On Writing #3

Periodically I like to gather various observations about writing, and share my perspective. Last time was in honor of my trip to Inkhaven. This time will be in honor of the announcement of Inkhaven #3, which I encourage everyone to apply to. I doubt I will be able to usefully be an advisor, but you never know.

This is not the ‘here is my core process’ post, although there are hints throughout as there always are. I’ll do that at some point.

Previously in series: On Writing #1, On Writing #2.

Table of Contents

1. You Still Got It.
2. How Scott Sumner Writes.
3. How Scott Alexander Writes.
4. How Jasmine Sun Writes.
5. How Various Famous Writers Write.
6. How Nabeel Qureshi Defines Great Writing.
7. Quickly, There’s No Time.
8. If At First.
9. Writers Have A Harder Time Influencing, But It Can Still Be Done.
10. It’s Not (Only) The Incentives, It’s (Also) You.
11. Beware The Fetish of the Desk.
12. How Orson Scott Card Writes.
13. Doing The Math Is Fun And Supererogatory.
14. Brevity is the Soul of Wit.

You Still Got It

I used to be like this more often, constantly doubting first if I had it at all, then later if I still got it, then finally if I’ll blow it.

Hitting post used to be absolutely terrifying, every time.

It’s no longer routine for me to freak out but I’m not zero percent like this now:

Dean W. Ball: By the way if you are a young writer and not routinely freaking out about “whether you still got it” you aren’t doing it right. That anxiety is what drives you forward. Anton Leicht: i don’t know anyone whose writing I love whose work isn’t the product of defying ‘it’s so over, i’m washed, no one will ever read anything else i write again’ over and over again Dean W. Ball: yes, Beethoven spoke of this too Nathan Lambert (fixed it?): By the way if you are a writer and not routinely freaking out about “whether you still got it” you aren’t doing it right. That anxiety is what drives you forward. Dean W. Ball: Agreed! But I figure all the older writers know this :) Andy Masley: Literally every time I sit down with a new thing Dean W. Ball: Almost literally every week I wake up and say to myself “everything good you’ve ever written is behind you, and it wasn’t even ever that good and now you have no more juice” and then I sit down to write

What about me?

I worry that I’ve had or used all my good ideas, that I’ll be repeating myself.

I worry that I won’t recapture the magic that goes into a really great post.

I worry that I missed the key windows of time to say the important things.

I worry I’m getting too old and my brain is slowly not working right anymore.

I worry I’m going to start making big mistakes and look like an idiot.

My real anxiety these days comes not from ‘I don’t got it’ but from ‘I could blow it.’ It stems from the opportunity, from the stakes of the games I’m trying to play.

Whereas I guess if I ‘don’t have it’ then there’d really be nothing to sweat, you know?

The worries mostly apply to me for my ‘effort posts,’ where I’m trying to assemble a clear thesis and create something that will last. For roundups and weeklies, I don’t much worry about it. Those are more like ‘clocking in.’

I notice that most of the time, when I do really have something, I stop worrying.

I draw the parallel to being a competitor in Magic: The Gathering. I would worry I didn’t have it. When I was really clocked in I didn’t worry about not having it, I worried more about blowing the chance when I did have it. So it’s a similar pattern.

Is the anxiety actually useful? Does it drive you forward? To some extent yes. In order to write well you have to reinvent, to do something you haven’t done before, and you have to struggle and push through bad ideas and false starts and so on, and be intolerant of things that aren’t good enough or could be better.

But I don’t think it’s necessary, at least not for everyone. You just need something pushing you forward, some form of dread about not putting in every last bit of effort. You can be driven forward by opportunity or risk, rather than anxiety about ability, and I’m sure there are other similar tricks.

How Scott Sumner Writes

Scott Sumner looks at Virginia Woolf on writing advice, and offers a list of his own.

Scott Sumner:

1. Do a lot of reading—that’s where ideas come from. Read in all sorts of fields, even fields you might assume to be useless, such as philosophy. But also try to be an expert in at least one field.
2. Do a lot of writing, as writing helps one to think.
3. Think of blogging as a form of conversation. When you converse with people, you presumably are aware of your audience. That should also be true of blogging—think about who you are writing for.
4. Even a great novel or painting is a form of conversation, Artists are reacting to what other artists have done. A blogger just starting out should probably not be presenting grand theories of everything, rather one should start by reacting to what others are discussing.
5. People complained that I often “bury the lede”. So, I guess you shouldn’t do that. Give people an idea of where you are going if you want them to stick with the post—at least until you have enough of a reputation that they can trust you to go somewhere interesting.
6. Avoid a self-consciously literary style. Write as you’d speak.
7. Ask yourself: Is it possible I’ll regret this? How might this post be misinterpreted? You might want to set it aside for a day, then reread before posting. When I do so, I often end up cutting out things that might have brought me grief.
8. Personality flaws are exposed in blog posts, but writing also makes it easier to fix these flaws. Keep in mind that your critics are usually right (apart from trolls.) If a smart well-intentioned commenter says the post was unfair or mean-spirited, it usually is.
9. Words don’t just have literal meanings, they have connotations. The term ‘ignorant’ doesn’t mean stupid, but it has come to have that implication. Don’t think to yourself, “I should not have to tell readers that just because I claim a certain politician is an idiot, doesn’t mean that I view his supporters as stupid.” You do have to say that, even though logic would suggest it’s unnecessary.

In other words, because people are stupid it is important to reassure them that you don’t view them as stupid.

I especially endorse #2, #5 (I’m working on it), #6 and the core of #9. You do not have to say it, but you do have to either say it or accept that the connotation is there. There is no universal law saying you have to disclaim every connotation you don’t fully endorse, but you do want to leave them there on purpose.

How Scott Alexander Writes

Scott Alexander offers a bunch of writing advice. Paraphrase summaries follow, then my reactions are nested.

1. Avoid micro dishonesty. Don’t try to smuggle in your point, hide the weaknesses, go with the slightly misleading introduction, or pretend things turned out a way they didn’t. The readers will notice and it won’t work.
   1. Agreed. Scott Sumner’s point on integrity is similar.
2. Avoid cliches, and treat them as opportunities to say a cool thing instead.
   1. Agreed when it would be cliche in the bad way, but not in the good way.
   2. That might sound unhelpful. What I mean is, don’t do the unfun straight man repetition thing, but playing around or using them like normal words is fine.
3. Follow writing rules until you understand the real underlying bad thing, which makes you good enough to get away with breaking the rule. Give yourself particular rules to follow that match what you need to fix.
   1. Agreed, but don’t spend too long on that mountaintop.
4. Untangle your sentences, just say the thing, unless you have a good reason.
   1. Agreed. Tangled is fine if and only if you know why you’re doing it.
5. He’s against explainers.
   1. My rule is, only write an explainer if you would often use it as a link.
6. The five paragraph essay is awful but the basic idea behind it is good.
   1. Yep. Tell ‘em what you’re gonna tell ‘em, tell ‘em, tell ‘em what you told ‘em.
   2. This is one I have to keep working on. I keep reminding myself, at least one of the opening and the closing needs to contain the central takeaways.
7. Don’t write your blog about other blogs. Introduce new info or GTFO.
   1. It has to be something that works if they’ve read your sources.
   2. It also has to be something that works if they haven’t read your sources.
8. Write brilliantly about anything, or decently about what people care about.
   1. Basically, yeah.
   2. As Scott says, most success will be from option two. Option one is harder.
9. [A parable that one can interpret a lot of different ways.]
10. There are tons of effort posts out there waiting to be written.
    1. I mean, sure.
11. The true blogger finds unexpected opportunities to start pointless but delightful conflicts.
    1. Strongly agree. Go forth and find them. It’s like abstract mathematics, in that you think it’s all pointless, and it works better if you don’t worry about the point, but it then turns out there was a point after all.
12. The point of poetry is that restrictions breed creativity, so you can strike gold.
    1. Maybe. I dunno. I don’t really get it.
13. You need to always buy enough runway to keep reader attention. Start strong.
    1. Strongly agree, in most situations.
    2. Notice that sometimes you actually have quite a lot of runway. Use it.
    3. Also notice that once you’re profitable, you can keep going indefinitely.
    4. Well, kinda. I test that theory a lot.
14. You need conflict and mystery but it can be pretty much anything.
    1. I think of it as interest or curiosity. Probably mostly the same idea.
15. Give readers things they can’t trivially generate from a prompt.
    1. Whenever possible, this is a good idea. Need it somewhere.
    2. But also, they mostly won’t bother to run the prompt.
    3. Often you need it to have been said as a reference point.
16. Just say the thing you want to say.
    1. Yep.

How Jasmine Sun Writes

She does her homework.

Jasmine Sun: When you become a writer, here is one of the first things you will discover: with about a month of dedicated effort, you can go from knowing nothing about a topic to being able to present fluently and converse with experts. The beginning will be painful. You will sound stupid for your first ten conversations, maybe fifty. At first jargon will look like gibberish, then it will be faintly recognizable, and then one day intelligible, like fog clearing from the hills; with every additional source you read, your story will reveal itself in greater fidelity and scope. “If you do everything, you’ll win,” as Robert Caro says. There is a comforting if-then determinism to Caro’s advice: research persistence is rewarded, success is guaranteed. You chip away at your hypothesis until something more interesting is revealed: an essay is a failure if you go in with an outline and it emerges the same shape. The world is both unimaginably intricate and more knowable than it looks.

Fifty conversations or interviews about a topic seems like a lifetime to me. That’s a very different alien mode of being and way of learning. I don’t know that I’ve ever even done two. But Jasmine is doing a very different kind of investigative or exploratory journalism, based largely on interviews. Whereas I depend on learning from what is written.

It seems so weird to impose on people and ask them to be interviewed, especially when Claude is right there, even though I get asked and I often say yes. I should go to the sources more often.

Jasmine is constantly taking notes about everything. On the margin that has to be right, yet I find myself unmotivated to do it.

Another alien thing to me is a work in progress ‘living in your head’ like this:

Jasmine Sun: At some point in every writing project, it stops living in the Google Doc and starts living in your mind. Everything else you see and hear gets filtered through the frame of the essay.

To me, to the extent work lives in my head, that is before you open the doc, not during or after. There are also a bunch more things she talks about in the full essay.

How Various Famous Writers Write

Useful pieces of writing advice from several writers, from 2010.

Elmore Leonard is all about what stupid things not to do: Never open with weather, avoid prologues, always use “said” for dialogue without adverbs, almost no exclamation points, no ‘suddenly’ or ‘all hell broke loose,’ minimal regional dialect, no detailed character descriptions, avoid long place and thing details, leave out things readers skip. If it sounds like writing, rewrite it.

Diana Athill says read it aloud to yourself, have no inessential words, be highly suspicious of your darlings.

Margaret Atwood focuses on physical tools like pencils and a thesaurus that I would say you don’t need anymore. She also steps back to earlier points if something went awry, does back exercises and hold the reader’s attention and has others look with fresh eyes.

There’s a bunch more. The biggest takeaway is they have some overlaps but are all different, and often very personal things that won’t generalize. The only commonality is ‘write.’

How Nabeel Qureshi Defines Great Writing

What makes text into great art? Nabeel Qureshi says good text is surprising and also it echoes over layers doing multiple things at once, it makes you think, it changes what it means as you read it. Great writing contains themes and layers most readers never notice. It has depth that is impossible to paraphrase, contains mysteries and is open to different interpretations.

I would say these are some of the things that make art great, and yes for now we do not know how to get AI to do this. Then again, I don’t know how to create ‘great art’ either, and have yet to produce what I consider even a short passable work of fiction or poetry.

Quickly, There’s No Time

Eukaryote points out that quick take posts are often about interesting things that require little context. This can be a lot more important than all the details and editing you pour into your effort posts.

In particular I’d note that having a great hook is a lot of what gets one to go viral and get posted to places like Hacker News, and that virality is how you get your hits until you’ve built up a big subscription base.

You can of course do even better by producing the cool concept post and then editing it within an inch of its life, as in ‘taking the time to write a shorter letter.’

Some, but far from all, of my historically successful posts did that.

If At First

Is it better to write quickly and revise, or write slowly and get it right the first time?

g a b y: the most liberating thing in the world is realizing you can just write it bad. you can write it bad and make it good later. does everyone know about this?????? Nate Silver: I write like shit sometimes but actually bad advice, writing bad and revising takes longer than chilling out and writing well once you’re good and ready. Kyle Bown: I think it depends greatly on the KIND of writing you’re doing. I find if I’m writing a script, I need to just get the first draft out. There’s so many moving pieces to telling a story in that format that editing is necessary. An article or think piece is better to do it the first time.

Here is another case of Nate Silver pushing back against ‘write the first draft quick.’

If you are struggling to write something, could it be you don’t know what you think? That is indeed my experience.

Paul Sax: Excellent advice for avoiding writer’s block. Just get the draft down as fast as possible. Revise later. Nate Silver: Not meaning to be critical. This will work great for some people. I just think there’s very wide variation among writers about whether you do your “thinking on the page” vs. “thinking in your head”. I’m on the extreme opposite end where like the first two drafts are in my head, maybe with a little bit of outlining or a few snippets written down (certainly for longer pieces). Not inherently good or bad, but that’s how my brain works for whatever reason. So for me, when I’m struggling to write, it’s often because I haven’t done enough reading or given myself enough time to let ideas marinate. Just “getting something down on the page” can sometimes make that worse especially if you grow attached to your copy.

I largely think on the page, as I write. Which means I often start writing before I know what I want to say. It also means I would never write a passage in order to have an early version, or move on from a paragraph that I know is not good enough. If it is not good enough, I am not done thinking. I am not ready to move on.

I also think that part of why my process works so quickly is that I don’t train myself that it is okay to write these ‘vomit drafts.’ If you’re willing to do that, you’re committing to rewrites.

Then again, I never write fiction. If I wrote fiction, I could totally see doing what famed Simpsons writer Swartzwelder says to do, writing filler dialogue and jokes to have a first draft, so you know what you are looking at, and then fill in from there. Maybe that’s a good idea.

It’s fine to write badly, so long as you notice, as an exercise, but then you need to throw most of it out and start over. You can keep actively good bits.

The key difference I have is, for my purposes writing nonfiction, I find that trying to revise bad writing is way harder than writing well in the first place. If I realize something is bad I’m inclined to entirely start over. In fiction perhaps it does not work that way? I’m not sure.

None of this gets you out of editing. Editing starts with good and makes it great.

Writers Have A Harder Time Influencing, But It Can Still Be Done

Noah Smith is worried that bloggers are losing influence.

Part of this is that we are in a populist era, when the right is led by Trump who acts on instincts built from TV and real estate and zero sum games, and listens to whoever talked to him last, and the left is largely insular and only listens inside its echo chamber. So you can’t much influence either.

I would respond that yes this makes it harder, but you can still influence both.

Dean Ball went straight to the White House and has had strong influence on policy. When the time comes to grope around for those instincts, or for whoever talks to Trump last to talk to him last, someone is still puppeteering that, to a large degree. I’ve had various doors opened for me as well.

You can still drive the online conversation. There are those in the administration very aware of sentiments on Substack and Twitter. Keep posting. But do watch framing. You can turn the administration against you if you provoke the wrong adversarial frame.

On the left, you have the moderates who listen to the intellectuals more than ever, and no they don’t purely fold to the left. As for the harder left like Mamdani, you still see him for example knowing his YIMBY facts. These people do sometimes listen. Other times not so much directly, and absolutely they can be off in their own world, but part of the job is to make sure that world doesn’t end up in charge. So there’s at least that.

It’s Not (Only) The Incentives, It’s (Also) You

Smith’s second concern is that Substack is a newsletter platform that rewards those who maximize subscribers, so you end up isolated and not in dialogue with other bloggers and thinkers. There’s no republic of letters.

My response is a mix of ‘it’s not the incentives, it’s you’ plus skill issue. Noah Smith has over 300k subscribers. He doesn’t have to be maximizing revenue. I have a lot less subscribers, largely because I don’t maximize, because I recognize that the impact is on a power law, by bringing key things to the key people who will act on that.

I do worry about the ‘silo effect,’ where you do better if you stay in your lane. I want to write about games more, or other fun things, but that gets punished. I pay some price for the tangents I do insist upon.

I find the silo problem even worse on Twitter or other social media than Substack. If you’re giving me 50% Tweets I want and 50% Tweets I don’t want, then that’s no good, because the cost of evaluating a Tweet is high relative to the cost of full consumption. Whereas if you give me 50% articles or books I want, and 50% that I don’t, that’s great. You can totally spam my inbox with the other half, because I spend one minute discarding the bad things and 1 hour consuming the good thing. It’s only a problem if the bad things fool you into investing a lot of time.

Smith’s third concern is that AI is hurting our attention spans, causing more skimming versus less reading, and generating a lot of its own posts. Here I worry less.

I strongly disagree with him that the AI-written articles are already good. You can use Claude to generate serviceable text if the goal is an infodump and you also do a bunch of work, if you want. It’s at best serviceable and like Noah’s example would likely show up at ~50% AI on Pangram. At least for now. Such text plays somewhat better to audiences who don’t yet have much AI writing exposure.

Matt Yglesias notes he took a pay cut and gave up his Bloomberg Opinion column to write in The New York Times in order to reach the audience only NYT can provide. As someone who did successfully get one piece into NYT, I can say the same except at even higher abstraction. I wrote that article in order to write future articles that would then get a key message across. The other advantage is I had a really great editor. It was a lot of work but also a joy to have someone who cared about making my work the best it can be.

Lulu Cheng Meservey charts new sales after various exposures. National outlets did little, substacks did relatively a lot.

This matched my own experience. I got into The New York Times once and if I showed you my subscriber graph you wouldn’t be able to notice that date at all. That doesn’t mean it isn’t impactful in other ways, especially in reaching a lot of otherwise hard to reach people who matter.

Beware The Fetish of the Desk

I am heavily on the pragmatist side in terms of your setup. You should indeed as Robert McFarlane says ‘beware the fetish of the desk’ and not worry about getting the details right or keeping things straight or what not. If you find clutter is bothering you, fix it. If you don’t like your pencil, get a better pencil, and so on.

The exception is your computer and monitor setup. Invest in that, whether or not you are a writer. You want at least two monitors, with the central one being large. I use three 30-inch monitors, all three are fully used at all times, and if I did it over again I would have four but have decided it isn’t worth redoing everything to fix that.

I find that when forced to write without my setup, I am at least ~50% less productive. I’ve given myself a rule that I don’t try to write elsewhere, unless I am going to be in a place for an extended period. If so I try damn hard to get at least one good monitor.

Then again, there are plenty of writers who don’t need secondary screens, and are perfectly happy to write on laptops. For some forms I think this is fine, for others not.

How Orson Scott Card Writes

Orson Scott Card on how he expanded a short story into the full book Ender’s Game in a matter of weeks. He says he did almost no prep for the book, because his prep was his childhood, which he transferred directly onto Ender, and things like a trilogy of books about the Army of the Potomac which mapped onto the other commanders at battle school.

This resonated with me. I spent a long time thinking about related ideas, but once I decided to write the Moral Mazes sequence I mostly wrote it within a week, and a lot of my other key works that have held up well were written quickly.

Orson Scott Card also explains that when you get rejected as a writer, you should treat that the same way game designers treat such feedback. When they say something is wrong, they’re right, and they can often point towards it. When they tell you how to fix it, you can mostly ignore them, and should figure out yourself how to fix it.

Another note is that when you are rejected, the editor usually won’t even know why they are rejecting you, so they definitely can’t tell you the reason. Very true.

The last note is that editors are dramatically different before and after you sign a contract. Before you sign a contract, they’re usually not concentrating attention enough to be general intelligences and don’t know much. After you sign a contract, now they have a stake in the outcome, and if they are good at their job then they’re great and invaluable. My pitching experience has been extremely frustrating, but every time an editor actually focused on improving the writing it has been great.

Doing The Math Is Fun And Supererogatory

Roach-works notes that if you are writing speculative fiction and you get detailed, some of your readers will point out your details are wrong and be upset about this. So Roach’s advice is to not give specifics. The city is old, the sword is heavy.

To which Misha correctly responds, screw the haters, you can ignore them.

Misha is correct, if you are willing and able. It is much better to make some mistakes than to keep everything vague. However, keeping everything vague is also totally an option, if you do not want to do the research.

You can make mistakes. You don’t want to be making glaringly obvious mistakes, or have your economic values be off by orders of magnitude such that the whole thing makes no sense. And you may not need the details at all. If you want to tell a story that fundamentally doesn’t physically work, then it’s time for that heavy sword.

Brevity is the Soul of Wit

How long is too long? Asking for a friend (named Dean Ball).

Dean W. Ball: I just wrote a 5,300 word essay. Should I yeet it out as one long essay or split it into two parts? A normal essay on Hyperdimensional is 2.5-3k words. The risk to two parts is neither half can sustain itself without the other; the risk to one is obviously overwhelming readers. Patrick McKenzie: The people who run the world mostly read, and people who read are not intimidated by pieces at their natural length. If the natural length is 5k words, ship 5k words. Ben: The public yearns for longpost Dean W. Ball: I’m kinda shocked

My take for posts that have a thesis, rather than being roundups:

You should only ever split an article into two small parts in places where the parts can stand on their own. If no one would ever want to link to or read Part 1 without Part 2, or Part 2 without Part 1, then you want to post it all at once, no matter the length.

That assumes you are on the internet, so you face no physical limitations. Except the Gmail clipping at ~102KB, but that ship sailed for me a while ago.

The reason The Sequences work so well is that while they form one large construction, you can link to or read each individual post on its own.

If you have a full book of content, then you have to split it into chapters, even if the chapters don’t stand on their own, but I would not consider that below 10k words, and would only seriously consider it around 20k, even if your normal length is 2k.

For roundups, I think it’s fine to split sooner, especially if that gives the sections better common themes. My new goal is non-AI roundups topping out around 5k.

So, while noting that everyone’s process is different and you need to find what works for you via experimentation, we’d better end here before we hit that mark.