How to research so the writing carries feeling

1 The bottom line

Research deeply, in person and at the source, then put on the page only the few details that let a reader see and feel. The days you spend learning are what give the writing its feeling.

2 Key takeaways

• Do most of the research before you write line one. You cannot write well until you know what you are writing about.
  • Research for The Tenth Circle sent Jodi Picoult1:33 by snow machine to a Yup'ik village in Alaska at 43 degrees below zero.
  • Susan Orlean50:30 spent three or four years on research for The Library Book.
• Go to original sources and to the people who were there. Documents and people hold what summaries leave out.
  • R.F. Kuang33:42 took "long back" for the long vacation straight from period sources, and some readers thought it sounded too modern.
  • Jimmy Soni10:50 called an unfamiliar PayPal investor, who said he invested because of how intensely Peter Thiel played chess.
• Hunt for the details a reader can see. A scene comes alive when the reader can picture it.
  • Eric Roth9:58 built a scene from a lemon on a gate that marked the winner of a citrus war in 1860 Sicily.
  • Robert Caro's repeated questions, as Paul Graham's interview relays them41:21, produced a Texas Hill Country cabin with no running water, rats and backed-up sewers.
• Cut most of what you learn. The reader needs the few details that serve the story, not proof of your effort.
  • Johann Hari4:25 does 200 to 300 interviews per book, and about 80% don't make it in.
  • Daniel Pink29:00 spent three or four weeks on child development research and used one paragraph.
• Treat a weak draft as a research problem. You cannot write your way out of a hole in your knowledge.
  • Wright Thompson41:05 says "All overwritten really means is underreported."
  • Ezra Klein32:20 found that a failed chapter usually meant he had not read enough books.

all the writing for me that moves me is made of feeling and time

Maria Popova, The Marginalian

2.1 Plan the research before you start

Decide what you need to learn, read what already exists, and finish most of the learning before you draft.

• Do the bulk of your research up front. Jodi Picoult1:33 does at least 80 to 85 percent of it before she writes.
  • Research until you are an expert on everything your characters know. By line one, Picoult has an outline, knows her characters and knows whether the research will fit.
  • Split the work into research, digestion and writing. Susan Orlean50:30 researches with no intention of writing, because she can't write until she knows what she is writing about.
  • Orlean works from tasks, not hours. Learning how arson is investigated took her six to eight weeks.
  • A reporter does not interview five people, draft, then go back for five more. At the Times, Nick Bilton47:54's editor first asked whether he was done reporting.
  • For realistic science fiction, Adrian Tchaikovsky26:04 treats current science as a wall and does the research first. The danger is unknown unknowns that force a rewrite halfway through.
• Start with a list of questions and a list of people to talk to. Johann Hari4:25 writes both at the start of every book.
  • List the questions you want answered. Then ask who could carry each part of the story for the reader.
  • End every interview by asking who else you should talk to. Each lead becomes a link in a chain.
  • Plan chapter by chapter. Fareed Zakaria6:14 decides what each chapter needs, which topic it covers and which books matter most.
• Learn the existing story before you test it. You then avoid reinventing the wheel.
  • Read the classic older accounts first. Diarmaid MacCulloch8:53 says "Get the cliches in your mind. Do not be dominated by the cliches, but know what they are."
  • Take the simplified version as a first pass. Tish Rabe57:18 took out every children's book on a topic, wrote notes, and only then wrote the rhyme.
  • Trace the history of an idea. Michael Mauboussin31:20 says knowing where it came from gives your writing depth.
  • MacCulloch's early Victorian edition of Bishop Burnet runs to eight volumes.
  • Ted Gioia1:40:35 reads 40 or 50 books on a subject, then moves on. That gives him authority without making readers tired of him.

2.2 Go to the primary sources

Original material shows how people described their own world, in words no summary keeps.

• Read secondary history for facts, then primary sources for voice. R.F. Kuang33:42 did this for Babel.
  • She read monographs on 1830s London first. Then she moved to campus novels and student diaries.
  • In a period piece, the narratives people spun about their world matter more than what we now know they were dealing with.
  • Let the sources sink in. Jayne Anne Phillips29:11 read letters from generals, infantry and women north and south, without writing things down.
  • One of Phillips's sources was a four-volume set on the Civil War, one volume per year.
• Go to the archives, not the surface page. Maria Popova2:51 says "I want the primary source materials."
  • The internet is only a thin layer of the human record. What lives in basements and reels has shaped the present more deeply.
  • Keep tracing an idea backward. Ryan Holiday46:30 looks for the oldest books on a topic, because current commentators rest their assumptions on them.
  • The 92nd Street Y holds undigitized reels of Pablo Neruda's first US appearance and Susan Sontag's first big lecture.
  • Holiday says that without reading Thucydides, you won't understand the people debating China and the US today.
• Mine old word lists, novels and films for the sound of a time or place. Precise words carry knowledge that standard language lacks.
  • Robert Macfarlane1:02:20 spent about two and a half years raiding 19th century regional glossaries. He also interviewed elders who keep threatened languages.
  • Mark Forsyth28:00 reads old slang dictionaries, which sum up a forgotten way of life.
  • For a novel set in 1954, Amor Towles6:32 read novels from that era by O'Connor, Baldwin and Chandler.
  • Dean Koontz40:42 watches good movies from the period, which hold great dialogue of the time.
  • A Gaelic speaker, Finlay MacLeod, gave Macfarlane a document with around 150 words for aspects of peat and moorland on Lewis alone.
  • Forsyth found "gongle", to stare idly at a canal, in a slang dictionary. Houseboat dwellers wrote to say they still use it.
  • Towles says the four books were written within about 18 months of each other, with totally different vocabularies and concerns.

2.3 Talk to people and go to the place

People and places give you what documents leave out.

• Research in person, both the setting and the people who work there. David Baldacci31:49 refuses to get his settings third hand.
  • Baldacci divides research into hardware and software. Hardware is things like parachute towers and walkalongs with police. It gives scenes and plot ideas.
  • Software is talking to people about why they do the work. It gives fuller and more honest characters.
  • Build the story from the ground up. For Train Dreams, Clint Bentley and Greg Kwedar19:56 rented a cabin on the river and paid a local naturalist to show them the trees.
  • Visit at least one real place per book. Jack Carr50:50 stays a week or two.
  • Bentley and Kwedar spent evenings in logging town bars. A Kootenai man restocking sturgeon fed a character and the opening image of a huge fish hanging from a pole.
  • When COVID kept Carr from Israel, he used books, photos and maps, hunted for details like a McDonald's menu, and had a family there read his pages.
• Interview the people who were there, and the people nobody interviews. Famous subjects have told their story so often it is rehearsed.
  • Jimmy Soni spotted an unfamiliar name on an early PayPal investor list, found a phone number and left a voicemail. He got a story no famous founder would have remembered.10:50
  • Ask people what they remember of a specific day. Bill Browder13:01 does this, then pulls on the threads and takes rigorous notes.
  • Call the people closest to the story. Tom Junod35:45 phoned families of the dead for The Falling Man.
  • Browder built his portrait of his traitorous former lawyer entirely from court transcripts and depositions.
  • Junod expected people to hang up. Instead they wanted any information he had.
• When a subject refuses you, write around them, and never rest on one voice. Patrick Radden Keefe5:54 interviews people who know the subject from other angles.
  • Subjects with PR and lawyers are scripted, so the outside view can be fuller and truer.
  • Read the literature, talk to several experts and check the data. Dan Wang29:35 criticizes pieces that relay one expert's opinions.
  • For the Sackler book, with three generations refusing to speak, Keefe talked to college roommates, assistants, yoga instructors and doormen.
  • For a profile of Mark Burnett, Keefe chose Burnett's two ex-wives over a sit-down with Burnett.
• Ask the questions that go past the first answer. The first account is rarely the deepest one.
  • Build an outline from reading, let the subject tell their usual story, then ask very personal questions. David Gelb6:33 also shares something of himself, which draws honesty back.
  • When two stories clash, ask the person to explain it. Brandon Stanton30:04 uses this to reach a new synthesis.
  • Ask every source the same questions. Lenny Rachitsky8:16 pastes every answer to one question into a single doc to compare them side by side.
  • Gelb's interviews run across about two weeks. Chefs say being on the show feels like therapy.
  • Lenny interviewed 17 founders or early employees of marketplace companies for one seven-part series.

2.4 Collect the details a reader can see

Look for small, concrete, sensory facts, and capture them while they are fresh.

• Ask what you would have seen if you were there. Robert Caro asked interviewees this again and again, as Paul Graham's interview relays it41:21.
  • Repeat the question even when people get annoyed. It pulls out visual details instead of generalities.
  • Mark Manson1:18:20 visited the church where Will Smith grew up and filmed it. He also had Smith draw a floor plan of his childhood home.
  • Manson's filming let him describe the wooden beams and the gospel song in one paragraph. He says "if you can make the reader see the scene that's when it comes alive".
• Research small physical details. They make the reader feel present.
  • Read the weather and the setting. Sam Parr12:12 praises Nick Bilton for using wind direction, Facebook photos and Ross Ulbricht's daily diary to build the arrest scene.
  • Walk your character through one day. David Baldacci1:05:40 researched everything his detective touched, from clothes to food to stores. He says that gave him about 90% of what he needed for the 1950s.
  • Keep digging for the telling detail. Eric Roth9:58 says "God is in the details."
  • Baldacci bought a 1958 Los Angeles gas station map on Etsy. He watched a 25-minute 1953 Las Vegas strip video frame by frame and found a mechanical cowboy that was dismantled because it scared people.
  • Books of Darwin Award stories gave Roth characters like a man struck by lightning seven times. Real World War II incidents and a tugboat maker fed his tugboat scene.
  • Sam lived in the neighborhood Bilton wrote about and says the bakery and street details were right.
• Use your senses and your own memory. Your lived feelings can carry a scene you never lived.
  • Rob Henderson39:01 listened to music from the era and opened packages of generic cookies to smell them. Smell especially pulled long buried memories back.
  • Jack Carr3:10 remembers what an ambush in Baghdad felt like and applies it to a fictional ambush in Los Angeles.
  • Learn your character's craft by doing it. R.F. Kuang14:27 took figure drawing classes for a novel about painters.
  • Henderson bought Dr. Thunder from Walmart and the same scented candle he had in Germany.
  • Drawing taught Kuang that "a lot of seeing is actually just filling in from memory". In prose she now asks what she has not observed.
• Write notes fast and uncensored, and keep them findable. You can't judge a fragment until you have the whole pile.
  • Yann Martel1:45 jotted quotes, small facts, a single word like bamboozle, and stray scenes without judging them.
  • Michael Pollan14:45 keeps one searchable journal per piece and writes in it every day. Each night he writes down everything he remembers of any experience, because vividness fades.
  • Write the full account the same day. Neil Strauss1:00:59 jots a few trigger notes in the moment and a full account at night.
  • Martel's two and a half years of research produced about 400 pages of notes.
  • Strauss and his son's mother compared notes on the birth and found they had forgotten nine tenths of it, including the emotional feel.
  • Jodi Picoult5:05 records interviews with permission. In her notebook she writes only the word that sparked an idea, so she knows which chapter each exchange belongs in.

2.5 Choose the few details that belong

Sort the pile by theme, keep only what gave you a jolt, and leave the rest off the page.

• Sort the material before you write from it. Shape the piece from your sorted notes, not from the raw sources.
  • Write each fact on a theme on a note card with its page number. Robert Greene8:20 then shapes the story from the cards instead of from the book.
  • Keep interview notes and reading notes apart, then combine them on an empty page. Derek Thompson5:22 does the synthesis there, not inside the notes.
  • Build a giant outline of facts. David Grann6:37 includes every physical detail of a person, how they spoke, every account of the scene and any photo of the place.
  • A 500 page Mary Shelley biography became five cards on envy for Greene's chapter in The Laws of Human Nature.
  • Grann's outline can reach 200 pages for a single 3,000-word chapter.
  • Tiago Forte33:02 highlights, then asks what is the very best within the best, and repeats. He says "the best one to five percent is so much more relevant than everything else".
• Keep the moments that gave you a jolt. Tim Urban9:14 notes the sparks while researching: a fact that clicks, a funny comparison, a connection.
  • Write a denser version made of those moments. Most research time is boring, confusing or already known.
  • Urban compares it to a stand-up special, the best material of a decade compressed into one hour.
  • He says three hours of research may yield only seven minutes of real dopamine hits. Raymond Chandler held himself to one insight per typed page.
• Stay on the path. Andrew Hunter Murray14:50 borrows Philip Pullman's image of a book as a walk through the woods.
  • You may look at the scenery. Do not leave the path to have a picnic.
  • Do the research, then show the reader only what moves the story forward.
  • Murray could have filled Breaking and Entering with pages on housing policy and made it an unreadably bad book.

2.6 Check it, and go back when the draft is thin

Verify every claim, and when a draft fails, fix the knowledge before the sentences.

• Get the facts right early to earn trust. Correct facts buy you room to be ambitious later.
  • Andy Weir11:35 makes his science accurate so readers stop checking after a chapter or two.
  • Nail down logistics so readers follow you into strange territory. Gillian Flynn41:30 writes first, then asks experts whether something could happen, not whether it would.
  • Reviewers tell Weir they skimmed the science paragraphs but were sure it was right.
  • Flynn fixed a plot hole about whether the FBI could date Amy's handwriting.
• Hire a fresh checker, and tell them to break the work. A reader's check is never enough.
  • Peter Attia1:10:35 hired an analyst who had never seen the book. The checker confirms the citation says what you claim and asks whether you misread the data.
  • Chris Dixon44:15 told his checker to find any mistake, no limits.
  • Tyler Cowen5:54 says any source must be double checked, whether it is a human, Einstein or the encyclopedia.
  • Attia's book has 38 pages of references.
  • Dixon's checker corrected dates and attributions and also turned up strong quotes.
  • Cowen tells his research assistant to "act like you're to destroy me". They always find things to fix.
• Test your story against details that do not fit. Reality has more nuance than your head.
  • Check each piece against three or four real cases. Henrik Karlsson14:08 rewrites when the example doesn't match the sentence.
  • Guard against confirmation bias. Ben Thompson38:58 checks that actual numbers and specifications match the narrative a company puts forward.
  • Keep lyrical language honest. Richard Powers25:55 lets his botanist speak in spiritual terms, yet every claim traces to peer-reviewed journals.
  • Much of Karlsson's case material never appears in the essay. It is the iceberg under the surface.
  • Thompson accepts overcompensating as the price of honesty, even if it sometimes makes him wrong because he is too eager to find a surprise.
• When a passage is purple or hollow, go back and report. The problem is a hole in what you know.
  • Get the missing facts instead of decorating the gap. Wright Thompson41:05 says every time he tries to write around it, it reads badly later.
  • Ezra Klein32:20 says the fix for a weak draft is intellectual labor, not better phrasing.
  • Klein says "it's not that you've messed up stylistically, it's that you haven't actually done enough reporting."

2.7 Where writers disagree

Writers split on when to research, what to research, and who should do it.

• Research first, or research as you write. Most writers above finish research before drafting, but not all.
  • Jimmy Soni29:33 found research was his way of dodging the anxiety of writing. He digs up a fact mid-section and drops it into the draft, and caps stand-alone research.
  • Johnny Harris6:27 found that over-engineering a story before a shoot shoehorned reality into his scaffold. He now goes in with access and a headline but no fixed framing.
  • Soni knows someone who has spent two decades researching a book about Plato and has not written it.
• Plan the research, or follow your curiosity. Planning saves time, and open wandering finds ideas others miss.
  • Riva Tez24:03 goes in with no goal and keeps poking until she finds something. She says few people do free-flow research anymore, which is where original essay ideas come from.
  • Her plan-free dive into the Codex Sinaiticus produced about 80 pages and an argument most people had never examined.
• Books and research, or firsthand experience. The sources differ on where the best material lives.
  • Tim Ferriss17:05 says his best work is personal experimentation or interviews with people who have done firsthand experiments. Anyone, and increasingly AI, can digest books and pull anecdotes.
  • Ezra Klein2:42 argues for books. He says it is more dangerous to think you've read something you haven't than to not read it at all.
  • Dean Koontz57:05 says to research the unfamiliar. Jack Carr3:10 instead draws on remembered feelings and does no research for his ambush scenes.
  • Koontz learned about gemstones buried in mountain soil to write a character who lives by a mine.
• Do it yourself, or brief an assistant. The answer depends on who can tell what will be useful.
  • Jodi Picoult5:05 uses no research assistants, because only she knows what will be useful.
  • Fareed Zakaria6:14 briefs a research assistant on each chapter and calls it almost a teamwork process.
  • For one chapter on the French Revolution, Zakaria read about 20 books plus academic articles and translated excerpts, and condensed it to a 40 to 50 page chapter.