First in a series exploring the workflow of noted authors and other creators.

For as long as I’ve been writing books, I’ve been in a constant battle with the design of word processing software.

Just the phrase “word processor” itself is soul-sucking. But the real problem lies in the common interface conventions almost all word processors share, locked in a framework that took shape in the early days of computing, when software was almost exclusively oriented around the business market. These tools are undoubtably well-engineered for producing an inter-office memo, or a short business plan that follows a conventional structure. But they are almost comically unsuited for long, complex, and structurally open-ended documents like non-fiction books or novels.

To understand why, you need to consider the entire the life cycle of a book, from the early ideas to the completed project.

In the initial stage, you have hunches and fragments: a hint of idea for a character, or an historical theme, a quote from a book you’ve read as early stage research. These fragments are often only tangentially related to each other, and usually have no intrinsic sequence to them. They’re just floating around in your head or your notebooks.

And then, over time, some of those hunches start to coalesce into larger things. They become coherent ideas. A character comes into view; an interesting explanation for some historical trend becomes clear to you. As they do, they send you off on further paths of investigation, particularly if you’re writing non-fiction: you interview an expert on the idea you’re exploring, or you uncover a long-lost gem of scholarship buried in the archives.

If you’re lucky and persistent, those ideas then take on a distinct linear form: they become arguments, or narratives. Sequence begins to matter; where you start and where you end makes a meaningful difference. Eventually those arguments and narratives mature into higher-level categories: sections, chapters, entire books.

Two things are worth noting here: the nests and the non-linearity. The different layers are nested in structure. Hunches come together to form ideas which come together to form stories. But sequence only becomes critical in the top layers: stories, arguments, chapters. The different between the two stages is like the difference between the pushpin evidence board from The Wire — a scattered network of clues and potential connections—and a prosecutor’s closing statement in a criminal trial.

Conventional word processors don’t have a natural way of managing that evolution from hunch to narrative. They lock you into a scrolling document structure from the beginning. They give you a line, not a nest. They give you a medieval scroll when what you need are Chinese boxes.

A linear scroll is the interface metaphor you need when you are sitting down to write the mythic first draft of a longer text, where the entire sequence of the novel or the chapter is already mapped out in your head (or in some other mode of idea storage — notecards, Evernote folders.) But I think for most authors, that archetypal first draft is itself a fiction. We venture into the new continent of a book with some small fragment of it mapped in advance, and then fill in the rest as we explore.

This means that as we write even final drafts of long projects, there is still a great deal of movement up and down the chain, from hunch to idea to narrative. New clusters continue to form as you write. Sequences get radically re-configured. Almost half of my books reached a complete first draft before I had even attempted to think about the order of the chapters. But executing all that conceptual movement is a pain, once you’ve switched over to the forced scroll of Word or Pages.

We haven’t noticed this limitation as much because many of us use different tools for earlier stages of the workflow for writing a book. We manage the progression of hunch to idea in products like Evernote and Apple’s increasingly useful Notes app. Dedicated outliners and built-in outliner tools baked into existing word processors serve as kind of middle layer, once you have started to think about actual structure. For many authors, the switch from idea generation to actual writing is marked by a shift from the physical world to the digital: they keep their ideas in notebooks or index cards, and only move into the realm of software when the fragments have turned into narratives.

But why do we need multiple tools to get from hunch to narrative? It’s a bit like splitting up a piano so you have one instrument to tease out a melody, and another to nail down the chords. I’ve been looking for a better system for book writing for more than two decades now; ever since I started writing my first book in the mid nineties. And I think I have finally found a solution that does justice to the actual realities of moving from initial idea to finished manuscript.

Over the past three years or so — starting with my book How We Got To Now— I’ve developed a workflow that matches the nested, non-linear development of a book, one that no longer crams that development into the forced scroll. It relies heavily on the indie application Scrivener, which bills itself as writing software for “long and difficult documents.”

Something about Scrivener elicits a lot of strong feelings from people who have used it, both positive and negative. It has a growing community of writers who swear by it, and a parallel community that is tired of hearing all the Scrivener-heads raving about their magic tool. I happen to think Scrivener has way too many features, and even the useful ones are often buried inside byzantine preferences panels that confuse more than they clarify. In fact, my allegiance to Scrivener basically boils down to just three tricks that the software performs, but those tricks are so good that I’m more than willing to put up with all the rest of the tool’s complexity.

Those three tricks are:

  1. Every Scrivener document is made up of little cards of text — called “scrivenings” in the lingo — that are presented in an outline view on the left hand side of the window. Select a card, and you see the text associated with that card in the main view.
  2. If you select more than one card in the outline, the combined text of those cards is presented in a single scrolling view in the main window. You can easily merge a series of cards into one longer card.
  3. The cards can be nested; you can create a card called, say, “biographical info”, and then drag six cards that contain quotes about given character’s biography into that card, effectively creating a new folder. That folder can in turn be nested inside another folder, and so on. If you select an entire folder, you see the combined text of all the cards as a single scrolling document.

You can see what the the panel/card interface looks like at the end of a project in this screen grab of the final draft of my book Wonderland.

These might seem like simple features, but they have profound effects on the writing process once you embrace them. Building the software around this foundation changes everything about the way you navigate from the early stages of the project to a completed manuscript.

In the research phase, you’re just creating a disorganized pile of cards, with quotes, ideas, links, fragments, hunches. There’s no order, no sequence; just a non-linear collection of vaguely related ideas. But as the project takes shape, certain themes begin to emerge, and those become folders housing other cards. Eventually those themes start to map onto actual sections of the book, or individual chapters. At this point, sequence does begin to matter, but you can change the sequence just by dragging cards and folders around in the left hand outline view.

Now, I realize that traditional word processors allow you do things like reorganize the sequence of a chapter. But just think about what most writers have to do when they’re in the middle of a book, and they decide that they want to move one three-page section from the end of one chapter and put it in the middle of another chapter. Most writers I know who don’t use Scrivener keep individual chapters as separate documents. So moving a three page section of text requires you to:

  1. open up one chapter document
  2. scroll down to the start of the passage you want to move
  3. select three pages of text
  4. copy that text
  5. open up the other chapter document
  6. scroll down to the part of the chapter where you want to insert the text
  7. paste the text

In Scrivener, that entire sequence requires one gesture: you drag a card from one folder to another. In a way, this part of the interface is more like creating a music playlist than writing in a traditional word processing application. Dragging cards from one folder to the next gives you a dramatic increase in efficiency, but it’s also a conceptual breakthrough as well, because as you’re writing, you can see both the specific section you’re working on and how that section fits into the overall picture. And if you find yourself looking for a quote or a sentence you’d jotted down in the research phase, you don’t have to switch over to Evernote or flip through your physical notecards. You can just visually scan the existing cards in the left-hand panel, or do a quick search inside of Scrivener itself. The interface doesn’t just let you to see the forest and the trees of your project. It also lets you see the seedlings.

Reading about how someone uses a piece of software is always a little abstract, so here’s an example from my own recent writing to give you a sense of how this works in practice. In my first year in grad school, way back in 1991, I took a seminar with the brilliant Franco Moretti on the 19th-century novel, during which we spent a few hours in class discussing an epidemic of kleptomania that erupted in Paris shortly after the opening of the Bon Marché, one of the very first true department stores ever built. (The outbreak was memorably known as the “department store disease.”) That story stuck in the back of my head for more than a quarter of a century, for no real reason other than the fact that it was interesting, and probably warranted a deeper dive. It made for a great dinner party anecdote during those twenty-five years, but I’d never had a reason to explore it professionally in more detail over that period.

But then a few years ago, I started thinking about a book about the history of leisure and play, and that story popped into my mind immediately. Here you had an origin story about the birth of shopping — the creation of a now-common leisure activity that involved crime and pathological behavior in the midst of previously unimagined luxury. That seemed like promising material, so I created a card in Scrivener that said something along the lines of “department store disease — what was the deal with that?” For a while, it lived alongside some other vaguely related cards on other leisure themes: some notes about the origins of Monopoly; the history of musical instruments; the cultural impact of taverns and pubs.

And then I dug around a bit, and tracked down Michael Miller’s definitive cultural history of the Bon Marché, and found a bunch of tantalizing quotes about the first department stores and their early kleptomaniac patrons. Almost immediately that single card devoted to the department store disease became a folder, housing a cluster of cards, each with a quote from Miller’s book, or some other meditation of mine on what the disease might potentially signify.

Thinking about the origins of the department store got me thinking about about the modes of luxury consumption that pre-dated the Bon Marché and its ilk. That led me to a number of less fruitful books on the history of shopping — most of which were short on historical detail — until somehow I found an interesting series of papers about the London shopkeepers of the late 1600s who first pioneered the use of luxury furnishings in the store design itself. That quickly generated a folder in Scrivener filled with a few dozen cards with quotes and images drawing on the existing scholarship about those stores and their influence. That investigation led me to the history of fabrics from that period — most famously calico and chintz — which soon became its own Scrivener folder, teeming with data about the East India Company and the political backlash against the “calico Madams” of British high society.

By the end of this process, I could see that the Bon Marché/new shops in the 1600s/Calico cluster was booming, next to other less promising lines of investigation, which tended to be a bunch of stray cards of quotes without an organizational structure. Seeing those ideas nestled together got me thinking about a larger idea on the connection between industrialization and leisure, and how the conventional account of the industrial revolution has the causality backwards. Now, for the first time, I could see a chapter structure starting to take shape. I could see it conceptually in my mind, but — and this is the important thing — I could also see it on the screen, in those bulging folders. Of course, the linear path of my investigation didn’t mirror the sequence that the ideas would ultimately take on in chapter itself. But that didn’t matter, because it took three seconds to drag the folders in Scrivener into the right sequence. And with each drag all my notes and citations and quotes came along for the ride as well.

At that point, I had four sections of the chapter arranged in the proper order: 1600s shops; calico; industrialization; Bon Marché. Superficially, these might have looked like entries on the same level of an outliner, but the difference was profound: they each contained the whole history of my investigation into that particular theme. This is the point at which it is probably fair to say that I started “writing” the chapter, though the lines are inevitably blurrier when you use one tool for the entire research/ideation/writing/editing process. But instead of opening up a blank document in a word processor, and creating a section header called “the shops of the late 1600s,” I could create an existing document immediately drawing on all the research that I’d been exploring for the past few months. I scanned through all the cards in the folder, and dragged out all the ones that didn’t seem essential into a generic catch-all folder called “research.” The remaining cards were the best of the batch: quotations I knew I wanted to use; snippets of commentary or description that I had written myself as part of the research phase. I dragged those cards into roughly the right sequence, and started filling in my own text around them.

One subtle advantage of this approach is that it helps you avoid the “blank page problem,” one of the major drivers of writerly procrastination. When you sit down to write a new chapter, you never confront an empty document; each chapter begins its existence pre-loaded with quotations and notes that already follow the general sequence you’ve envisioned. I often think of those fragments as a kind of archipelago of ideas; when I finally sit down to write a chapter, I’m just building bridges between those islands, and not venturing out into an empty, uncharted sea.

Replacing the forced scroll with nested cards allows the project to grow and mature over time, inside the same application. You end up using the same tool to explore and organize your early ideas when the project its still in an embryonic stage and when you are developing a full manuscript with chapters and sections and endnotes. After two decades of writing books, I finally feel as though the tools I’m using to write are properly aligned with the life cycle of the book itself. I no longer feel as though I’m being compelled to word process. I’m just writing.

Like many people I know, I have always enjoying tinkering with my workflow, experimenting with new tools and approaches. And I’ve always enjoying talking to other creative people (sometimes other writers, but not always) about their habits and technology. My Scrivener-based workflow seems to fit the way my mind works better than any approach I’ve employed in the past, but it’s always amazing to see the range of strategies that people adopt in the creative process. And so it seemed appropriate to launch a new series here at Medium, called Workflow, where I’ll interview some of the most interesting creative people I know, and get them to talk about their process and their preferred tools. Next week, we’ll start the series with another kind of writing — songwriting—in a conversation with one of my favorite songwriters of all time, Liz Phair. In the coming weeks, I’ll visit with Kevin Kelly, Rebecca Skloot, and Steven Pinker, among others. Stay tuned!