AI is solving equations and passing exams. Getting it right is no longer scarce.
But that is in domains where there actually is a right.
After sitting in a room by myself writing books for fifteen years, I have found the “correct” way to write. Often when I read something, in the background I am thinking about how it’s written wrong, and how I would correct it.
It’s only my opinion, of course. There is no correct way to write. But to me, there is. I have my weird version of right.
I have very specific ideas about building and releasing tension, withholding information until the moment of maximum impact, what words with what sounds should be put down in what order.
No matter how much I wrestle with an LLM, explaining why I make the choices I make, it will always write glaringly, embarrassingly, unequivocally wrong (to my eyes). AI will never write like my weird version of right.
That doesn’t guarantee my success as a writer. There’s a whole lot of luck in creative work. But I do believe that gives me a better chance of doing anything worthwhile than a field where there really is a right answer.
As getting it right becomes cheap in domains where there are answers, it becomes more valuable to have your weird version of right in domains where there aren’t.
Book:The Bachman Books (Amazon) is a collection of novels from, and the story behind, Stephen King’s secret pen name.
It almost never seems like a good idea to quit a project. If you never finish, you never unlock the rewards that await.
But in creativity especially, quitting is a gift to your future self.
It’s hard to see that when you’re in the muck of something that’s not working. All you see is what’s in front of you – the many details you could keep trying to salvage.
You can’t see what it’s costing you because the alternative hasn’t had a chance to present itself.
The seed of an idea takes time to sprout and grow roots. It needs to get quite large before you’re bothering to dodge it with the lawnmower. You can’t even see it when your primary goal is still cutting the grass.
Mark Twain had a good approach to this: If a book wasn’t working out, he’d leave it be. That was easier to do because, as he explained,
There has never been a time in the past thirty-five years when my literary shipyard hadn’t two or more half-finished ships on the ways, neglected and baking in the sun…The minute that the book tried to shift to my head the labor of contriving its situations inventing its adventures and conducting its conversations, I put it away and dropped it out of my mind. Then I examined my unfinished properties to see if among them there might not be one whose interest in itself had revived through a couple of years’ restful idleness.
Maybe the most valuable thing I’ve learned writing Finish What Matters (now available for pre-order) is that a big part of finishing is knowing when to quit.
Quitting is a gift to your future self. You can’t unwrap it until what’s inside has had a chance to grow.
Aphorism: “It’s when you are not doing much of anything that you are accomplishing most.” —Pierre-Auguste Renoir
Cool:Whispr Flow is really, really, really good speech-to-text.
People wonder what will happen to writers and writing now that AI is so “good” at writing.
Since AI can crank out what they think is good writing, many conclude that there’s no longer value in actually sitting down and writing.
I have found that AI actually makes writing much harder, rather than easier.
The paradox is a bit like what happened as labor-saving devices made it “easier” to keep a clean house or clothes. Maybe the first people with a vacuum cleaner had an edge, but eventually everyone expected their house to be that clean. Trace that trend back to the era of dirt floors, or remember that even Kings rarely bathed, while we shower every day.
If you don’t see the point in letting AI write for you – as I don’t – the things it can do for you just increase your standards of quality. That image of perfection is a bit closer. You find yourself striving harder and longer to try to achieve it.
AI makes it much easier to:
Explore more potential examples for illustrating a concept
Find the work of others who have talked about similar ideas
Get multiple rounds of feedback in a single session, instead of waiting weeks for test-readers (if you can find them)
Connect your ideas to more complex fields, in which you can achieve a level of proficiency that wasn’t previously practical
Get a reaction to more nebulous ideas
When you’re striving for the impossible, these create more work, not less. It means you have to:
Consider more examples, finding the one that subtly edges out others in appropriateness and emotional tenor
Read more of others’ work to verify your point-of-view
Mentally slice-and-dice your own writing at multiple levels of abstraction – from the macro-level down to individual words
Read more source material to get a deeper understanding in subjects that would have been too complex to bother with
Follow ideas that were previously too hard to turn into something useful on your own
For my upcoming book, that meant I achieved a level of expertise in the Italian Renaissance that would have been out of reach before. The alternative would have been enrolling in a Masters or PhD program, which would have led me to writing a book on history, not creativity. AI essentially sent me on a self-directed research trip to Italy!
Previously, you were limited by how much mental heavy lifting you could do in a given session. Now AI is there to spot you to get in more reps after you’ve reached your limit. (AI makes Mind Management more important than ever.)
As it becomes more feasible to achieve a higher level of quality, the ways we find writing to consume also increase the importance of that quality. Algorithms are so sophisticated, you no longer need hashtags (thank God). The number of people you can reach is getting bigger – if you can edge out the growing pool of competition. The media environments of recent years rewarded quantity, and now the premium is increasing on quality.
If AI is making writing easier, you are Not Gonna Make It. It should make it harder.
Aphorism: “What is called ‘computer literacy’ does not involve raising questions about the cognitive biases and social effects of the computer, which, I would venture, are the most important questions to address about new technologies.” —Neil Postman (1985)
Every time we make a thing easier, we lose some crucial element that was wrapped up within it.
I remember many mornings in San Francisco being struck by irony on my way to work. As I passed the open windows of a gym, I heard pumping music and the amplified shouts of a spinning instructor. The students were riding bikes, exercising while going nowhere. I was riding my bike, exercising while going somewhere.
As AI and robotics advance, we will stop having to do some things. At the same time, we’ll stop doing some things that are good for us.
I’m finding that with these developing powers I now opt for many things I would have once considered drudgery.
I’m growing food, making soap, cooking new recipes, and doing plumbing.
Now that the informational bottlenecks that prevented me from doing these before have been widened, doing more things in the physical world has changed my perspective on other chores, such as laundry, washing dishes, and cleaning.
So when I see a video of a housekeeper android, my immediate reaction is, Damn, I want one of those!
But then I realize that in some version of the future, it’s not the physical drudgery of keeping a household running that we outsource.
The other day, my water distiller broke. The old me would have just spent $125 to buy a new one. Instead I asked Claude to guide me through troubleshooting.
Instead of sitting in an office chair, clicking a one-click-order button, then sitting in an office chair even more to figure out how to make an extra $125, I went into town.
It was a bad capacitor. I didn’t know what that was but I walked around and asked people because you can’t find many businesses on Google Maps. I found an electronics-repair shop, where they soldered in a new one. Parts and labor: $1.25.
My distiller works like new. I got to move around and use my hands and body, and isn’t that what they’re for? I used my mind, but didn’t have to go through the mental drudgery of learning electrical engineering to know it could be a broken capacitor.
Given the trajectory of the past several decades, we’re biased to believe that automation will next lift us out of physical drudgery, but it’s not done lifting us out of mental drudgery. Maybe it’s actually been heretofore pushing us into it.
Maybe we’re made for moving around and doing things in the physical world and shouldn’t be so eager to outsource it to an android.
Aphorism: “There is no habit you will value so much as that of walking far without fatigue.” —Thomas Jefferson
Book:In This Economy (Amazon) is Kyla Scanlon’s fun introduction to money and markets.
I have done just enough jogging to say that pacing yourself is a lot like mind management.
If you’re a beginner and just keep jogging, your lungs soon start to burn and it sucks and you never want to jog again.
By contrast, if you keep yourself within a comfortable heart-rate range – by walking when it’s high and jogging when it’s low again – jogging actually becomes kind of enjoyable. String together enough days of that and you adapt to where you can just cruise along in that range.
In the days before I learned mind management, I would frequently peg myself out mentally in the same way. I’d end up saying, “I never want to edit a video again. That was too mentally exhausting,” not recognizing that of course it was exhausting because I tried to learn a new thing while also powering through getting the thing done.
If you keep monitoring your heart rate, you stop needing to check what it is. You can just feel in your body whether it’s high or low.
Mind management becomes the same, where you can sense that you’ve just been presented with a task and face a choice. You can power through and probably burn yourself out. You can switch mental states and get off the track you were on. Or, you can set that task aside for a better suited pool of energy and stay in the state you were in.
The challenge is, there’s no quantifiable metric you can monitor. The four stages and the seven mental states give you a vocabulary to guide you. From there, you learn through practice.
Aphorism: “What I was given to study in school I have forgotten; what I decided to read on my own, I still remember.” —Nassim Taleb
I’ve been asking myself an insanely simple question that does a surprisingly good job of protecting my attention.
“What’s over there?”
When I reach for my phone: “What’s over there?”
When I switch to a social app: “What’s over there?”
When I get the urge to open my laptop: “What’s over there?”
There’s an arresting simplicity to What’s over there? It forces me to think for a second. In rare cases, I have an answer, and that helps set my intention to fight against being sucked into the maelstrom.
But most of the time, I have to admit, I don’t have a clue what’s over there. I just went there five minutes ago – do I expect something new? Maybe I should create something instead so there’s actually something worth checking on.
I end up fumbling for an answer, and it makes my misguided urges seem kind of dumb – which is the desired effect.
Give it a shot: What’s over there?
Aphorism: “Simple, clear purpose and principles give rise to complex and intelligent behavior. Complex rules and regulations give rise to simple and stupid behavior.” —Dee Hock
Cool: This quiz can tell U.S. people where they grew up in just 25 dialect questions
Pleasing the algorithm too quickly is a path to mediocrity.
If your ideas gain immediate support, they probably aren’t new. If they gain resistance, you’re still in familiar territory. But when your work is met with crickets, that can actually be a good sign.
One of the best ways to do impactful work is to be the first to see the world in a particular way. That’s how you catch the attention of the proverbial “innovators” and “early adopters.”
The algorithm shows work that most people react to. But if most people are reacting to it, it’s already old news. Go ahead and post an Impressionist painting and enjoy the likes, but you’re 150 years too late for it to mean anything.
Obviously, if your work is ignored that very often does mean something is missing, whether it’s craft, legibility, or cultural significance.
But the litmus test is whether it’s still interesting to you. Are the feelings you’re expressing real? Are you still curious about what you could learn by digging deeper?
Comedian Louis CK said, “Always, the best bits I have in a special, the ones that are like, ‘That just kills every time,’ started with silence.”
I know the pieces I’ve felt the most vulnerable to publish are almost always met with crickets at the start. It’s as if you can hear the collective fear of people who say to themselves, “It feels risky to be the first to ‘like’ this.” But when they catch on (if they do) they’re the ones with the widest reach, eliciting the most sincere reactions.
Humans are pack animals. The innovators and early adopters have to lead the pack. But to reach them, you have to endure being ignored, until you have something interesting.
Aphorism: “Will public resistance prevent the introduction of technological control of human behavior? It certainly would if an attempt were made to introduce such control all at once. But since technological control will be introduced through a long sequence of small advances, there will be no rational and effective public resistance.” —Ted Kaczynski
I don’t know if it’s more remarkable that book sales are down lately, or that they stayed as high as they were for as long as they were.
Tim Ferriss recently reported an 80% decrease in book sales, and cited an industry report that self-help sales were down 26%.
Ferriss thinks people are using AI as a replacement for “prescriptive nonfiction.” I think there’s also an overall cultural malaise around ambition.
But I also see an exciting opportunity in the drop in book sales.
Because it seems obvious that self-help and business books for the past decade or so have been written to be sold, not read.
They’ve been entire books about one compelling idea – mostly summarized by the sticky title – stretched out for 250 pages. The text has been broken up by a subhead every 200 words, and littered with one-sentence paragraphs and one-word sentences. There have been diagrams every other page. Concepts have been broken up into bullet-points, the flow of text interrupted by sidebars, each chapter ending with “key takeaways.”
I personally hate this kind of “book.” I don’t know that I have in the past ten years once finished reading a book in this category, aside from the rare ones absent these features. Subtle Art and Four Thousand Weeks come to mind.
But that is what you get when the book industry responds to a market of readers who don’t actually like to read. James Clear has said he only expects people will read the first fourteen pages, so that’s where he concentrated value in Atomic Habits.
I like to read books. I enjoy prose. I do not enjoy sifting through a stew of bite-sized ideas that aren’t tightly integrated with one another to build support for a meaty central thesis.
Info-snacking is what social media is for (and now also what AI is for). If the concepts can be summarized in a series of bullet-points, it does not deserve to be a book.
So as more people get answers through LLMs, fewer books will be sold. But as someone who actually likes to read books, I’m optimistic about what happens when books are no longer bought by people who don’t have the attention span for reading actual books.
Maybe books can be written not to be sold, but read.
Aphorism: “Whenever anybody tells me they want to write a book in order to help other people, I always think, Oh, please don’t. Please don’t try to help me.” —Liz Gilbert
While writing my latest book, I conducted an experiment on myself.
When I interviewed reader Lisa Van Gemert, she told me about her secret Amazon Wishlist. That she’ll reward herself with items for finishing certain tasks.
I decided to try it for myself. So to motivate myself to finish my manuscript, I promised myself a new Kindle.
It was strangely difficult to set up this reward for myself. My current Kindle really was getting old and slow, and I had $120. Why not just buy the thing?
But I held off, in the name of science. A couple things happened.
As I struggled through using my slow Kindle with little battery life, I kept reminding myself, Finish the manuscript, and you get a new one.
Then, once I finished the manuscript and got the Kindle, I was even more excited about it than I would have been had I just bought it.
Lisa’s wishlist has really inexpensive items on it. As she told me, it helps her tell herself like, “‘You finish indexes. They suck, but you did it. And you have this cute eraser that consistently reminds you of that.’”
So while the reward is valuable, so is the physical presence of the item after you’ve accomplished the goal. Indeed, every time I use my Kindle, I’m reminded that I earned it.
Since reading on my Kindle is a big part of my business, it’s going to be a tough reward to top. But the experiment worked so well, I’m trying it again. I’ve already picked out the pair of pants I’ll earn when I’m done editing.
Have you ever set up a reward for yourself?
Aphorism: “If you’ve truly created an innovative work, it’s likely to alienate as many people as it attracts. The best art divides the audience.” —Rick Rubin
Book:Meditations for Mortals (Amazon) is Oliver Burkeman’s meditations on embracing finitude.
The shift that has happened in how media is consumed has made it way more comfortable to create.
So long as you recognize that shift.
We used to be in follower mode: People would follow you on a social media platform, or even subscribe to your RSS feed. They would see everything you made.
Lots of people miss those days. They complain their followers don’t see what they create. But that actually sucked.
Follower mode put way too much pressure on the creative process. You had to get it right the first time, because everyone would see it.
It would be too embarrassing to publish two variations of the same thing. You’d scare your followers away. You just had to guess.
In feed mode, the pressure is off. You can try it a dozen different ways. Your followers will only see it if it’s better than the other things the algorithm could have shown them.
Add to that, your follower count is no longer the limit. Beat out the other stuff on the algorithm, and your reach is nearly unlimited, even with 0 followers.
You can be as cringe as you want, and hardly anyone will notice.
This is especially good news for the Leonardos. Those who find their process through exploration. Follower mode was great for the Raphaels – those freaks who could somehow plan out one version of a vision, then execute.
Follower mode was harshly judgmental and stiflingly rigid. Feed mode lets you be improvisational and iterative.
Book:The Origin of Science (Amazon) presents evidence from persistence hunters that explain how humans evolved to be scientific.
Cool: David Attenborough’s documentary of an 8-hour persistence hunt in the Kalahari.