“[O]ne has only learnt to get the better of words for the thing one no longer has to say, or the way in which one is no longer disposed to say it”  —T.S. Eliot

The doing and thinking required to write and revise means that writers are constantly calculating the output of their subtractions and additions. 

We’re counting on precision, but there is no exact answer. We can’t use perfect words; we can only use words that serve the moment. But, blink, and, as Eliot points out, that particular moment has passed. Those words, “shabby equipment always deteriorating,” which were so apt, are already wrong.

The attempt to fix moments in time, with words, frequently feels impossible and pointless.  This discomfort can coalesce into an unmovable obstacle, encountered by some as writer’s block. 

Writer’s block does not mean that you’ve failed. It means that you’ve stumbled onto the failure of words.

Such failure is constitutive of language because language is not commensurate with meaning. Our words always say less (and sometimes more) than what we mean. We can never really say just what we want to say–first because we don’t always know exactly what it is we mean, and second, because if we do, we don’t usually have the just-right words to convey it.

In other words, our thoughts and words can’t coordinate precisely. Writing lets us pretend otherwise by offering itself as a tool for facilitating closer connection. But its mechanism merely extends the variable of time, which magnifies imprecision.

There’s a solution to this problem, but it’s not without remainder. We must free ourselves from the tyranny of exactness by acknowledging our future failure. It’s not a personal shortcoming: It’s a consequence of communication, which is only approximate. 

Similarly, esprit de l’espalier. is linguistic melancholy. The perfect words–like the perfect comeback–often only arrive (if they arrive at all) when the moment has passed.

We can still write, we just have to tolerate that it’s almost always wrong.


While some writers approach revision as a set of fun problems to solve, most approach it as a set of high-stakes, anxiety-producing riddles. 

Revision, like all writing, requires both thinking and doing, or, in John Warner’s more evocative words, “expression and exploration.” Because we tend to subtract when we think, but add when we act, the work of writing is particularly difficult. To write–and revise–well, we must strategically wield apparently opposed forces.

Our subtractive thinking is represented by heuristics, those mental shortcuts that slice through the abundant inputs we take in every second of every hour of every day that we’re alive and awake. Heuristics subtract; they help us move from the potential paralysis of contemplation to the decisive movement of action (while also abetting our many and diverse biases). 

Meanwhile, our actions, particularly when directed toward object transformation, are frequently biased toward addition. In the linked study, participants were asked to change a Lego structure to make it more stable. Most participants, unless explicitly cued with directions to “streamline” their structure, added to it. They only rarely used subtractive strategies, though those typically yielded greater stability, and were more efficient. 

If (to adopt a gritty heuristic), we subtract when we think and add when we act, how should we handle the work of writing and revision? 

Primarily, we should separate the work into separate processes. Writing and revision both mean doing transformative thinking, but addition and subtraction serve each activity differently. Writing is all about adding–if we prioritize subtractive thinking and heuristics, we risk subtracting meaning.

Revision, however, should be understood as an implicit directive to “streamline” the object of our thinking. We practice efficient revision when we apply our subtractive powers to the writing we wish to transform.

Ultimately, when facing the challenge of revision, we may need to subvert our instinct to add more to stabilize our work. We should first consider the sum of subtraction. Subtractive revision makes the additive force in/of further action possible. Once we’ve taken out the extras, we can see what else should be included.

Though a project outline can be a helpful roadmap, for an outline to be radically useful, it can’t be traditional. It must be interrogative. 

We often take it on faith that all outlines are useful: If you’ve got a destination in mind, you want (and often need) to know how to get there. After all, aimless exploration isn’t ideal or enjoyable when you have an end in mind and a timeline for your arrival.

Although for a big project like a book, the destination is typically “done,” the kind of outline we use can get us there or it can get us lost. 

Traditional outlines get us lost. They’re passively structured and inert, articulating the themes and topics that need to be covered and connected to get a project to done.

But their structure makes it easy to lose your way. They’re simple and feel satisfyingly productive to create. We frequently confuse the effort we put into the outline with the effort required to turn the outline into a finished project.

However, our attention to it is reinforced by the traditional outline’s serial structure. An A, B, C pattern makes it incredibly obvious what comes next–D.

The obviousness poses a problem. By emphasizing structural connection, a traditional outline leaves unanswered the question of why A precedes B, and how we should make our move to B. The structure is maximally authoritative but fundamentally inert.

Question-based outlines, or interrogative outlines, on the other hand, are radically useful. Simply posing questions transfers the exploration of topics and subtopics and their connection to one another from the outline and onto the writer.

For instance, a theme like AI ethics in biomedical research, offers a nearly infinite number of topics and subtopics to be explored. A traditional outline encourages us to list out all the topics and subtopics that fit under this theme and connect them via seriation. It provides shape to infinitude, which makes it satisfying. But what is the shape of linearity? And when is it complete?

A question, on the other hand, solicits completion. If we turn the theme above into a question like, How will ethics shape the use of AI in biomedical research in the short-term?, we limit the topics under discussion while still allowing for maximal experimentation in our response.

Questions are dynamic. They imply not just one but many answers. They assume–by their very structure–an argument. Questions are also future-facing: A question mark solicits our future participation in meaning-making. And when we pose a question to ourselves, that question mark demands our participation.

So turn your traditional outline into an interrogative outline by reformulating your outline entries into questions. You may be surprised by how easy they are to answer.


An outline is a roadmap to a complicated project. It marks out the destination, as well as the big and small side trips you want to take along the way. An outline offers orientation and direction. With it in hand, you can see where you’re going and how to get there–you don’t need to wander around lost.

However, an outline can leave open the question of how, exactly, you’re supposed to get to where you need to be.

To answer this question–a question implicitly posed by the blank page or blinking cursor–consider the call sheet. It’s a tool that can help your execution.

A call sheet is typically used to organize the production of a film. It’s the daily memo from the assistant director to the cast and crew that describes the day’s shooting and production schedule, as well as related logistics like on-set participants and call times.

Like an outline, a call sheet breaks down a big project into its component parts. Unlike an outline, which provides more of a map toward a destination, a call sheet breaks down each leg of a trip into its component parts.  

Consider it an itinerary— a companion document to support your on-time arrival. Its daily schedule includes the day’s most pertinent details, making actualization straightforward.

If you’ve created an outline and are wondering why the project isn’t really easier to complete, first turn your outline entries into questions, and then create a call sheet to guide tomorrow’s work. Include on the call sheet the date, the project’s title, the number of words completed, and the number of words to complete that day. Include, too, the title of the part of the outline on which you’ll focus, the segments you’ll write, and the research required to support/complete those segments. Then, specify the times you’ll allot to the work and your daily schedule, including anticipated interruptions and other necessary breaks.  

When tomorrow comes, review your outline, consult your call sheet, and start writing as fast you can.

Parthenon temple on the Athenian Acropolis, Attica, Greece.

No matter how dense the subject, complicated the field, or convoluted the material, every interested reader should be able to access and understand the argument in any nonfiction book.

This can be a difficult imperative to accept. When we’ve spent years/decades/a lifetime gaining expertise, we usually bury the assumptions, connections, and relationships that make up the foundation of our work. If we condense that work into a book, we implicitly demand our readers do the work of excavation.

But readers won’t.

Even so, authors often resist the directive to make their argument more accessible–protesting that it’s a directive to dumb things down or pander to casual passersby.

This is not true. Accessibility is not synonymous with simplicity; it’s synonymous with functionality. When it comes to argument-driven books, functional means readable, and making a book readable is an authorial responsibility.

Authors of functional, readable nonfiction books adopt the conventions by which thinking can be shared. Importantly, they explain the foundations of their argument and expose the scaffolding from which they’ve built its tenets.

This is harder than it sounds. The foundations of complicated arguments tend to be deeply buried and are hard to unearth. Many authors give up their search during the drafting stage, deciding that if readers can’t do the work themselves, then they’re either not sufficiently motivated or the author’s thinking is too complex.

Possibly. More likely, though, this is what we tell ourselves to avoid what we prefer to see as unnecessary effort.

While it’s true that not every reader will be interested in evolutionary biology and the future of genetics, or in the philosophical foundations and future of AI, those who are interested enough to purchase our books are already motivated to follow the most complicated of thoughts.

We write for these readers–interested, motivated readers–readers who have sought out our work and want to know more. However, to understand our thinking, they must be able to access it.

Feedback is an integral part of any big project. Ideally, we solicit feedback from functional experts, neutrally review their notes, and integrate their applicable suggestions. In practice, however, we often solicit feedback from our friends, review their notes somewhat defensively, and search in vain for usable insights.

Feedback is always helpful, but it’s not always helpful in the ways we expect. Though we typically use feedback as a tool for finding solutions to our project’s problems, it’s more effective (and more reliable) to use feedback as a tool for verifying our project’s problems (and determining which of them require our attention). 

We do this by looking for the feedback behind the feedback. Readers’ suggestions are often motivated by the emotional friction they experienced when encountering our project. When we look in the background, to the feedback behind their feedback, we can identify this friction and deduce the problems that generated it.

Let’s take a comparative look. Here, a list of solutions from a reader of a working draft:

  • Consider taking out chapter 3–it doesn’t seem to fit.
  • Chapters 8 and 9 seem a bit long and meandering–consider combining them into one chapter.
  • Some chapters start with stories and others don’t–consider using the same structure for every chapter.
  • There are so many citations–I’m not sure where your argument begins or ends.
  • The story in the conclusion is very interesting–move it up.
  • The chapter examples are repetitive–consider mixing it up more. 

These might be helpful, but they might be arbitrary. Is deleting chapter 3 a good solution? It’s hard to say when we haven’t identified the problem beyond “fit.”

If we look behind the feedback, though, we find more generative feelings:

  • I’m confused, and I’m not exactly sure why. Chapter 3 seems confusing.
  • I’m confused. Maybe it’s because some chapters have different forms than others.
  • I’m confused. Maybe it’s because there are a lot of interruptions in the sentences. 
  • I’m having a hard time following this argument. I’m confused.
  • I’m not interested in this argument until it’s too late. / If I’m totally honest, I find this a little boring.

What’s the friction motivating our reader? Confusion and, potentially, boredom: They can’t find the argument’s throughline. They don’t find the argument interesting. They may not find the argument relevant.

The feedback behind the feedback can feel harsh (which is why readers don’t offer it and writers don’t seek it out), but it points the way to the underlying issues keeping our project from completion. Sometimes, useful solutions are in there, but in the background. We need to look behind the feedback to find them.

When beginning a new project, especially one that requires skills not yet acquired and experience not yet gained, we often encounter a gap between what we envisioned for our project and what it seems poised to achieve.

This chasm is an unavoidable feature of the creative landscape. It’s there, and we know it’s there, and if we’ve ever before created something, we know that sooner or later, and typically when we’re just about ready to release our new project into the world, we’ll arrive at its edge. 

The crevice is the beginner’s gap, and Ira Glass of This American Life candidly defines it as the space separating our work from our ambitions for it. Encountering this gap is demoralizing–and arriving at its brink over and over again makes it seem unnavigable.

Plus, there’s the irritating truth that the gap remains open for a surprisingly long time. “Beginner” is somewhat of misnomer here because the gap is always present, it just goes by other names.

Luckily, the gap gets easier to navigate. In fact just the work of creating a lot of material over an indeterminate but necessarily long period of time builds the bridge required to reach the other side of our efforts and feel real satisfaction.

Unfortunately, most of us don’t get there. We might encounter the gap once or twice or more and decide we never want to encounter it again. We experience the disappointment of the gap’s darkness as a message to turn back. 

We should instead experience it as a message to keep going. The beginner’s gap is just one element in a larger scene: It’s true that there’s no way to really close it (in part because disappointed ambitions are a frequent companion to creation). However, we can prepare for it and build a bridge across it by expecting our projects to fall short of our ambitions, and by keeping going anyway.

Semiotic Triangle

When it comes to writing, AI can generate poems, songs, stories, and fiction and nonfiction books. It can produce interviews and summaries and evaluations and copy of all kinds. As it gains more, better, and potentially multisensorial training, it will be able to do much, much more.

For some, the sudden surge in applications uncovers previously unexploited conveniences. If, for example, you spend too much time writing articles to refresh SEO relevance, AI offers a convenient solution.

For others, however, the purpose of writing is not always—or not only—to get the work done. It’s also to do the work. This is the case even though, as a proxy for thinking and reflection, and/or as a means for information exchange, writing is an inefficient, inconvenient medium.

It’s also often annoying, irritating, unpleasant, and very, very hard. Even writers consider writing torturous—a point made in Hemingway’s oft-quoted description of writing as “easy”—you just have to “sit down at a typewriter and bleed.”

But inefficiency and inconvenience—and annoyance and irritation (probably not the blood)—are important parts of the process. They’re cause and effect of the friction created when we attempt to match what we want to express with expression. 

AI can make the match easy by smoothing away this friction, but the convenience comes at a cost. In fact, Tim Wu writes in the still-relevant “Tyranny of Convenience” that although convenience helpfully and necessarily sands down some of life’s rough corners, if we sand away too much, we lose the edge.

Making easy our primary goal radically limits our choices, and thus the individuality we express in the act of choosing. Yes, AI can make the work of writing easy, but it smoothes away the friction that invites (perhaps requires) individuated expression.

We can and will turn to AI for a wide variety of tasks. But when it comes to writing, the hard work of enduring the annoying, irritating, unpleasant, and terribly inconvenient friction of writing is (part of) what makes us meaningful.

Reaping machine, 1880-1925, New Zealand, by Crombie and Permin. Identifier: B.079705.

By now, we’ve all seen the hype cycle welcoming and lamenting AI’s advancements. It’s true: Its astonishing innovations are a source of wonder. And, at the same time, ill-conceived incentives and unintended consequences will probably lead us, led by AI, in a meandering race to the bottom, in some areas at least. 

We’re in the grey space of before, awaiting potential regulations and experiments in implementation that will determine ethical and practical usage. In the meantime, we can draw from AI many useful lessons. For writers and others, AI offers a lens for understanding and responding to creative anxiety, maybe even the creative anxiety provoked by AI.

Creative anxiety is the stress that follows from the pressure to think expansively and improvisationally. When we feel it, we freeze before our task, work superficially through a tough problem, or avoid whatever is causing our discomfort. 

As an existential threat to creativity, AI is a legitimate cause of this (and other) anxiety. But AI is not yet so much an existential threat as it is a reinforcing mechanism of two critical biases: It supports both our tendency to confuse excess with meaning and our assumption that creativity is limited.

Our online lives encourage the conflation of excess with meaning in many, many ways. Such conflation is an efficient mechanism of/for the attention economy, in part because it eliminates the firsthand, active, participatory, and also time- and body-consuming experiences that typically inform significance. Our online experiences are gained second- or third-hand, passively and asynchronously. Their value depends not on impact but on accrual. 

Yet, the pervasiveness of digital ennui suggests that accrual can’t really lead to the kinds of significance on which meaning depends. That’s one reason why the online scroll feels endless, futile, and doomed: We seek, infinitely, some meaning.

AI is a new tool for making excess meaningful. In fact, its ability to accumulate the furthest pixels of the digital world suggests its output is particularly authoritative. This has implications for creativity, too: Those of us who experience creative anxiety often implicitly assume that creativity is a limited resource. It’s out there and acquirable mostly through discovery. AI’s capacity to access everything out there suggests a claim on locatable creativity.

But this isn’t quite right. Excess can inform significance and meaning, but it must do so by way of an interested interpreter. And creativity isn’t contingent on everything: It actually depends on nothing. By some measures, creativity is the improvisation that follows restriction—it’s an internal potential. In fact, it’s the possible basis of our evolutionary capacity and is therefore inherent, as possibility, in every living thing. 

Ultimately, while AI provokes anxiety, it also suggests strategies of response. We can, for instance, create significance and meaning by seeking out firsthand experiences (perhaps using AI as a tool to inform these experiences). We can also structure our work in restrictive ways that require improvisation (perhaps using AI as a tool to help set restrictions). 

AI can be a meaningful and creative, if fundamentally derivative, producer. But for now it requires interpreters to cull from its excess and respond to its nothings with flexible improvisation. It can help us to channel the very anxiety it provokes, even—or especially—when we consider it a stone against which to sharpen our response.


Books fails for many reasons, but nonfiction books fail when they fail to find an audience. Although this is a common consequence after publication, it can be hedged against in the early stages of book development. 

When nonfiction books fail to find an audience, it’s typically because they were developed, accidentally or purposely, for everyone. A book read by everyone sounds like a worthy aim, but it’s a reflexive and counterproductive goal.

Why? After all, everyone sounds like a lot of readers, and most people associated with books (rightly) believe that the more readers the better. Also, the concept of “audience” is inclusive: When we describe our book as a book for everyone, we mean it’s a book for anyone. No one should not read it. 

This is the everyone reflex: the natural and potentially even logical assumption that when we write, we write for everyone. The reflex is powerful, and when it guides development, it leads to books that fail to find real readers.

The everyone reflex is partly a manifestation of our confirmation and egocentric biases: We assume that others are as interested in our subject matter as we are, even if they don’t know it yet. 

In this way, the reflex lets us sidestep the responsibility of explaining our book’s relevance. Though this seems unnecessary—surely interested readers will find our book?—it’s a critical part of argumentation and one of our most powerful tools in positioning our book for success.

To circumvent the everyone reflex during the development stage, we must ask and answer the question of our book’s relevance. We must explain, on the page, our book’s “significant and demonstrable bearing on the matter at hand.” This explanation serves as our book’s reason-for-being, giving shape to our argument and pointing toward its most appropriate audience.

By identifying our book’s significance to the matter at hand, as well as the interested readers who already do and should care about it, we write not for the nameless, featureless everybody, but for the very particular readers who need and want our book.