Michael Dean: The 27 Patterns To Fix Bad Writing | How I Write
01:51:30 Share

You ever sit down to write and you just feel like something's off with your writing, but you can't quite pinpoint what it is? Well, this episode is the antidote. And my friend Michael Dean, he's a former architect and VR specialist,

And then he got obsessed with the structure and the patterns of great writing. The guy's a wizard. I worked with him for years and he's got this truly unique way of visualizing the writing process. And it's going to give you X-ray vision into exactly where your writing can improve whenever you sit down to edit. And then if you stick around to the end, Michael's going to share this wild outlandish vision of how he thinks AI is going to change writing and how AI will help you master the craft.

Alright, so why do you think writing quality is subjective?

So it's definitely both objective and subjective. I just think that objectivity, rationality, reason, they're all severely underrated in our culture of writers. So when it comes to grammar, right, everyone will agree there are objective rules. That's why grammarly can exist. But when you ask what makes writing good or great or world class, that's obviously more ambiguous, right? Because readers have different lenses. They prefer topics, different time periods, different styles.

And so I think it's kind of comforting for a new writer to say, "Oh, if there are no wrong answers, I can just show up, follow my taste, and there'll be somebody out there who will like this." But I think it can pretty quickly get frustrating where you look at a draft and it feels off. You don't have the language to know what's wrong with it. So I think

The idea of this objective system of quality can really help writers find their blind spots and get better. Okay, fine. But somebody might say, "Hey, but like how do you do an objective system of quality? So much is missing." But you've really looked into this. So tell me about what you found that

that you see that I think most writers are missing? Yeah. So I think I believe this because of the volume of stuff I've seen, right? There's thousands of drafts from new writers. I've read a hundred classic essays. I've written almost a million words trying to get better. And I think underneath this tremendous variety, underneath all possible tastes, there are fundamental design problems that we all deal with.

And so this is kind of a map of essay architecture. That's the project I'm working on. Regardless of the topic, everyone has to scope ideas and set boundaries. Regardless of the topic, everyone needs to think about form. You have to start and end somewhere and then craft a path between them.

And regardless of the idea, you need to craft a voice that permeates every sentence. And I'm not saying that we should all emulate a single voice or that there's one template that'll solve all of your problems. What I'm saying is that within your genre, there are fundamental design constraints that you can't ignore. So let's just go through them quickly. So under idea, we have material, thesis, and title. Talk to me about those.

So material is the substance of an essay. It covers your stories, your references. How do you shape that into an argument? Where thesis is the central idea that all the material revolves around. And once you know your thesis, you have to cut some of the material. And then title is eventually how you compress the thesis into a single package that can get somebody to even enter this idea space. And then under form, you have paragraphs, structure, and tension.

- Yeah, tension is what I use instead of storytelling. But in my mind, paragraphs are the atomic unit of form. We can talk about that 'cause I know people have some trauma with five paragraph essays, but paragraphs are the atomic unit. Structure is like this mid-level sense of how you move between ideas and maintain cohesion. Whereas tension is this more mysterious thing where how do you build suspense and get someone to care if the main character dies, right? Like how do you actually get readers invested in what you're writing about?

Now with voice, and I'm excited to hear about how you think this is objective because the first one is spirit. - Right. - Spirit, sound, and sight. - So spirit is the element of voice that AI has most issue with because spirit is all in the subtext. Underneath every sentence,

there are implications and tone effectively is the combination of those implications. Like what is attitude if not across multiple paragraphs, we can sense how you feel. Whereas sound is the way prose comes across when you read it out loud and that taps into repetition, rhyme, rhythm. Whereas sight is obviously imagery. It's the words, it's how do we make things concrete? - So what we're looking at here is we have nine pillars

Or we have three pillars here. Yeah, three dimensions. Three dimensions. Then we have nine elements. Right. And then we have 27 different patterns of writing. Exactly. And what you're saying is that we can look at any single piece of writing along the dimensions, the elements, and the patterns, and we can instantly see what's going on to then get an objective measure of the quality of a piece of writing to see where it's really strong and where it's really weak.

Yeah, this sounds crazy, but when I read essays, I score every essay one through five on 27 criteria, sometimes 81 criteria because there's even three more in every pattern. It's ridiculous, but it's part of the project. When people first hear this, I think they freak out a little bit because they think that I'm assigning templates and these very rigid rules. That's how writing advice often comes across.

This is very much inspired by Christopher Alexander's system. And the one thing to keep in mind is that a pattern is not a solution. It is a question, right? So there are 50, hundreds of different styles of repetition, right? There are 80, who knows how many possible sequences there are. I'm not telling you, you have to use the hero's journey, right? It's a question that arises

in essays across time and you have creative liberty to answer every pattern however you want, but we can still come up with objective criteria to measure if a solution is good or bad. So that's kind of how it's trying to fuse. That's why I say it's both subjective and objective. So can you show me a piece of writing that you think is really good? And I want to see how you analyze it, how you break it down, because I want to be convinced. I want to be convinced that

We can take something where I'm like, oh, no, it's really good. But I think it's just like artistry. It's just artistry. And I want you to show me how there's an objective quality to it. So we'll definitely break down two essays today in high detail. They're very different essays. And you'll see how despite the differences, they actually are, you know, working on the same foundation. But to really explain this with a good metaphor, I think we should start with architecture because architecture is so instant and visual and you can get it right away.

I think when we see everything at a taste level, it's very easy to see the differences between things. But architects have a specific way of analyzing very different styles and finding similarities. And so I want to show you how an architect does this, and then we're going to do it for writing. And so we have an example here. These are two buildings in Manhattan. On the left, you have the Guggenheim, 1959, frankly, right. On the right, you have the Apple Store on Fifth Avenue.

Right. And like what's your first impression on how these are different? I mean, everything. One is circular. One is glass. One feels sort of mid-century. Another one feels very contemporary. One is boxy. One is opaque. The other is transparent. There's a lot that's different. One is a very subtle signage and the other is a levitating logo. These two things seem opposites. They're half a century apart. One's a museum and one is selling iPad Pros. They're very different.

different things. So what architects do is they'll cut plans and these are section drawings. So these are views from the side. They're flattened so there's no perspective. There's no materials, right? So you kind of just see them kind of one for one what they are. So this is already starting to give you a view where you can compare them analytically, but it doesn't stop there.

What architects then do is they will make these diagrams over drawings and they'll color code them. So you can see right away that both of these projects, the Guggenheim and the Apple Store, they're all about the central atrium. You walk into a grand vertical space with a majestic skylight on the ceiling

And then you have circulation running around the perimeter. And then off of this core, you have the core spaces, which don't have a lot of lights because they need to control the lighting for art or for products. But then they each, through the ceilings, they have very subtle ways, different, but they're ways to let light in without it directly hitting the space. All right, so when you...

reduce the complexity and just isolate certain features, you actually realize that these two are very similar and we can understand that they're both successful in creating a powerful and clear lobby. And if we were to look at, say, the Whitney Museum, for example, you could see, oh, there's a reason this is confusing because it's not following the same patterns. We've seen the Guggenheim, we've seen the Apple Store. Now connect that with writing for me. When people kind of question how we can look at

everything along the same framework, I usually present them these two essays. So you have "Consider the Lobster," David Foster Wallace, 2003, and then "Shooting an Elephant" by George Orwell. I think it's 1936. And at the surface, these are opposites again, right? One is about saving animals. The other is about shooting an animal. One is about a meandering thought process during a festival. And the other is a extremely precise

Adventure, right? Very linear. One is maximalist one is minimalist. So again at the surface These are very different essays, but this is a radar graph that is showing how you know on my first pass I scored each of these essays and They both score 9 out of 10, you know, I scored one a nine five and one a nine one but effectively they are both

masters of the fundamentals, they just have different ways that they will answer the same questions, right? They both do sequence very differently. They both do experience very differently. They both do imagery very differently, but they are mastering them in different ways. Cool. So let's start with Considered Lobster. Great. That sound good? Yeah. So what is the one thing, if we're going to deconstruct this, what is the one thing that we can learn from it?

Yeah, I think there's something very simple that could fix a lot of problems that new writers face. I mean, I faced this when I started. When you write, the scope just endlessly expands. It's like a scope balloon. And so you start with a thousand words and then it's 15,000 words and you have 10 essays in one thing. And so I think David Foster Wallace in this book of essays, Consider the Lobster, not just this essay, but all of them, he is amazing at zooming in to something specific, like a very...

oddly specific thing that represents a much bigger idea or theory. And so what I think he's great at is shaping a microcosm, right? And usually when people want to write about an idea, they'll pick a topic, right? So this is considered the lobster. He could have written

an essay on animal ethics and covered everything, but he continuously zooms in. He says, "Let's focus just on lobsters. "Let's just do lobsters in Maine. "Let's do a Maine lobster festival. "Let's do the Maine lobster festival in 2002." So that's an extremely tangible thing. You can see that festival through his eyes and through all of the details

you understand this point of view he has on animals and consciousness and pain, and you absorb it easier because he's not trying to preach to you something. He's zoomed into a microcosmic experience and he's letting you come to that conclusion. - Yeah, this is a mistake that writers make all the time is like from all the time I spent teaching Rite of Passage, maybe the number one lesson that I had to give people was just

write about fewer things. Get more specific in what you're writing about. There's this strange way that by getting really specific, like the Maine Lobster Festival of 2002, that you're somehow able to actually write about more. It's less but better. Yeah.

There's another theory that's really helpful for this. So Umberto Eco, he wrote a book called How to Write a Thesis. And he has something called the Pick One Volcano Principle. And so he'll have a student come to him saying, I want to do a dissertation on volcanoes. And he says, all of them? Like, you need to narrow your scope. So the student comes back and says, OK, I'm going to do active volcanoes in Mexico.

And he says, "Okay, fine, but there's still 48 volcanoes. Are you gonna actually research all of them and include them all in your scope?" He says, "Pick one volcano."

So the student comes back and says, "Oh, I'm gonna just do a thesis on Popocatapetl, a specific volcano that has multiple types of eruptions, it has history with the conquistadors, it has a very intricate notification system." And so by zooming in on the right volcano and looking at it in detail, you can actually understand many volcanoes, but you get the essence. It's not just about zooming in, but it's zooming in on the thing that is emblematic of more than it is.

I want to zoom in because you said something earlier about maximalism and minimalism. Can you break down how the maximalism of David Foster Wallace happens? I actually really want to know because I love his writing. But once again, that's the kind of thing that kind of feels subjective to me. And you're like, no, dude, it's pretty objective. There's a way that he does this.

Yeah, so I want to look at two paragraphs, two different paragraphs by him. Maybe we should look at a minimalist paragraph first, and then we'll do a maximalist one. Is that cool? Yeah. Okay, so this gets into kind of a theory of paragraphs. So E.B. White says that the paragraph is the atomic unit of composition. And you see a lot of modern online writing advice, and they're saying, oh, a paragraph should only be one or two or maybe three sentences because attention spans are shorter. But it's

but the paragraph is actually a really important, almost a punctuation device, because every time you see a paragraph, that should be a distinct idea. And when you shift to another paragraph, you're signaling that you're shifting topics.

Does that make sense? Yeah, yeah, yeah. And so I think it's helpful to understand like what is the fundamental nature of a paragraph? And this gets back to the five paragraph essays because in school we learned this very rigid formula where it's state your topic sentence and then three supporting points and then say the topic sentence again. Like it's incredibly boring. It's so boring, yeah. And so...

I think we need like fun rules for paragraphs. And if I were to try and make a fun and flexible rule, it would be every paragraph needs a hook and it needs to end with a punchline. And with that, you have so much variety. There's different kinds of hooks. And when I say end with a punchline, it doesn't mean that it needs to be funny. It could be sad or onducing or disgusting. What a punchline means to me is that there's something in the subtext that is like an explosion. And that's how you want to end a paragraph because the hook, it

it gets you interested to read it, and then you're rewarded. And if you get that loop right, the reader's gonna wanna read another paragraph. If you break that loop, they're gonna say, "Why am I gonna keep going? "Reading is friction." - Yeah. - So I wanna read a simple paragraph and just kinda show you how it kinda fits that format. - Cool. - So here it is. This is-- - Here, let me read it. - Oh, you wanna read it, okay. - Yep. - Let me preface it by saying this paragraph is about him explaining how luxury food is actually kinda gross sometimes.

The point is that lobsters are basically giant sea insects. Like most anthropods, they date from the Jurassic period, biologically so much older than mammalia that they might as well be from another planet. And they are, particularly in their natural brown-green state, brandishing their claws like weapons and with thick antennae a whip.

not nice to look at. And it's true that they are the garbage men of the sea, eaters of dead stuff, although they'll also eat some live shellfish, certain kinds of injured fish, and sometimes each other. All right, break that down for me. - Yeah, so if you look at just that opening sentence,

The point is that lobsters are basically giant sea insects. So what we think is seafood delicacy are actually massive bugs. So the thing that is expensive is actually incredibly gross. So right there you have something that is, it's kind of unexpected. - Yeah, crazy. - It's crazy. So you're gonna kind of read to figure that out. So that's sentence one. Sentence two is saying, well, they're gross because they're ancient Jurassic creatures. Three is they're gross because they're just

And then the third is that, well, they're gross because they're cannibals, right? And so the idea that lobsters are cannibals is like such a good example of why they're gross. And what's interesting is he says, sometimes each other. On its own, that's a very innocent phrase, but in the subtext, he's saying the word cannibal without saying it. And so that's what I mean by paragraphs wanna end with a kind of subtextual explosion.

Right. So basically what you're seeing here is what? It's that he opens with a sentence that both explains what we're going to read and there's a bit of mystery to it. Then he kind of flows through a few different points. It's the strongest example at the end and it kind of rings. You're thinking about lobsters eating each other and it's kind of a reward for reading that paragraph. And remind me, what are the meta takeaways that we can now apply to our writing based on what we're seeing here?

Yeah, I think when you're editing, it's a good lens to go through each paragraph and say, is that opening sentence? Is there a sense of mystery to it? Can you consider it like a micro hook? And then look at the last sentence and say, is this a powerful example? And there's a lot of different ways to do that last sentence, but...

There in that last sentence is there something in the subtext that makes you go Wow And what are people not understanding like when you read just crappy crummy writing? What is the thing that people are missing about the nature of the paragraph? Yeah, I think so often you'll see a paragraph that starts somewhere and then midway through you realize oh, wait a second This is what they're talking about. So they're not setting the frame. So you don't even know why you're reading the thing you're reading Yeah, so this minimalism

Yeah, that's a in my mind, that's a fairly straightforward paragraph. It's four sentences and he's kind of listing off examples. It's pretty traditional in terms of what a paragraph is. And talk to me about David Foster Wallace Maxwell. OK, so he does this in a lot, if not all of his essays. But, you know, people say be specific, add an example. Right. That's like very general writing advice. And so, you know, in the second paragraph of the essay, he has 37 different examples in one paragraph.

Wow, right? And so, you know, this is him when he first arrives at the festival and he's used the phrase floating I essays. So basically he's just rendering, he's taking notes on every single thing that he sees and then he's going, he's integrating it into prose.

And there's a real arch to how you can include 37 details without being boring. - In one paragraph. - In one paragraph. If you were to just list them off, I saw a parade, I saw lobsters, I saw this, I saw that, it could be very repetitive and mechanical and boring. But there's a few ways that he's able to include this much detail and make it dynamic. And I think the obvious one, so this is what's color coded, is that

17 of these details are general. You'll say the big parade, carnival rides, lobster ravioli, right? They're kind of like non-specific nouns. But then he has 16 examples that are absurdly specific. So he's mentioning different events, the William D. Atwood Memorial Crate Race or the Maine Sea Goddess Beauty Pageant, right? They're one of one instances that you wouldn't know that thing existed unless you were there.

So by mixing, by going back and forth between the two, you're kind of at a syntax level, you're always shifting. And then sometimes he has numerical examples too. He does this a lot. He's telling you 25,000 pounds of fresh caught Maine lobster. He's telling you the attendance is 100,000. That's just one way. If you really wanna show something, gather all the possible details, but then show them at different resolutions and mix it up.

I love this because the experience that I have reading David Foster Wallace is Venkatesh Rao once wrote that reading David Foster Wallace is like looking at a 15 megapixel camera, just giant resolution when your brain is like a three megapixel camera. And it actually hurts your brain. At least for me, it's kind of hard for me to read David Foster Wallace for a long time because

He's just capturing so much of reality and so in such a small space and what's so cool about this what you're saying is it's just 37 different details that are super specific that he could basically walk around with a notebook and just write 1 2 3 4 36 37 and then what he's doing is he's squishing all of those things into one paragraph and that's how you get this kind of hyper reality effect in his writing mm-hmm and

You'd imagine walking around the festival, it's a bunch of random noise, but he's trying to cluster the details. He starts with, let's talk about the musical acts there. Then he talks about all of the attractions that the festival themselves sets up. Then let's talk about the types of food there, right? So he's creating categories of details and arranging them that way. - So what's the lesson here? Why is he able to make this interesting and engaging and entertaining when you were kind of getting it earlier,

"Ah, a list of details could be kind of boring, dude." - It's maybe a similar lesson to how you vary your sentence lengths to keep it interesting.

you can also vary your example types. As in sometimes you want a three paragraph example that's immersive, but if you do that again, that could be too much. Then you wanna compress it and say, let me just do one paragraph examples. And then after that, you can say, let me put 17 things in this one paragraph. So there's a way in which you're dynamic in the resolution of your examples. And like you said, if the whole essay were like this, it would be unreadable. You know, he only has three,

paragraphs in this 33 paragraph essay that are this maximal. And I think it works. Whereas if you look at some of his unedited essays, you can tell that he's really just trying to put his whole consciousness in the page with no breaks and it becomes unstable and it's hard to process. So I think the lesson here is sometimes you go hyper detailed and then sometimes you zoom out and tell a single story slowly.

Tell me this, how does the observational merge with the personal? So how can you observe something as a writer in a way that feels personal and unique to you? Because usually when we think of observations like this, we think of them as being generic. But David Foster Wallace has a way of doing it where you're like, I get that these are just observations, but I feel like I could only get these observations in this sequence in this style from one person ever.

So this paragraph is interesting because the middle of it is filled with observations, but it actually ends on a personal note.

He says, "Your assigned correspondent saw it all, accompanied by one girlfriend and both his own parents." So it's not just this barrage of detail. You now picture a guy and a few other people just totally overwhelmed by all of these things. And so this taps into one of the key pillars of personal writing. I think with essays, with a lot of writing, it's knowing that you want to put yourself in the page, make it something only you can write.

But it's also very intimidating right to actually reveal the details of your life into this very public medium and It's tricky to do when we don't have a good definition of what personal writing is because sometimes we'll point to the the mastery of Memoir for example and we'll see these really confessional things and we'll think that personal writing has to be these all revealing expositions, right? Mm-hmm and instead I think you can be

There's a way to put yourself on the page in a way that's a little bit more comfortable. I know when I was writing at first, I told myself I was writing personal, but I wasn't really. This definition might help. One of my patterns is experience. Then the question is, well, what is an experience?

I break it down into three things. It's biography, interiority, and then outlook. Define those for me. So biography is about, imagine there's a camera in the room that's looking at you, and we can actually see your life in high detail. We can see you moving through space. We can see you in dialogue.

We can see your body language. It's more than resume level details. It's seeing the specifics of your life. - It's like materialism taken to the extreme. These are all the things that happen, all the movements that you made, exactly how many words per minute you spoke, exactly what chair you were sitting in, stuff like that. - Yeah, and a lot of memoir is about this where

Like Charlie Bleeker, for example, she never wants to tell you how she's feeling. She wants to show you. Right. It's almost like a film-like cinematic sense of what is it like to be a fly on the wall in your life. Okay. So that's biography where interiority

is the opposite. Interiority is what's going on in my head right now that no one else can see. And you can't really capture that in cinema, but with literature you can. You can attempt to render your consciousness on the page so that readers can know what was it, what were you experiencing at the festival that maybe your girlfriend and parents weren't aware of? So you have biography, interiority, and then you have outlook. So outlook is

"I believe this." It's confessing a belief you have in the world that might not be standard. Sometimes it's a risk to say what it is you believe. And I think this is probably the pillar that most people default to and they think it's personal writing. Just by saying, "I believe this," it's not enough to do that, but you have to go upstream and show us the experiences and the interior calculations that led you to believe that thing.

And there's a lot of different ways that you can use these three things. And in a lot of cases, you then do the loop again. So because you have this new outlook, you now have different experiences that you process and it changes your outlook. So this is also a model for like transformation, which is really at the core of personal writing.

Where does this end up going wrong? Like what is the kind of writing that somebody would write? And you're like, I see what you're going for, but that's not the kind of personal that really works. Right. So I think there's a style of personal writing where you can be very gushy and sentimental and you're, you're telling us how you feel about something, but you could take off your name, put someone else's name on it and no one would notice. Right. So it's, it's not necessarily about a depth of feeling that is part of it, but you also need to

put us in the context of your specific life, right? We want to see the things in your life that are different from any other potential writer. Biography, interiority, outlook. Break them down for me. Yeah. So this is showing how these two essays, Consider the Lobster and Shooting an Elephant, they both have these three things, but they execute them in very different ways. And I scored both of them a five out of five.

So this is another example on how there's no one way to satisfy the experience pillar. And with Consider the Lobster, I'd say 50% of the paragraphs have no personal in them. It's just him geeking out on etymology or like animal consciousness. But, you know, we saw that second paragraph here. We see him online at the festival. Then we see him in a cab talking to somebody. So we get these flashes of things that

only David Foster Wallace saw, right? They're inside his head and he's telling us. And then these green, these are moments where he's kind of questioning, like, wait a second, like we're boiling lobsters at scale a hundred at a time. Is this weird? And it's not until the very end, right? That he actually shares his outlook.

A bad essay would have just started with his outlook and would have been very preachy, saying, "You shouldn't eat lobsters." But he doesn't do that. He just kind of renders his experience. He thinks through some ideas. And by the end, he's just asking the question, "Hey, is Gourmet Magazine promoting a medieval torture fest?" So he does the work to get you to his outlook. So "Consider the Lobster," you know,

Shows experience in one way, but in shooting an elephant you can see that all of these cells are shaded, right? He starts with biography. Then you see him Reflecting he's kind of understanding how he's in this weird position where he both hates the Empire that he works for and yet He's also really pissed at all the locals in this town who torment him endlessly. So that's kind of interiority, right? He's showing this conflict in high detail He's just not saying oh, I'm I'm annoyed. He's being specific and

on how he's annoyed. And then the action continues and then elephant escapes and it flipped the car, ate all the fruits, killed the cow, killed the person. So you have this action developing and then you go back to interiority and outlook. And this is when he realizes like, wait a second, I need to shoot this elephant and I am a cog of the empire. And then there's another action. There's another biographical scene at the end

where he shoots the elephant, and then at the end you see him reflecting on that again. So right here we're seeing how Orwell is different from David Foster Wallace. Can we get into specific sentences and paragraphs and break that down? Let's do it. Cool. So I think we can look at how these two essays are structured differently because Shooting an Elephant, you know, it's written by Orwell. He writes, I mean they both write a lot of fiction, but Shooting an Elephant at least is pretty traditional in terms of how you build a conflict where

consider the lobster has a different organizing principle. So let's look at shooting an elephant first. So this graph shows from start to end how Orwell moves from intention to obstacle, to consequence, to climax. And this is in shooting an elephant. Right. And so in the beginning, you understand

who this character is, right? So you understand that he's stuck between an empire he doesn't want to serve. He's trying to help the people, but the people don't like him either. So he's in a hard place. So intention really is about modeling the character, whereas the obstacle comes, right? So in this case, an elephant escapes, it ravages a bazaar, and we know that this character is now going to have to deal with the situation, right? And so

It gets to a point where this elephant eats all the fruits in the town, flips a car, kills a cow, kills a person. Orwell has a gun and he's chasing the elephant and it's running away. And he says, oh, I probably don't have to shoot it. But he turns around and he has 2,000 people behind him and they're all expecting him to shoot the elephant. I think it was Aaron Sorkin who said at any point in a movie or a TV show, you should be able to stop

Stop the film and just point who is the character? What do they want and what is getting in their way? Right? What do they want is their intention? What is getting in their way is the obstacle and I think intention and obstacles really simple two words that you can use to just make up a story Hey, I just graduated from college. My parents called me they said hey, love you so much We're gonna surprise you in town tomorrow. We'd love to see your new place and

unfortunately, you and your friends from college have been smoking a bunch of weed in your place and you have this couch that just completely smells like Mary Jane. Uh-oh, mom and dad are coming in 24 hours. I don't want my parents to know that I smoke weed. They're going to be here. What do I do? You can just make up stories by just using intention and obstacle, intention and obstacle. Boom, boom, boom, boom, boom. Yeah, I think those two work

really well as a kind of generative tool. And you can do the same with obstacle and consequence. Okay. Right? So an obstacle is really well done if you don't know how the obstacle will resolve. Right? A good obstacle has multiple consequences.

paths that it could take. - Of course. - So for example, in this case, we're seeing Orwell and he has 2,000 people behind him and they're all pressuring him to shoot the elephants, yet he doesn't want to, right? And so there's two roads. He shoots the elephant and he makes the town happy and he's guilty.

Or he does the right thing and he doesn't shoot it and the town hates him. And so for a good period of the essay, you're suspended in this unknown. Is he going to do it or not? And then finally, the climax is when one of those two consequences are realized.

And am I right to say that the better that you do in tension, obstacle, and consequence, those three are basically creating the tension that then leads to that climactic moment being a big thing? I think so. Yeah. These three work together to really, these are the three kind of criteria of a conflict.

And they work in sequence too. You have to, I mean, of course you can have previews of consequences in the beginning, but you generally, an obstacle and consequence aren't going to really register unless you really understand the character and their intentions. So Orwell kind of structures

his conflict in a very traditional textbook kind of way. But when it comes to nonfiction that is idea-based, like Consider the Lobster, it's not as clear what is the intention, what is the obstacle, what is the consequence. And something that I've noticed in a lot of nonfiction is that it's organized by questions. Basically invisible questions that are behind the page. And it's really helpful to think of essays and writing in general as questions and answers.

And there's a really interesting, well-known kind of situation with startups and businesses. You have founders come in and they present all these features, these amazing features. And then someone bursts their bubble and they say, yeah, but what problem are you trying to solve? Right. Those are a bunch of answers without questions. Right. And so when it comes to an essay, you want to figure out what is the question I'm actually answering. So whenever you start a piece, your first draft, your question is fuzzy.

and therefore your answer is going to be fuzzy too. But that whole thing is really just a vehicle to help shape a more specific question. So through redrafting, you get a tighter and tighter question until the answer just is there because you're asking the right questions. And with David Foster Wallace, I've

dissected two of his essays according to questions. And what he actually has is a whole hierarchy, his three tiers of questions beneath his structures. So if we can go through that really quick, consider the title, consider the lobster. And so behind that descriptive call to action is a question, is eating lobsters wrong? Simple question. But then he organizes the whole essay into three kind of mid-level questions.

The first is how have lobsters gone mainstream? That's where he talks about the festival, talks about how 200 years ago it was prison food, but now it's luxury. The second section he's asking, is it all right to boil sentient creatures alive?

And then the final question, which is only the last three paragraphs, he's asking, "Is Gourmet Magazine promoting a medieval torture fest?" So you have these three big questions that are organizing the piece. The first one he doesn't say out loud, but these two in green, he does say out loud. But what's really interesting is if you want to really break down this section,

he has a final level of questions that organize the paragraphs. So this is an outline of that whole section, the bulk of the essay, and I can explain to you what's in there through a series of questions. So, first he asks, "Is it all right to boil sentient creatures alive?" Then he asks, "Well, how does PETA protest this?" Then he asks, "Can lobsters feel pain? How do cooks feel about this? Are there alternative methods to do it that are more humane? How do we interpret their lack of endorphins? And do they exhibit preference?"

So each of these questions have maybe two to three paragraphs where he answers it, but he's not even asking these questions out loud. But what he's able to do is in the subtext, he's effectively planting these questions in your mind in one paragraph, and in the next paragraph, he actually answers it. So what should we be doing as we're writing? Should we be thinking about these questions deliberately? What I'm hearing from you is every piece will have one driving question. Maybe that's worth making clear. But at what point

do you get to a place where it's like, hey, this is a good thing to study that you just sort of have in the background of your mind versus if you want to write better, get clear on this, write it down. Where does that boundary exist? Yeah, I'd say like a good starting point is always at the top of your document. You can ask yourself, what is the core question I'm answering?

And then throughout your draft, maybe you have 2000 words, look at what you've written and try and figure out like, oh, there are actually sub questions. Oh, these first three paragraphs, they're actually answering this question. These next three paragraphs, oh, they're answering that question. I don't even want to be discussing this question. I can remove it. So it's like through drafting, we have these fuzzy questions, but through

reverse outlining and making those questions explicit, you can decide if you want to double down on a question or cut it. So what you're saying is basically tension and the thing that engages people is a series of good questions followed up by answers and that's the resolution. So at the level of peace, you have

the tension of a driving question and the resolution at the level of a piece, but it also shows up continually. But here's the key point. You don't need to constantly pose questions by asking those questions explicitly. You can have them be implicit, but as a writer, if you're looking at a paragraph or looking at a section, it's good to ask, what is the question that I'm trying to create here? - Exactly. So whichever of the two frameworks you use, whether it's intention, obstacle, consequence, or question and answer,

They work in the same way. You're establishing something and something is unknown and then you move to resolving it. So you do that with characters or with ideas. - I love that. - Yeah, and that's why I called the element tension instead of storytelling. People think stories are about characters, but there's tension behind characters and there's tension behind ideas too. - Right, cool. - And so what I wanna look at now is the hook of shooting an elephant. And hook is one of the patterns in tension too.

And so I'll just read it really quick. "In Moulmein, in Lower Burma, I was hated by large numbers of people. The only time in my life that I've been important enough for this to happen to me." And it's a really simple sentence, but what's interesting is that underneath this descriptive sentence, there are invisible questions. There are at least three questions in the subtext here. One is, why is George Orwell in Southeast India?

Two is, why is he hated so much? What did he do to be hated en masse? And then three, since the essay is called "Shugen Elephant," how will this hatred of him lead to the elephant situation? So it's interesting that when I'm trying to write a hook for a piece, sometimes instead of trying to write the hook itself, you need to ask yourself, what are the important questions that you want to bake into that sentence?

I think it's interesting that in this hook is almost an outline for the essay. Underneath this hook are questions that he'll use to organize the content of the essay. - Cool. - I can give you another example too. - Yeah, go for it. - Okay, so this is a Charlie Bleeker example. So I'm working with Charlie Bleeker, she's a very personal writer, she's working on a memoir right now, and we're working on this prologue, and the hook of the prologue will be the first thing anybody reads, right? And so it needs to have, it needs to really draw you in.

And we've rewritten this 17 times or so. And this is kind of loosely where it is. She says, usually I only steal one or two pills of Adderall from my best friend, but this time I stole the whole orange bottle. Man, that's so good. Yeah. And so what are the invisible questions in this? Who is she and why is she stealing Adderall? And then two, what is the relationship with her friend where she's stealing? And then three, is her friend going to find out?

The reader might not even know it, but you've incepted those questions with that very short sentence. And those questions are now the outline of the prologue. First, you understand who she is and why she needs the Adderall so bad. Then she explains the relationship with her friend. And then we get into the situation where her friend confronts her. It's often hard to start with the hook.

You need to actually write the chapter or the essay or the draft, see what are the main pieces, turn those pieces into questions, and then bring those questions up and put it in the subtext of the hook. So tell me this, people talk about rhyming in poetry, but do you see it show up in prose?

Definitely. So rhyme is one of these 27 patterns. It covers alliteration too. It's any sort of phonetic resonance. And you can really hear it when you read it out loud. So Hunter S. Thompson, for example, he would pay his friends to come over. He would give them beer and say, "Read my work out loud." He would hear how it sounds coming from someone else, and he would go, "Thank you very much," and he'd go rewrite it. Really? So there's a whole dimension of editing that you do with your ear.

And, you know, I can't say for sure how Orwell composed this, but this is in the first paragraph and it's just rings. And I wanted to read it out loud to show like what poetic prose can sound like. And you can see the color coding, right? So you can track how the S's move through it, how the B's move through it, the F's. So here it is.

Huh, so break this down.

So when it comes to poetry, you have the end of the line and the beginning of the line as obvious points where you locate your rhyme. But with prose, it's not like that. You have an ocean of sentences and it's all interline rhyme. And so the way I see it, something you can do is when you're editing and you're trying to find the right word, sometimes you try and pick a word saying, what is the right meaning? But you can also pick synonyms that have a certain sound.

Right, like referee happens to land on a football field with laughter. You could have picked different words, but these all have an F. Right, GH creates F. Exactly. Football field, referee, laughter. So you think that this, that process is basically happening intuitively whenever you're listening to a piece of writing. I think even when you're reading it silently, your inner ear is picking up on prose that is phonetic.

I think there's two ways to think about it. One is you can just aim

aim to have poetic prose. I mean, prose poetry is a whole genre and say, "Hey, I just care about having melodic sounding prose." So if you want to be more strategic with how you use rhyme and prose, you can find three areas through your essay that are important that you want to spotlight. And you can edit those words and find synonyms that do rhyme. And in this case, I think it's interesting that you have all of these S's that point to hideous and then several F's that point to laughter, right? So hideous laughter

is an important moment. It's an important theme in this piece because it's laughter that drove him

It's fear of laughter that drove him to shoot the elephant. And so in order to make this phrase stand out even more to the ear, you're planting all of these words above it so that lands. Let's talk about writing process. Let's talk about where you start, where you end, how you edit. And I want to hear just from your method, based on all the writers that you've studied, all the writers that you've worked with and taught, where have you landed on process for your own writing?

It's funny because when it comes to patterns and the end point that we can edit to, I'm extremely specific. I'm saying we have these 27 patterns. Every essay should aim to do this. When it comes to process, I refuse to commit to anything. So I don't even share what I think is a good process for other writers. And even personally, I will try to never approach the same thing twice. And so basically what that means is I'm anti-template as a

As in, I don't think I should be trying to find some perfect structure that I use every time to make it faster. And a really good story tied to this, or a good concept, comes from Frank Lloyd Wright. He was also an essayist. He's known for his essays. And he wrote these manifestos against floor plan factories.

So he says we should never take a house and just stamp it everywhere. Another fun fact is Howard Roark, the character in Ayn Rand's novel Fountainhead, is based off of Frank Lloyd Wright. So it's almost this like violence against mechanically reproducing design solutions. When it comes to process, I think you need to write your first draft for yourself and your second draft for your reader.

And what that means is that this first draft, it's actually helpful to not think about any of these patterns, right? Don't try and structure it. You need to chase the material down and see what it is. And so Frank Lloyd Wright, he had this concept called organic architecture. And organic doesn't mean curve. The better word really is emergent architecture. And what he means is that you shouldn't copy a floor plan from site to site. He said,

He says every site is dramatically different and you're bringing a unique client to that site and you need to actually see what unique opportunities emerge from this place. And the same is true with writing. Every idea is different. You want to understand what is possible.

It's a brand new thing. So the first draft, you're just trying to discover what the thing could be. And then once you tap into the essence, once you know the specific question you want to answer, that's when you start using essay architecture to make that idea more legible and resonant with readers. So what I'm hearing you say, because a lot of your method is actually very analytical, very objective.

But what you're saying here is, yo, but at the beginning, just, just write, get stuff out there. Let it be emergent. Let it be intuitive, whatever. You can worry about that other stuff later on.

- Right, you need to, I think it's worth honoring both mindsets. At the beginning, it's like a mystical process and it's better if you don't know what's gonna happen, where it's gonna lead. And so that means essay architecture really is an editing philosophy. It's what you do when you're stuck with a draft and you know something's there, but it's not coming through right. These are a set of tools that can help you get it over the finish line. - Hmm.

There's a phrase you shared years ago, which is practice analytically and then perform intuitively. And I think nobody takes that idea seriously enough. I really think that is maybe the most important thing. I think we all want to be intuitive performers and writers, but you don't just kind of get there by feeling it out. You need to kind of practice and study.

And my favorite example for this comes from the Grateful Dead. They're a psychedelic bluegrass improv band. They go on stage, they rarely have a set list, and they figure it out. And the songs take on different roads each time. If you were to ask Rick Rubin, what makes the Grateful Dead so great? He would say, well, you know,

they're not chasing perfection. They're chasing real moments. They're following their intuition, right? And that is all true. But if you go back five years and you watch Jerry Garcia, he was a competition banjo player and he would be in his room practicing scales for 10 hours a day. And if you look at Phil Lesh, he was going to music school, studying music theory with trumpet.

And the early Grateful Dead, they kicked out Bob Weir because he wasn't practicing hard enough. He was partying too hard and he had to work his way back into the band. It's funny that this band that is the ultimate symbol of improv and intuition, they were practicers. They studied, they were analytical. And it's almost like you need to be in your head to understand how it works in order to eventually get out of your head. Dude, I love this. I mean, the way that I would think about it, this is obviously a metaphor, is

You think analytically, you really study it, and eventually the things that you learn get absorbed into your DNA and you perform those things without even thinking about it. I mean, look at Stephen Curry, man. Stephen Curry is working on the form of his jump shot. He's actually, if you watch him dribble before the games, he's practicing all these little dribbles. And then you watch him during the game and it's like, choo, choo, choo, choo.

It's super fluid and he's in flow. But this is the key point, that that begins with a very analytical and a very intentional method of practice. And a lot of people look at people who are in flow and who are super intuitive like the Grateful Dead once they're actually performing, but what they don't see is all of the

very purposeful work that went into it beforehand. Right. So Jack Kerouac, he's the symbol of, "Oh, I'm just going to take amphetamines and write for 10 days straight in a massive scroll and it's all just going to pour out of me." And then you just try and do that from day one and like, good luck.

But mimicking that, but what we don't see is that for 10 years, Jack Kerouac wrote a million words in very plain and traditional novels. And that's just not part of his lore because that's not exciting. But I think with all of these figures who are kind of have this like in the moment genius, it's because they tried to understand the basics before they kind of went off the rails. Hunter S. Thompson rewrote every single word of The Great Gatsby so it could feel what it was like to write a great novel.

And you see the same thing when I was in Barcelona and there's a Picasso museum in Barcelona. So I was like, oh, this is going to be cool. You know, I'm going to see a bunch of funky, strange Picasso paintings. And this museum was focused on his early work.

And so I'm in there, I'm checking everything out and I'm expecting kind of the mid 1910s Picasso style. And no, I got super formal, much more classical styles because what Picasso was doing is he was learning from the greats, trying to copy the greats, trying to mimic the greats. And it was only way later that he developed his own strange style. And so even with somebody like Picasso, who you look at,

And I mean, I sort of feel like that sort of seems like a third grader painted it. I don't really like Picasso that much. But what I really admire about him is that he mastered the basics. Like what you're saying about Kerouac, he was doing that sort of traditional painting before he ended up inventing his own form and style. Totally. And the way I see it is

My project, Essay Architecture, is an attempt to map out, here are the boring basics that are probably worth studying. And by kind of having patience to little by little practice these patterns, eventually you can forget them. The whole point is to forget the system, but it's a method to help you understand what are the basics you need to shore up. Yeah. So how do you edit? I think when a lot of writers edit, they basically start from the top

and then they have a very general sense like, "Oh, maybe I'll find things as I read that I need to fix." But instead, you can have a razor lens just on one pattern at a time. So you can say, "Let me just go through my draft and let me highlight the moments where the voices pop through." You're not looking at structure, you're not looking at your idea or your thesis. Let me just get the color orange and I'll highlight every single moment my voice pops through.

So it's a very reductive lens, but then you can look at the color coding and say, oh, this paragraph is doing great, but I have this whole section with no highlights. That means the voice is disappearing in here. Let me rewrite those. - How do you get more of spirit, sound and sight in the editing process? - It can be really hard to take like,

a three out of 10 paragraph and then edit it, edit voice into it, very often you just need to rewrite the paragraph with some of those patterns in mind. I really think that voice is something that you, you can reverse engineer voice on famous pieces of writing like we just did, but

something is different when you are trying to make it. So it's not like you can, you know, mechanically bring voice into something. You need to just try the paragraph again and then you look at it and you say, "Oh, okay, well, it's working. Maybe this is a more phonetic paragraph, but there's no imagery. Cut it. Let me try it again." So you can almost just keep looping and rewriting that paragraph. And then when you analyze it, if it works, like, okay, that's a keeper.

Why is voice so important to find? Why is it such a crucial part of writing? Why can't we just communicate the facts, the information? I think reading is inherently friction.

So unlike a movie where it just comes at you and you can see it, there's a process and we're obviously literate and familiar with it. But if you're reading thousands of words, that is mental work. And so in my mind, when you consider voice, what you're doing is you're shaping language in a way that it is

more compelling for us to digest. Like when you're more concrete than abstract, we're seeing in visions, we're hallucinating. And that's kind of like the level of our nervous system. So I think there are like biological reasons why you want voice to be visual, why you want to hear it with your ear. I think when you have attitude and personality and emotion in the subtext, you get the sense, oh, there's another person at the other end of this.

I think a lot of good writing is almost putting on somebody else's brain for some time. Seeing the world how they see it, thinking how they think, and being like, whoa, that's what it would be like to be in that person's brain. You know, like, I really like reading essays about New York because I lived in New York for so many years, and people will write these essays about

Hell's Kitchen or about what it's like to be a cab driver. You're like, oh my goodness, I lived in New York for so long and never had that thought and that thought and that thought and that thought. And all of a sudden you're like, wow, that's what it's like to be in somebody else's mind. Right. And I think there's something profound there where maybe writing voice should be to honestly

transfer your consciousness onto the page. Or if you're thinking about, well, who's the audience for this? Who am I speaking to? There's some use in that, and I think that will affect voice. But maybe the ultimate place to be is that what is on the page is pretty one-to-one with what you think. And

And what I find is that there's a lot of stress when you're, not stress, but there are stakes when you're publishing to something that will be read by 100 people or 10 people or 10,000 people. And those stakes will very much slant the voice. And so this is why the favorite part of my writing practice is actually not essays. It's logging.

And so I have this practice called logging. And so for example, already today, there are five or six things that I've written down through my morning coming here of just thoughts that I had that were mildly significant. And I wrote them out in full sentences. And then I published them onto an unlisted page on my website. So everything that I thought in the last three or four years is just kind of in these monthly text files. One of the core features is that I can't get any feedback

on my logs, right? They're all in just a massive wall of text for the month and you can't like or comment on it. And so I'm not writing with any self-consciousness. There's no expectation for these things to perform. This is very much just a rendering of what is going on in my mind and me putting it on the page. And because I have that practice, I feel like

I'm pretty attuned to what my voice is like when nobody is watching. So by keeping that practice live, you can put me in a high stakes scenario and I have it in my bones like what it should sound like if I'm being authentic. One of the things I've noticed about what makes for good writing is

You can almost think of there's like these layers of consciousness and what writing is getting you to do is to dive deeper and deeper and deeper and deeper into what is happening in your mind. And the surface level tends to be fairly cliche. It tends to be things that mimic the way that other people talk, the way that society thinks. But as you get down and down into the way that your brain works and the way that you think, you end up

finding these trap doors of surprising ways that you're thinking ideas that actually you've been having but for whatever reason they haven't been made conscious and

I think when voice works really well is when you're articulating things that you've always known and always felt, or at least for some time, but you haven't actually made explicit. And it ends up having the feeling of surprise, this sort of epiphany of like, oh my goodness, I always knew this. I can't believe I just...

Put that to language, I've known this for like two years. - Right, it's been in there, but it hasn't had an outlet yet. - Yes. - And I think-- - And that feeling of surprise is how you can basically verify that it is something from within and not from the cultural consciousness.

Being aware of who is could potentially read your work is the easiest way to shut down your voice And so when I started writing five years ago I did it under my full legal name and writing was not fun because I was aware of who could potentially see it And so what I did is I dropped my last name. So Dean is my middle name Michael Dean is a student It's a half pseudonym. I'm really Michael but once I kind of was able to shed that I

writing became so fun and the voice became unleashed and unchained. But then I grew, you know, a small audience and now I have other people watching me. And so I burrowed again and there are now secret corners on my website where I don't publish this stuff, but you could find my thoughts on death and religion and politics and all of these ideas that I might not blast the audience of essay architecture, but they're all there. And that's where I do a lot of my

Kind of like the edge of my thinking happens in places that are less public. And so it lets it grow and marinate. And then eventually it pushes public. What I'm taking from you is people say that, hey, you're writing under a mask. That's bad. What you're saying is just put on a thousand different masks and you'll have a thousand different voices. And somehow the stew of all those things comes together to create a voice. Fernando Perso had 75 pseudonyms. Really? Yeah.

And I think the idea is through creating these contexts where you can explore different sides of yourself, they come from the subconscious to the consciousness, and then eventually they become integrated into the main self that you do present to the world. So I think it's always important to have both of these processes going. You need to be sharing yourself and be out there and be highly public, and you need another more secret part of your practice to let ideas marinate before they make it there.

Tell me about your method of getting feedback like a comedian. So there's a few different ways you can ask for feedback Often when you talk to an editor you say hey help me fix this tell me what to do But instead you can say I just want you to read this take it all in and when you like something Highlight it green give me an emoji say lol Make it very simple because I want you to be as granular as possible. And if it's confusing tell me you don't like it. I

So I ask for flash feedback, your most granular reactions. And the key thing is I give each reader a separate document so they can't see each other's responses. And then what I do is I have a master document and based on

what they like, I color it green in my master document. If they don't like it, I make it red. And as more readers come in, I am changing the hues. So if three people really love something, it'll be a dark green. And that's a really good signal. If three people independently said this same sentence rocks, it's like, okay, well, that's a good sign. Like, I definitely want to keep that. Maybe I move it up, right? Anything that's red, that's what you need to fix first. And then sometimes, you know, I'll have five readers come in

And I'll notice there's a stretch of five paragraphs and no one has said anything about these paragraphs. There's nothing notable for one person to even make a remark on. So that's a sign to me that, hey, maybe I should cut this or compress it. And this is effectively how standup comedians

kind of like test jokes on a circuit. They will take the same joke and they'll, independent audiences, they'll make sure that it works in every single setting. And so by the time they do the big show, they're pretty confident that it's gonna work because in 50 isolated instances, it worked well. - And what do you do when you read? Do you sit down and read the whole thing or are you highlighting the first time you read? What do you do to actually see what's going on in the writing and get x-ray vision?

I think it's always good to try and read the first draft kind of normally. Just try and take in what's happening and highlight things that are remarkable, that are articulated well. And then I try and do a second pass where I only go back to the highlights. And then I try and deconstruct and figure out why is this working? And I try and come away with a principle that I can use in my own writing. And so, you know, by doing this a hundred times, you have a big list of

principles and then the question is well how do you then organize that into a framework and so that's kind of how essay architecture came about it was all of my scattered insights and then I used Christopher Alexander's framework of a pattern language to help me think of how you organize all these things when what is a pattern language explain that so that's a book by Christopher Alexander in 1978 and

Basically, it's him trying to map the universal patterns across all of architecture. So he has 253 patterns. The first one is how you scale cities between two to 10 million people. The last pattern is how you hang stuff in your living room. So it goes big to small. And so you can actually map any pattern just based on that hierarchy.

But what makes it a language is that you can zoom into any pattern and it's not isolated. It is connected to other patterns. So if you're on the pattern of a greenhouse, for example, it'll say, well, in order to solve this, you need to understand shelving and storage and compounds and vegetable gardens, right? So it's this interconnected web of design constraints. - Yes, and the thing is,

with these patterns, once you see them, you kind of can't unsee them. So one of my favorite ones is that a balcony that is less than six feet deep, people just won't use. And you just see it all over the place. If there's a balcony that's four feet deep, there's no furniture out there. If there's a balcony that's eight feet deep and you gotta assume some warm weather or something like that, people be out there hanging out, having a good time. And it's not like you need to have a balcony that's more than six feet deep. That's not the point of a pattern language.

The point is, it's something that's a little bit more than a heuristic, but much less than a rule. And there's a bunch of them. And by combining them, you begin to get x-ray vision into how a space feels. And that's what you're saying with writing, that if you vary your sentence lengths, then you end up with rhythm. If you paragraphs could be a whole language.

element of patterns and you would say, hey, you want the maximalism in the middle, you want the minimalism on the edges, maybe you want the introduction to ask a question and the final sentence in a paragraph to answer the question. And now all of a sudden is you can look at any piece of writing through these different patterns. And all of a sudden, if you have a problem, you have a series of

patterns that you can use to then solve those problems. Exactly. And Alexander says that within a pattern, you can have a million different solutions in the same pattern and none of them will be the same. Because it's just the constraints that are recurring, but the solutions can be infinite. What are some of your favorite patterns for writing?

Well, I'd say all of essay architecture is a map of patterns. And so I'd say that's a difference between my system and Alexander's where he has 253 of them, I only have 27. And my case is that you can find all of these in any essay. And within each pattern, there's a whole dictionary of types. None are better than the other.

So I think writers can have subjective preferences on their solutions, but I'm trying to adopt a lens that all 27 of these patterns are equally important for me to master.

And of course there are sub-genres of the essay. So you have the lyrical essay, the personal essay, the historical essay, the narrative journalism essay. And if you're in a sub-genre, they might say that, "Hey, we actually only care about 15 of these 27 patterns." And sure, there are contexts where it makes sense to focus on a sub-genre, but for the ideas that really matter, that you want to be timeless, I think something is timeless if it

if it touches on all 27 of these because these touch basically all the corners of the human psyche, right? So this is why I love essays is because I think when you write essays, you become a writing generalist. So, you know, if you're writing fiction or poetry, you're specializing and isolating, but

But an essay is a fusion of the soul and the experience of a memoirist, the research of a journalist, the thesis and argumentation of a philosopher, the hook and the call to action of a marketer, the stakes building of a novelist, and the image and the rhyme of a poet, right? So an essay in a way, you are these six different archetypes in one thing. And regardless of the context, it has this universal and timeless appeal to it.

Walk me through if I want to I really like we're saying about tension tension as opposed to storytelling Tension is the thing you want to focus on and storytelling is very common way to create tension So if I'm looking at a piece and I'm like, I just don't feel like there's that tension And now I see that the three patterns are conflict threads and hook How do I look at my essay through those three prisms in order to improve the writing? I

Sure, so conflict where you have the intentions, then the obstacle and the consequence, right? Whereas threads,

you can actually have multiple conflicts overlapping. There should never be a point in your essay or your story where there's no conflict being advanced. And then hook is effectively the opening conflict. These are three different patterns where you can look at your essay and say, oh, wait a second, I have a good overall conflict, but the hook is wrong. So for example, in "Consider the Lobster," I only gave the hook a four out of five, I believe, 'cause the first paragraph didn't really set us up for the core dilemmas that were coming.

Now walk me through a title mystery target phonetics, right? So Mystery in my mind. This is about the title having these invisible questions that draw the reader in whereas target is about

Are you properly putting the thesis into the title? It'd be like if we called Star Wars Wookiee Warriors. You're not optimizing the title around the right thing. You're not distilling it correctly. So you want to both build mystery, because that brings people in,

And then you want the target to be right because when they leave your essay, you want them remembering the right thing. We're remembering Star Wars, not Wookiees, right? And then the last one, Phonetics, this is about just how the title sounds out loud. Are you actually coining this as a piece of catchy language that can be remembered? I think there's a lot to do with things that

are shaped with like rhyme and rhythm are more easy to stick in your memory. Jane Austen, love phonetics, pride and prejudice, sense and sensibility. So when I hear- Snakes on a plane, Star Wars. Snakes on a plane. So what do you make of snakes on a plane? Because with that, it seems like it's less as strong with phonetics, I might be wrong, but better in target and mystery.

Well, I'd say with phonetics and there's a whole science to phonetics where every single syllable technically has three things. It has an onset, a nucleus, and a coda. So snake is a s and then a a and then a k. So even though snake and plane have a s and a p, they both have a a a. Snakes on a p. Like they both hit that a. And it's also symmetrical. Snakes on a p.

Plain so on and are actually simpler syllables. So on only has a nucleus in a coda It's like rest on and then a is only a nucleus. It's only a single one of those three Oh, that's a good answer. So I mean you can go infinitely deep on any of these patterns, right? Like there's a whole universe in studying phonetics But the question is what are like what is the the practical basics in terms of how you shape the sound of a title? All right

Alright, so we've done one with form, we've done one with idea, let's do one with voice. Let's talk about sound. So I look at my writing and I want to look at sound, repetition, rhythm, and rhyme. How do I look at a piece of writing through those three layers? Well what's interesting with sound is you can calculate if something is phonetic even if it were complete gibberish.

So sound, it has a part of it that is just about how it sounds out loud. So repetition is this idea of loopage. Like how often are you using a phrase to loop and expand an idea, right? Whereas rhythm, it has to do with the containers for your ideas and how you change them over time. And rhyme has to do, it's similar to repetition, but it's not about stamping the word exactly. It's about...

two nearby words that have a similar sound. - And do you want more repetition, more rhyme, more rhythm necessarily, or no? - No, well, so all of these rules kind of in isolation are wrong, right? So like--

- It's the ensemble of them. - It's a whole composition. Like you could say, ask GPT, "Hey, just make this whole thing incredibly rhyme-like and it'll be ridiculous." And so I think a part of the rhyme pattern, for example, is are you just splattering it everywhere? Or are you finding the high leverage moments where rhyme matters and you're using it there? Same with rhythm, right? You don't wanna just alter for the sake of it. You want to expand ideas that deserve to be zoomed in on, and then you wanna do spotlights on transitions.

Another good example is word choice. There's a whole pattern for word choice, which is funny because that seems like such an infinite thing. But within words, there are different ways you could think about it.

William Zinsser has a quote that says, "clutter is the disease of American writing." And so just from that quote, you say, okay, concision is important. I want to be economical. I want to make it very clear and legible, right? But then there's precision, which is kind of clashing with concision. Precision is let me pick the word with the best undertone, right? There's a reason why you would use illumination instead of light.

If you're optimizing for concision, you would always pick light. But there are some situations where you would pick illumination. And then there is another criteria of word choice called inventiveness, right? Inventiveness is about rare words, foreign words, words that are typically that a reader might have never seen. But if you put it in the right

context and you give clues, you can actually introduce them new words. So there might be a case where I use alpenglow instead of illumination, but I can, if I'm talking about mountains, alpenglow suddenly makes sense. And so this is just an example of how any rule kind of on its own is incorrect. You need to understand that there's a tension between these different rules that all work together and in different scenarios you use one or the other.

I want to do this. I want to go through some maxims and mantras that you have. Okay. And we'll just go one by one, okay? Sure. Understand the fundamentals of your genre before you develop taste. Yeah, I think we're in a culture that is very excited about

taste. Taste matters. Find your taste. It's a one-on-one way of finding value. And I think that is important, but it is possible to get taste blindness, where you are determining like some specific pattern or some aesthetic that is important to you. And you're so focused on that, that you totally lose sight of the basics. And an example is like

you know, it's a trend to make these big spectacular buildings that are curved and made with software and you'll see them presented. And it's like, but where do you walk in? Like, where's the front door? Like you're so obsessed with a certain aesthetic that you just lose touch with the basics. So I'd say that instead you should start with the basics and then from there your taste will arise. So make this concrete in the world of writing for me. Like,

Like what do writers do wrong writers think like you sort of had people in the culture but writers think this when actually you think that when it comes to Writing taste. I think people will cling to specific solutions. They'll say I always use the hero's journey I'll always use this kind of personal experience where I'm doing confessions or I will always

write in these short blocky paragraphs. People just develop this idiosyncratic sense of preference that might work for one random context. But I think by zooming out to the fundamentals, you'll realize you don't have to always use the hero's journey. There's actually

50 there's hundreds of different kinds of sequences you can use and there's patterns that underlie all of them And so if you understand the patterns then you can break out of that rut So I think when you're stuck in a taste you're very much like limiting yourself to doing the same thing over and over Let's do this. The main point of composition is to establish readers trance. Mmm

So I feel like reader's trance is when the writing is so well done that they're in this state of concentration where they just kind of tear through until the very end. The default state is someone will read a paragraph or two and say, I'm not interested. Reader's trance is when you're

Just you can't put it down. You're not even supposed to be reading you missed your train because you're just so locked into this thing and I think that's like a big aspiration everyone wants readers trance But I think the value of a system like this is you can you can make a technical definition You can say oh actually by combining these three or six patterns you can you can create this trance thing and so I'd say

You can think of readers trance at three scales. One of them is at the scale of tension, right? You need to have something that's unknown so that they keep moving forward to resolve it. Humans hate the unknown. And so that's kind of the root level of the trance. And then you have this mid-level where you have paragraphs as like your microstructure. And we talked about how you have mini hooks and then

kind of punchlines. And so that's a sort of repetitive trance-like thing with each paragraph. And if you break that, you might go, "I'm not sure what I want to read anymore." So you have something that is very stable. And then while this is doing, you have a separate element that is constantly changing. You're changing the length of your paragraphs or the immersion of your examples. So you have this like mid-level part of the trance. And then at the micro level, it's about, you know, imagery and sound, right? Are you actually

rendering language so it's a concrete hallucination that you can see. It's like when you're reading Harry Potter and it's like you're so immersed in it. So I think that's like the deepest level of the trance where you can see it. It's almost like you're not even reading, you're just kind of internally rendering.

concrete hallucination that you can see. That's a badass way of describing it. Okay, let's completely switch gears. Let's talk about what's happening in AI right now. Great. Why don't you give us your big picture overview on how exactly you're using AI to write? So what parts of writing are you being influenced by AI? What parts of writing are you like, nope, I'm still doing it solo. And let's just take a snapshot of here and now, and then we'll end up talking about the future later on. Great. I'd say I'm

still stubborn in the sense that I still insist that I write every single sentence. But that said, I am using AI a lot in the process. For example, I might have a draft that's 3,000 words. I'll say, hey, tell me this in 1,000 words. Tell me it in 100 words. Tell me it in one sentence. So it's a great way to

Structurally compress your drafts and understand the essence of it right like sometimes I'll realize Whoa, hey, I did this thing in 10% of the space and I get the idea Like why am I rambling and then I'll just rewrite it myself with shorter context So AI shows you that something is possible that you didn't realize was possible, right? And you'll say I just saw that a I did it like a 7 out of 10 now I know that I can make it like then I do it from scratch on my own and

what models you're using, what interface are you using for different projects that you're working on? And then how exactly are you actually prompting the models? I'm really interested

in the specifics here? - I'd say most of what I do is within ChatGPT. I feel like it has so much context and memory. I also have my own custom GPTs in there that even though some models are better at different things, there's something nice about having a single environment for everything. Sometimes I'll use Claude for little experiments, but I'd say I'm mostly in ChatGPT. And then there are all sorts of different parts of the writing process that you can use in this one space.

And so we can go into each one. So one of them is research. And who Googles anymore? I mean, I think you can effectively put in a draft of yours. And then you can say, well, I'm looking for examples of this concept between the years 900 and 1200 that exemplify-- you can give extremely specific constraints for the kind of references that you might want to weave into your essay.

Right? And then often that is just helpful in identifying a person or a time period or whatever it is. And then I can manually research that added into my essay, but it's almost like a highly specific matchmaker where it has read everything. And so I can put in my draft and it's able to, if I just prompt it specifically, it can connect me to the figures that make sense for my thesis. Okay. Let's talk about the other pillars now. How do you use GPT for other aspects in writing?

Yeah, I think there's like a gradient of scale, right? So you can use AI at the structural level to compress things, to make outlines for you, to help understand your main idea. Then there's kind of a mid-level tier

where you can be throwing in paragraphs saying, write this paragraph in five different ways. And you're just seeing what comes out and that sparks ideas. And then there's down to the word where you can have a custom GPT that has like 20 different things you want in each definition. So what I'm getting at is that

It really covers the full stack from the big picture thesis of your piece down to precise word choice. I think AI is good for all of those things. And what I think it's not quite good at yet is you can't give it a list of instructions, right? You can't say, hey, I want you to keep in mind these 25 writing principles, write an essay on this. It's not able to synthesize all of that into a work

that's cohesive. It often like short circuits and focuses on rule four and ignores the rest. Right. Sometimes too, if you say, give me a Hunter S. Thompson impression, it'll just start talking about Las Vegas, right? It'll mix the voice of the person with the content of the person. So tell me this, where is AI currently not good that it's going to get really good at?

And where is AI currently not good? That there's something fundamental about the tool and the technology that it's just never going to get good at. And humans are always going to have to do that. If you think about what an LLM is, it is binged.

trillions of words, and it's almost spatially mapping every sentence in some sort of hypercube, right? So every sentence has a position. And so then when you prompt it, you're basically shooting a vector and you're finding, oh, here are the similar things that have been said before. And it's able to kind of serve you up something that is likely or probable. So LLMs are amazing association machines. What they don't really have is a true sense of judgment of quality.

There's a funny example where someone used AI to rewrite Paradise Lost, a classic epic poem. And then they had AI judge the two. And one was slop. And it preferred the slop. The slop one is obviously a masterpiece. So it doesn't really have a consistent barometer of quality. It is...

for getting feedback it is great at sparking ideas but if you want like a ground source of truth it is almost completely random i am i'm developing the uh a system to test prompts and i can tell you that like from some of my testing that it's like 50 accurate which is almost like a random guesser um so i think over time we'll be able to have more reliable

like quality is, I mean, my project is basically about this on how can we more reliably have these LLMs score patterns, right? So I think that's one thing that's missing. The other thing that's missing is AI models just write so fast, right? So they write it like 400 words per minute. And so even if you're letting it think for a minute or 30 minutes or whatever, as soon as it starts writing, it just blazes forward.

400 words per minute. It has no ability to recursively say, oh, actually that, now that I'm on sentence eight, sentence two is not as good as it could be, let me start over. Right, it can't assess itself, it can't challenge the original prompt that was given. So I think this is more of maybe a product

thing that could be solved. But I think if you let these things loop in on itself and think for hours on how to compose a piece of writing, it would be a lot better. We're basically telling it to think on the spot. When I think of AI, the word that comes to mind for me is volume and possibilities. It's basically, I like to prompt AI to basically get a lot of possibilities at once. And then what I can do is I can say, okay, this

this is a menu of things that I could potentially do, but I just asked it to rewrite one paragraph 10 times. I'm not, obviously I'm not going to copy and paste that same paragraph 10 times, but I might see something really good in the second output, the seventh output, and the eighth output. Things that actually I see could kind of weave together and then I'll end up rewriting it. But just seeing the possibility when I didn't see that before just really sparks momentum for me. And

increases the range of ideas that I can possibly have. - Right, it's a creative ideation machine. And I think even if it, I think if you use AI, right, it doesn't necessarily make it faster, but it can make it better because you're exposed to more of what something could be. Of course, there are use cases where things can be dramatically faster, but I think if you're working on a piece you care about, you're actually exploring so much surface area that you never would have if you had to do it manually.

And tell me this, for learning to write and improving your skills, why do you think that we should ban chat GPT in schools? I think software is always changing. It's getting easier to learn on yourself. And so schools could be really the last place where we protect the fundamentals. And I think, sure, GPT can now make outputs incredibly fast and eventually very high qualities, but it's not just about the output.

there is something that changes when you actually commit to writing and editing. When you're slow cooking in an idea, your mind changes, right? It's kind of like a person who makes a chair doesn't just have a chair. They become a chair maker and they see their whole reality in a different way now. And so I really think that, you know, kids of course should be using

chat GPT to ask questions, to learn. But I think if you don't have a context where they have to actually go through the slow act of writing, they're not, something is going to be off in how they conceive ideas, edit ideas. There's a good phrase by S. Kelly Harari, editing is not rewiring words, it's rewiring your synapses. And so the idea is

The reason we need to protect writing and editing in schools is that it's not about the outputs, but through writing you change and you develop. And you develop skills to critically examine yourself and your identity and your ideas. And kids are gonna resist this like hell. And I can give you an example. So I went to architecture school

It's 2009, my freshman year. And we had software where we could, for example, build a 3D model and then automatically cut our plans and sections through it. They would not let us use software for the first three years. So we had to manually draw every single line. We couldn't use computers to get perspectives. We had to actually calculate the angles to create three-dimensional views. I had to hand trace Vitruvius or hand...

transcribe him. The most extreme example is I had like a four by five foot map of Manhattan and I had to render it strictly out of dots. So for over 100 hours, I was just drawing dots to make a photorealistic image of Manhattan. And that's like that's printer. That's what printers do. And they're making humans do this. And, you know, I resisted it. I said, I need to be using the technology. And

And so I transferred and I went to a modern design school. And only then did I realize that the slow and inefficient way of learning actually changed my mind. And now that you put me in software, I was so much better, right? Because I knew, because I drew through every line of the building, I knew how buildings go together. Where someone who's automating everything from the start, they're just divorced from the fundamentals. Let's talk about the future now. So you mentioned that

in 15, 20 years, you feel like a lot of our canon will be created by AI. Well, that's an extreme example. And one of the reasons I'm thinking that is because every three or six months, some little feature comes out and it's like,

"Oh, no, well, it can do this." But you know what? Humans are always gonna still have this advantage, no matter what. And then three months later, something new comes out that proves that wrong, and we just continuously keep shifting it forward. And so I'm thinking, you know what? Like, what if you just took a long view, like 20 years? What is the most extreme possible situation of what AI can do? And you just kind of think of any possible scenario to just go to the extreme.

And it's a helpful thought experiment because if you still would write in that writer's apocalypse, then you don't have to stress out over the little month by month improvements. - Well, also you think about what happens if this tech is taken to the very extreme, what remains true, and then how do I just focus on those things? - Right. - I think that's the thought experiment. If you're scared about AI as a writer, you say, okay,

we're not going to quibble with, oh, is this technology there? Is it not there? We're just going to say, of course, it's going to get there. Like we already see that the arrow of technological progress is vertical enough and it's going in this direction. So we can assume that AI is going to be there. Now, okay, we'll assume that, say, AI will write poetry that

Humans like more than human written poetry or study that just came out that AI is better at making memes I saw that and only the very very very best meme makers are Better than the AI I mean you just have to assume in five years as gonna be better at making memes than just about every single Human being on planet Earth. Okay, so let's assume that those things are true now and

what skills are still going to be valuable. Yeah. That's how I like to think about it. Right. And so it's worth trying to render a really specific

Image of what that extreme state could be. Yeah, so I kind of have three ways to think about it So first on that like AI poetry example I think that study is different because it's it's being taken by people who don't really know poetry If you took the world's best poets, they would look at this and be like, well, of course that slop and this is timeless work But I think the first flip we have to make is that the future is not filled with slop but I think the quality will be

unimaginably good. It could get to a point where it's not that it writes better than us, it's not that it writes better than the best living writers, it will be extraterrestrially good, right? And there will still be a volume problem, but it's not a tidal wave of slop. It's there is so much work that is so good and that will be equally disorienting, right? If there's just infinite

I wonder if the major lesson that we've learned so far is that humans actually just kind of like slaw. And the reason why I say that is think of the average beach read. It's already the kind of thing that would be produced by an AI. And in basically every single field,

The kind of art that the people at the very top of the field like tends to be pretty different from the kind of work that the masses like. You know, I think of a lot of people who we talk to really like David Foster Wallace. David Foster Wallace is pretty popular, but he's not as popular as more mass market people who a lot of big name writers say, ah, it's not that good. You see the same thing in music. I mean, how much of pop music is sort of four chords and, uh,

People who are musicians are like this is not good music, but people love the music and AI Will be able to produce that stuff very easily Well, I think we're already at a point where AI is very close to producing things to capture mass tastes Yes, right. I think like it's probably far less people who will appreciate like stuff that is like canon worthy and in a lot of ways I think

Don't trust everyone's opinion equally. So it's like I'll trust the opinions of three people I respect more than a million people I don't mm-hmm and so to me that study on a slop is kind of irrelevant to me because I don't even know who the readers are Right, you know, so I'm not so indexed on mass taste. I'm really chasing like what is Beauty what is quality?

And if you have a trained eye, I think it's important to talk to the people who are trained in these fields and understand how they perceive the quality of this stuff. But I think the point I'm getting is that we're gonna get to a point where even the masters will look at this stuff and be like, whoa, this is better than anything I've ever seen. And I think there's a world in which eventually it will both impress the masters and the masses.

And so the first point is that the quality could become unimaginable even to our best human artists. And the second point is that it might actually be insanely popular, right? People use the example like, well, you know, chess beat humans years ago and humans still prefer, you know, watching humans. I think there's a world in which like some of the most popular creators are AI. You know, it could be extremely relevant.

The thing that stands out there that we haven't spoken about is new forms that are birthed by AI. I mean, this is just what we see all the time. What new technology does is it takes certain things that have always been true, that have been very laborious, and it automates it. And then new things begin to emerge, new forms of art, new ways of doing something. And you just have to assume that there's going to be

kinds of writing that we can do with AI that we were never able to do as human beings. I'll give you an example. Yeah. The autozine. What's that? Right? So you can imagine a company putting out a prompt, right? And then you'll have 10,000 writers each answering a prompt. And then an AI can ingest that all and then synthesize it into a single work that features ideas and writings of the writers. So it's like

It involves, it's almost like a hive mind where never before have you been able to synthesize 10,000 voices into a single glowing object of quality. - One of my theories is that niche humor will be done very, very, very well for AI. So for example,

If you, let's just assume that AI is an eight out of 10 in comedy. So it's not going to be as good as the world's best comedian. But what I can do is I can prompt it and say, hey, give me humor that's based on the culture of Austin, Texas, where I live from 2020 to 2025, the five years that I've lived there. Give me humor that's just about that. I would-

offer two revisions. Yeah. One is that you won't even have to prompt it. So already have your data and know what kind of humor you would like. Right. And so make me laugh. I want to watch comedy right now. Yeah. It's just, I mean, there's that. And I think you also said, oh, it'll be eight out of 10. Like, I think it could be like 50 out of 10. Like,

I don't think there's a limit to AI producing the funniest thing you've ever imagined. Like that's kind of part of this thought experiment where it's like we set these limits of like, oh, well, it'll never really get that good. I just really want to challenge people to say, oh, it could be extraterrestrially. People say it's an alien intelligence. It can just be good beyond our imagination, not just intelligent, but creative and every dimension at full force. You know, the vibe of what you're saying reminds me at the very beginning of COVID and I

I had a phone call with a friend who was tracking really well and he was like, "We might have lockdowns." Right. And I was like, "People not being able to go outside? No way." And he's like, "Yeah." And I was like, "Well, okay. I think that we can just do one week or two weeks to flatten the curve."

He was like, no, no, no. This could be many months or maybe even years. And I was like, there's just no way that could never happen. And you know, if you remember how that whole conversation unfolded, it was like, okay, we'll do two weeks. And then we got a month and then we got three months and it was sort of just, it, it, it extended beyond what anybody thought possible.

And this whole AI thing is really similar where it's like the technology is getting better. We're like, well, okay, this is sort of as far as it can go. Let's just hold on to it. And what you're saying, it's like, no, no, it could really, really get better than you can possibly imagine. And that's kind of what we just saw with the lockdowns a few years ago. Right. It's scary to imagine this. I think that's an underlying thing.

premise behind a lot of our assumptions. It's scary to imagine the world being locked down for two years. It's scary to imagine synthetic writers that can tap into all of our emotions and just be unbelievably good. So I think in that moment of fear, we're very easy to focus on some technical limitation and be like, well, it's never possible because of this. And therefore humans will always do that. But what I'd say is that

I think people are locked into thinking about the LLM paradigm, right? And I think people talked about the scaling laws and now there's, oh, we can let it think for longer now too. And I think even just on those two vectors, I think there are limits. I think we're going to hit a block from that, but I think it can incorporate so many other architectures and transformer breakthroughs that

LLM will just be, LLMs will be one element. It's an orchestra of things. Basically like in the origin days of AI, they were trying to use symbolic logic to create intelligence and it just didn't work.

And then they went to neural networks and now it is. But I think, for example, if you just layered symbolic logic now on top of these neural networks and you had some fusion of the two, suddenly now this thing has a sense of math and physics and quality. And people, I think, aren't anticipating what could emerge from just new architectures. You have to imagine that's being heavily researched. How do you think about the intersection of personal writing and education?

because you're, we're saying two things here. The first is personal writing is really important. And I think that you can assume that there's going to be certain details about your own life, your emotions, the things that you've done that at least for the foreseeable future, AI won't be able to replicate. It just, you know, your life better than the AI as well.

And then at the same time, AI is becoming super powerful. It's almost becoming this like super intelligence genius. So what do you see in terms of getting that personal onto the page and using AI to enhance that? So I started a list. I wrote out like 250 cornerstone memories of mine. And the whole idea is to write them out in paragraphs. And what this does is it takes the deepest experiences that have happened to me and it makes them machine legible, right? Every single day I'm...

Trying to at least write a few logs of here's what happened to me today. So people say like oh language models don't have your memories That's only if you don't write your memories right and so I'm kind of in this I

I have hundreds of thousands of words of my detailed personal thoughts in the last three years. And that means if I were to build a replica of myself, it has so much context on all the nuances of the personal thoughts that I've had, where if you have just a single corporate essay, it's kind of a very, you know, off sense of who that person is. So I think the first point is that the quality will be beyond what you can imagine. It will be terrifyingly good.

I think the second point is that it won't be obscure, but it will dominate commodity markets and it will be popular on social media and will even replace our canon, right? Like it will be very relevant. And I think the third kind of reframe of the writer's apocalypse is that we could have

replicas that mirror our own mind and memories to an alarming degree. As in, if I have a million words of my day-to-day thoughts and I've written up all of my core memories, this thing will be very much like me. And so I think something interesting might happen where once people realize like, oh, whoa, like there could be a lot of value or even just creativity or interest in having this kind of like mirror of me,

you could all of a sudden find the masses saying, "I need to start writing because the more that I write, the more personal and aligned this thing can be." There's a reason to keep writing every day and sharing your experiences and your evolving beliefs, and then this thing evolves with you. What you're saying is that the benefit of writing nonfiction becomes to make your thinking legible to machines so that you can then work with the machines

to articulate those ideas. - Yep, ideas and also your own personal experiences, feelings, and memories too. Like the more of yourself you can make machine legible, the more you're able to have a thing that you can work with that is like a mirror of you, and also the more you can leverage yourself

So I have this funny thought experiment where you and I, if we each have our own replicas, we can effectively be in 24/7 conversation, 400 words per minute. So that means, you know, synthetic Michael and synthetic David, we generate half a million words per day.

And you could be having that conversation with everyone in your audience. So you personally are generating half a billion of words per day, which is a ridiculous scale. Of course, you're not gonna read any of that. And so it would then have to create these digests for you or real world David to make sense of all of the collaboration opportunities and new ideas. And then you would have to decide, well, okay, well, how do I want to then deal with this information? But it just,

that creates a situation where it's leveraged so beyond what like our own processing can handle. It's exciting, it's alienating, it's weird. - Why aren't you discouraged by this? - Discouraged in what way?

You called it the writer's apocalypse. AI is here. It's going to overtake us. And you don't seem discouraged. You seem very excited and at peace with this. Oh, why am I not discouraged? Yeah, exactly. Well, so I think I've just accepted it. I've run this thought experiment. And I said, if this goes as extreme as it could, these three things will happen. And then I asked myself, would I still write?

Would I still write if the quality was good, if it was insanely popular, if I had a replica? And like the answer is yes. I would still write through all of that. And to me, that's a good signal that there is nothing that could come out that would dissuade me.

because I am very much aligned with the reasons of why I write in the first place. Of course, there are currently external reasons why I write. I'm trying to make a career out of it and it will be disrupted, I'll have to figure it out. But even if all of that just flops, I still write to make sense of my life, to document my memories, because I love being immersed

in a puzzle, right? Writing is like a linguistic puzzle. And I enjoy the process of every month watching myself get better, watch myself overcome challenges. So it's a thing I'll be doing my whole life regardless if the technology is around. So I figured, let me always kind of remember the core reason why I do it

And no matter what gets disrupted in the outer world, I'll figure out how I can plug into it and I'll keep doing this. So there's a world where I have replicas that are writing 2 million words a day, handling all the business side of things. Great. That means I can spend 100 hours locked in and editing the one idea that really matters to me. Yeah, that got me thinking about why do I write? And what I get an enormous amount of satisfaction from is there's something in my brain, whether it's an idea or a story,

that feels fuzzy and then through the process of writing it becomes clear and once it becomes clear it becomes educational becomes entertaining it becomes pragmatic it becomes poetic and there's a utility to it but it's also like a joy to read and then there's a way that

Because of how it's written, it's like easy for me to remember. It's simple. And it gives me like a lens, like a clear lens or a framework that I can use as I go tackle the world. Right. So it clarifies my thinking. It hopefully helps other people. It's hopefully fun to read. And that process is.

I love doing. It's just one of the most enjoyable things for me. Now, one of the things that's changed is now I use AI to do that in a way that I didn't use to do it. But it's still, I'm still doing the exact same thing of trying to say, what is an idea that I have that has a lot of potential? What is a story that I need to share that would really resonate with other people? And now instead of doing it alone, I'll ask AI, I'll say, hey,

I'll go in argumentative mode on Grok, which I have so much fun with. Whenever I try, I'll just start arguing with Grok. We just get me and the Grok lady. We get in crazy fights. I'm like, what the heck is wrong with you? All that sort of stuff. And what I'm asking her to do is help me poke holes in my thinking, which is super useful because I want to know

what assumptions I'm holding that aren't true. And then also, okay, now I want your help, lady. Can you help me structure this idea? Give me a clear one-liner and help me say, all right, this is what I'm really trying to say. Because so often, I mean, in my mind,

I'll want to say something, but I don't even know what I'm trying to say. And writing helps me figure out what I actually am thinking, what I'm actually trying to say, and what's fundamentally true. Right. So people have said that writing is about one half expression and then one half communication. Right. There's what you're talking about is like, I need to communicate with myself.

I need to actually write out my fuzzy thoughts, edit it, and oh, I've actually taught myself some valuable principle that's going to help me. And I think even if you just offload some agent to write millions of words for you, it almost doesn't count if it doesn't loop back and actually affect you. Yeah, because I'm not really interested in AI doing the writing or thinking for me.

That's not really fun. I'm interested in finding the truth and writing something of quality, and AI can definitely help me do that. So then I'm down to work with the tool. It's almost like through learning these concepts and composition, it's not just about getting better essays, but I think as you learn these principles, you round out your own psyche.

Think about pop writing, personal, observational, playful. I think people tend to usually lean towards one of

the three and there's a big weakness. But then if you actually take that weakness seriously, like, hey, I'm typically not personal, but I really want to focus on that. You don't just change the essays, but it changes your whole lens as you learn these tools of composition. So my hope is that, you know, it used to take 10 to 20 years for someone to master writing. But if you use AI to act as a coach that can find this like proximal zone where you're stuck,

and you still insist that they write all of their own sentences, it can take that 10 to 20 year process and hopefully compress it down to six months or a year or at the very least, it'll make their rate of learning

Steeper steeper so they stick with it what I'm hearing from you is you're getting a much more comprehensive picture of where you're strong and where you're weak so we romanticize oh, but school you learn how to write everyone thinks of but you get a teacher who's like the dead poet society that cares so much about you so passionate I mean I went back as My parents are moving so I went back and I looked at a lot of my childhood essays and the grades and the teachers are like Hey, here's some grammar. They cut stuff off. They're like be

And they're like, whatever. But I think what you're saying is actually the level of resolution that you will have on the quality of your writing, not just grammar, but how your ideas are structured. How well do you tell stories? How well do you communicate tension? And then all of that gets modeled inside of a system that then you work with from kindergarten through 12th grade or through college or for your entire life. And the AI has a very low

legible ability to see what's going on in your writing, but also remember from all the years of the things that you've written. And it becomes a companion that knows you intimately. Right. Yeah. So I think education has been this mode of mass dissemination where every kid learns the same thing at the same rates. And I think curriculums of the future will be

Self-driven and so I think the way to teach writing in the future is you start by doing so you actually write a draft and Not until you upload your draft. Do we know your weaknesses and know what's worth learning? You know, so I think it becomes a very active Form of learning where you just are always on the edge of drafting and you're always learning about stuff That is very relevant to what is gonna fix the thing you care about. Okay, let's do this. I

We started with essay architecture, 27 different parameters, and your thesis is that writing quality is objective, not subjective. So based on everything we just spoke about, restate that thesis now that we have context on how you think about writing. So there's an interesting tie-in between this and our whole AI discussion, where if we're saying that there are an objective series of patterns that underlie everything,

then we can also train an AI to see and score those patterns. So after I've made this framework, now I'm going through and I'm reading hundreds of classic essays and I'm scoring each pattern one through five. But the next trick is to now see if AI can match how I score.

right? And so in the end, I'm creating this orchestra of different AI models. So for example, scoring your experience, it might be CLAUD37, but then scoring your arguments, it might be GPT45. For RIME, it might use O1Pro. And so I have this sandbox now where I'm able to test a thousand different prompts at a time, and it's telling me how accurate each pattern is. So I think one of the problems with

ChatGPT is it's always bullshitting you and you never know how much. You never know if it's 5% bullshit or 50% bullshit. So each of these patterns will have a confidence score, right? Like, oh, this pattern is 95% accurate. This one's only 72% accurate. But over time, I'm tweaking the prompts so that this becomes a reliable proxy of quality. And so what that means then is that

this becomes a kind of mirror where you can be uploading your drafts and it's showing you what your strengths are and what you need to work on. And so the hope is that this

Helps make learning writing way more approachable. That's one of the hardest things to do I think it's one of the most important skills to teach I think it's the best way to learn composition like you can learn composition better through writing than through music or through architecture or through film writing is so accessible and democratic so my hope is that this can make learning writing easier and It can teach you these fundamentals

without replacing the act and so you can become this well-rounded writer who now have access to these profound tools and you can kind of be a fusion between these really classical baseline skills and these crazy new possibilities with technology. Rock on. Thanks for coming on the show. Yeah, this was fun.