls a New York Times bestseller. Creator first, author second. You can tell which job wrote which chapters.

I build organic social and content systems for food and beverage brands. My shop runs AI-first every working day. I live inside the same dashboards he describes, deciding whether a hook that pulled views was worth the voice it cost the brand. I read this one as a practitioner with skin in the game.

Data current as of November 5, 2026.

The book on one index card

Harper published it September 22, 2026. Hardcover runs 288 pages.

Goodreads has it at 4.12 from 96 ratings as of November 5, 2026.

Seventeen of twenty-three written reviews give it four or five stars.

Four sit at two stars or lower, and every one of them cites repetition or missing depth.

What is Pay Attention arguing?

The feed rewards whatever keeps you scrolling, and over time it rewrites the people who feed it. Pakman walks through how creators get trained by their own numbers, the formulas behind thumbnails, audience capture, and why he thinks the political right fits the attention economy better than the left. He closes with ways to make yourself less predictable to the machine.

The line readers keep lifting is his. One four-star reviewer quotes Pakman describing a world "Where outrage is engineered, pleasure is optimized, and truth becomes optional." That's the whole book in eleven words, and it's a good eleven.

The five-star crowd loves the insider view. The most-liked review on the page, 15 likes as of today, says it plainly: "The algorithms are like invisible bosses." Another five-star reader came away with a suspicion confirmed about "why video thumbnails often rely on the same formulas, visual cues, and gimmicks."

None of that is news to anybody who runs a brand account. I don't need a book to tell me thumbnails follow a formula. I need one to tell me what the formula costs the people who follow it. On that question Pakman has something to say, because he paid the bill himself.

Why do readers keep calling it repetitive?

The complaint shows up at almost every star level. The book earns it.

The one-star reader didn't hedge: "SO UNGODLY REPETITIVE early on that I quit." Then the kicker. "Stopped at page 80." A three-star named the mechanism better than I could: "Felt like he was using some of his content creating tweaks to just say the same thing over and over." A two-star went further: "This started to sound like it was written by AI which is ironic considering it’s about algorithms."

That last reader blamed a chatbot. I'd blame the retention graph.

Anybody who's cut a daily show knows the move. You restate the premise every few minutes because a fresh slice of the audience just wandered in from a recommendation and has no idea what you said at minute two. On video, that keeps people watching. On paper, it reads like a radio station out past Marfa looping the same weather report for the one truck on the road.

Good AI-assisted writing doesn't loop like this. We put AI on client copy every day, and the standard is that nobody can tell. Saying the same thing four times is a human habit. An editor gets paid to cut it. Somebody at Harper let the YouTube cadence through.

The three-star review is the sharpest thing on the page. The book is about audience capture, and the author's own prose shows the symptoms.

The confession is the best part

Strip out the repetition and what's left is a working creator admitting what the numbers did to him. That's the chapter worth the cover price.

A two-and-a-half-star reviewer who otherwise had real problems with the book still put it clean. Pakman shows "how creators slowly adjust their behaviors over time to please the all-powerful algorithm." It happens a post at a time, "Until they lose their identity and become an algorithm-optimized version of themselves."

A four-star reader got to the same place from the other side: "Pakman is at his best when he discusses his own experiences and lessons learned as an internet content creator."

I've watched versions of this hit food brands, and it never starts with a bad call. A salsa brand posts a founder talking about her grandmother's molcajete. Fine numbers. Then one week a shocked-face hook about ghost peppers pops and the dashboard lights up like a Friday night football field. Next month every post opens on a shocked face. By spring the grandmother is gone and the brand sounds like every other jar on the feed. Nobody chose that. The scoreboard chose it, one post at a time.

I'm not immune either. I mine my own posted hooks against real view counts, and the data's useful. It also pulls, every single time, toward the loudest opener I've got. Knowing that is the only defense I've found.

Is this a politics book or a marketing book?

Both. And the politics will cost Pakman readers he could have kept.

He argues the right is built for the attention economy. One two-star reader pushed back hard: "I think there's plenty of click-bait tactics across all spectrums of political beliefs." Another two-star wanted the opposite, more partisan comparison: "I really wanted him to expand on that especially the differences between different sides of the political spectrum." Same chapter, opposite complaints. That's a book that picked a side halfway.

One four-star reader docked a star over the publisher. HarperCollins belongs to Rupert Murdoch, so "every copy of this book helps Rupert Murdoch spread BS on Fox News." That review tells you more about the attention economy than some chapters do. People grade a book on the logo on its spine now.

If you're here for brand work, skip the partisan fight. The mechanics he describes don't vote. Same thumbnail math runs a cooking channel, a sneaker drop, and a congressional race.

Where does the book stop short?

At the part where you do something about it.

A four-star reader put it best. After laying out the machine "he gives no real assistance in learning how to tackle it on its own terms." Pakman's fixes stay personal and analog. Go running without your phone. Read a paper newspaper. Be less predictable. Fine advice for a person. Useless for a brand that has to post on Tuesday.

So I wrote the brand version he didn't.

What should a food brand do with this book?

The dashboard starts making creative calls the minute you let it. Build the guardrails first.

Four steps. Boring ones. That's why they stick.

1. Write the voice down before you open analytics. One page. Words you use, words you never use. Drop it in a brand rulebook. You're finished when anyone on the team can name three things the brand would never say.
2. Test one hook pattern for 30 days. Change only the opener. Hold the rest still. Then read what came back next to the voice page. You should be able to say which opener won and whether it still sounds like you.
3. Keep one slow format on the calendar the feed doesn't care about. Long recipe. Founder story. Plant tour. It runs every month, views be damned.
4. Line up your last 20 posts. Find the week the brand started sounding like everybody else. A boring product with a clear voice beats an exciting one with a borrowed face. Point to the first post that slipped and write down why.

None of this fights the algorithm. It leaves the last word on voice with someone who can taste the salsa.

Should you read Pay Attention?

If you run a feed for a living, read Pay Attention with a pencil.

The creator chapters are the best first-person account of audience capture I've seen in print this year. Skim the rest. The readers who quit at page 80 aren't wrong about why.

I said the same thing last week about Open to Work, from the opposite direction. That book's readers caught an ad. This one's caught a YouTube habit.

It's worth the 288 pages. The loudest critics proved the author's thesis when they recognized the feed in his prose.