Writers on starting, finishing, and bad drafts (Tim Ferriss #878)

A compilation episode in which seven writers answer the same four questions about practice — how to choose a project, how to begin without inspiration, how to produce pages reliably, and what to do when the pages are bad. The answers agree on more than they disagree, and the disagreements are the useful part.

Source: The Tim Ferriss Show #878 — How to Start When You're Stuck, Finish What You Begin, and Stop Waiting for Inspiration, published 7 August 2026. A compilation cut from earlier interviews with Elizabeth Gilbert, Anne Lamott, Joyce Carol Oates, Jerry Seinfeld, Mary Karr, Brandon Sanderson and Seth Godin. Archived under sources/2026-08/tim-ferriss-878-writing-craft-compilation/.

This is the first source in the notebook that isn't about agents, and it is worth reading on its own terms before it is read as anything else. The one place it genuinely meets the rest of the material is filed separately in Separating drafting from judging.

Ferriss names four questions in the introduction, and they make a good enough spine: how do you choose a project and stay with it, how do you begin when inspiration doesn't show up, how do you build a practice that reliably produces pages, and how do you keep going when you think the pages are terrible.

Choosing, and staying

Gilbert's answer is the only one in the episode about selection, and it has two parts.

The first is a rule for the moment a better idea arrives mid-project — which she says happens to her constantly, a quarter or a third of the way in. Her rule: you may abandon the current project for the new one only if you have a proven track record of finishing things. If you have thirty unfinished projects, the new idea is not a better idea, it is the same event repeating. Her reasoning is that the new one will be stale in a few months too, so the only thing the switch reliably produces is another unfinished project.

The second is a way of interrogating ideas before committing. She makes them pitch her — "I'm the angel investor and these ideas are like, 'We want your time and money for this'" — and reports that a lot of them evaporate under the question, because they turn out to be "something about birds" rather than a project. The ones that survive get told to come back when they are more formed.

Two things worth separating here. The pitch meeting is a real filter and it costs nothing to run. The finishing rule is doing something different: it is a deliberate refusal to re-evaluate on the merits, on the grounds that the in-flight judgement is known to be corrupted. That is a strong move and she states the cost of it plainly — sometimes the new idea really is better, and the rule tells you to ignore that.

Beginning

Oates is the flattest on this and the most quotable. Ferriss reads her a line from a Paris Review interview (The Art of Fiction No. 72) — "One must be pitiless about this matter of 'mood.' In a sense, the writing will create the mood" — and asks her to expand. She declines to expand: "I definitely feel that you create your mood by working." Don't wait to be inspired, don't overthink it, go and work.

The part of her answer that is actually actionable is the second half, and it is about protection rather than motivation. The problem is interruption; the people around you have schedules and expectations, and "while you may love your family, they can be the ones who drain you of energy." She is explicit that this falls hardest on women — the requests to read manuscripts, sit on committees, work pro bono all arrive, and the instinct is to say yes. She says yes to a great many of them, names Philip Roth and Flaubert as the counter-example of people who guarded their time, and does not pretend to have solved it herself.

Godin's version of beginning is the sharpest formulation in the episode: "there's no such thing as writer's block. Writer's block is real, but it does not exist." What it actually is, he says, is fear of bad writing and of what the world will say about it — so the way through is to do the bad writing, without shipping it. His model is Isaac Asimov, who typed for six hours every morning whether or not it was any good, and looked through what was left at the end of the shift; Godin's reading is that the subconscious works out that since it is going to type anyway, it may as well type something good.

A practice that produces pages

Seinfeld and Sanderson answer this one, and they answer it the same way: the output is a by-product of a system, and the system is boring on purpose.

Seinfeld's unit is the bounded session. An open-ended writing session is "a ridiculous torture to put on a human being's head" — his test case is hiring a trainer who says the workout is open-ended, which nobody would accept. So: pick a duration, and the end of it is the reward. "The alarm goes off, and you're done." When his daughter said she would write all day, he told her nobody writes all day. He gives the same treatment to meditation — if fifteen minutes doesn't work, do ten, the point being to find a length that will actually happen. His other two lines are worth keeping: writing is "a game of tonnage", and you should "learn to accept your mediocrity", because the people who are great are the ones who put in the hours.

Sanderson's unit is consistency across days, with numbers attached. Nine writers in ten do better writing regularly than in bursts; the one in ten is a binge writer who disappears to a cabin for two months and doesn't write for the other ten, and he says he doesn't understand that mode well. His own target is 2,000–2,500 new words a day, which he works out to a theoretical 500,000 a year, minus the third of his time that goes to revision, so around 300,000 — and a Stormlight Archive book is eighteen months for that reason. His class asks for 35,000 words over roughly a third of a year, i.e. a novel's length at that rate.

The mechanics he teaches are habit-stacking and environment: a specific place used for little else, music reserved for writing, and attaching the session to a habit that already exists — go to the gym, think about the writing there, come back and write for an hour. He also gives exercises, but only as diagnosis: the person who keeps re-revising chapter three instead of writing chapter four is told to draft longhand and type up yesterday's pages as the way into today's, which physically separates the two activities.

Note that these two answers are the same shape at different scales — bound the session, repeat the session — and that neither is about what to write.

When the pages are bad

Lamott's claim about Bird by Bird is that it worked because it promised nothing: not that finishing would get you published, and not that publication would heal anything, since "all of that is an inside job." What she teaches is narrower than writing well. "I just teach them to stop not writing" — keep your butt in the chair, write badly, and know that every book you admire started as an unreadable draft. Her grandson, aged five, caught the scope honestly: you said you'd teach them to write a book, but you only taught us how to write one page.

The two images that carry the chapter are both about scale and both worth having. Bird by bird is her father's instruction to her brother, who had a term paper on birds due Monday and hadn't started on Saturday: read about chickadees, write a paragraph on chickadees, draw a picture, then pelicans. E.L. Doctorow's is the one she calls the most profound advice she can offer on any topic — writing is like driving at night with the headlights on, you can only see a little way ahead, and you can make the whole journey that way.

On the critic itself she is more specific than the usual advice. Perfectionism is "the voice of the oppressor", and Bird by Bird's KFKD chapter is about a station that runs twenty-four hours a day telling you how far short you are falling. Her husband Neal Allen's practice, in Shapes of Truth, is not to remove the critic but to reassign it: thank it for keeping you alive at six and seven by keeping you small and controllable, note that it is no longer needed for that, then give it a new job — ethical consultant, sent off to a library with a good chair, to be fetched when an ethical consultation is actually wanted. Her honest caveat is that the left-hand speaker never turns off completely; it turns down.

Karr's answer is about revision, and the number is the thing. She threw out roughly 1,200 pages of Lit — "finished pages too, that's not draft" — written over five or six years, on a contract she was seven years late for. What she did about it was go to bed for two days, watch television, and then call Don DeLillo to say she thought she was writing a bad book. His reply was "Well, who doesn't?", and her conclusion from it is the useful part: every writer she knows has written a bad book, so maybe this is a bad book, "but it's the book that's standing in line to be written." She became willing to fail in order to say what happened.

Her method underneath that is sensory rather than analytical. Her test on a sentence is a list — can I make this less boring, more interesting, prettier, more colourful, more true — and the way she gets material worth revising is to re-occupy a former self in a place: the icy water from the under-dash air conditioner spilling onto her bare feet when her mother turned right, the peaches from Arkansas, the vodka and orange juice. The exercise she teaches from it is to pick a room you grew up in, get the smell first because smell is the oldest and most emotional memory, then interrogate the body for what it can taste and touch, and finally for what you wanted and what was in the way. The claim attached: as in trauma, the body remembers — and it remembers pleasure and beauty as well.

Godin's contribution here is the one aimed at a specific person, since Ferriss is describing his own hundreds of unpublished drafts. Two moves. The first is to challenge the premise: not good enough to publish — says who? The real statement, he suggests, is that the writing doesn't match the promise you'd like to make to the market you think you're in, which is a different claim and not obviously one Ferriss is well placed to judge; hand 5,000 words to someone whose actual skill is knowing when a thing is ready, and ask where it's thin and what to cut. The second is a replacement for "just do it", which he reads as licence to be a hack: merely do the work. The time spent narrating and catastrophising the work is not helping. He is describing eight consecutive years of failing as an entrepreneur while dramatising the problems, and the shift out of it as dropping the narrative rather than solving anything.

His last point is the one that generalises furthest. Twice he decided he had written the best book he had in him and stopped. After Permission Marketing he sat in the dark for a year, and what ended it was accidental: Gladwell, then unknown, sent him The Tipping Point for a blurb, and reading it in one night he realised he had already been writing a book about how ideas spread. He wrote it in twelve days, offered to throw it out if Gladwell objected, and got a blessing and a foreword instead. What he took from it is that he could not write Permission Marketing again — "that day was gone" — but he could write this one, and someone would benefit.

After Linchpin he felt the same way again and did not get an accident. Six or nine months later he decided the stopping had been selfish, which is where generosity comes in: not free work, but spending emotional labour on somebody who could use a hand. Once there was a specific person on the other end, "it wasn't about me anymore", and it got easier to merely do the work.

Where they disagree

Sanderson supplies the frame for reading the rest of the episode, which is a useful accident of the edit. He opens his class by telling students they will have to learn when to ignore him, because he is an authority standing at the front of a room and survivorship bias means he cannot actually tell which parts of what he did mattered. He presents the material as a toolbox that is deliberately self-contradictory.

His own example: Stephen King says don't outline and don't use a writing group; Orson Scott Card says an outline is fundamentally vital. Using George R.R. Martin's terms, King is a gardener — he plants two characters in conflict and finds out what happens, and says an outline kills his motivation because the discovery has already been spent — and Card is an architect, building the structure first and then polishing pieces of it. Sanderson's point is that you cannot weight both equally and get anywhere; you can hybridise, but at some point you pick one and test it on yourself.

The compilation contains at least three more contradictions of the same kind, none of them flagged in the episode:

• Bounded sessions vs. tonnage. Seinfeld's hour with an alarm and a reward, against Asimov's six hours every morning regardless of quality. Both are systems; they are not the same prescription, and Godin's is much closer to Sanderson's daily quota than to Seinfeld's.
• Consistency vs. the long collapse. Sanderson's spreadsheet and daily word count against Karr taking five or six years and discarding 1,200 finished pages. Sanderson would call Karr the one-in-ten he doesn't understand; nothing in her account suggests a word count would have helped, and nothing in his suggests hers is a method that could be taught.
• Finish what you start vs. throw it out. Gilbert's rule is that abandoning is the failure mode. Karr's turning point was abandoning most of a book. The two are reconcilable — she stayed with the project and threw out the pages — but only if you are careful about what the unit is, and the episode doesn't do that work.

What a compilation like this can't tell you

Worth recording, because the format flatters the material. Seven successful writers describing the practice that preceded their success is a sample selected on the outcome; nobody in the episode kept a daily word count for a decade and produced nothing, and those people exist. Sanderson is the only one who says this out loud, and he says it about himself.

Two smaller things. The advice is about practice, almost never about the writing itself — Karr's sentence test and her sensory exercise are the only craft content in the episode, and Lamott is explicit that she teaches people to stop not-writing rather than to write well. And the answers converge partly because the interviewer asked converging questions across the original interviews; the four questions in the introduction are his, not theirs.

The common ground that survives all of that is fairly small, and it is the same three items in every account: begin before you feel like it, bound the work so it can be repeated, and let the first pass be bad. That third one is the one that connects to the rest of this notebook, and it has its own page in Separating drafting from judging.