Clear writing
How to write prose that a busy reader understands on the first pass: the core principle, the canonical rules (Orwell, Strunk, Zinsser, Pinker) and where they go wrong, structure, sentences, word choice, punctuation, readability formulas and an editing process. Applying it to memos, emails and pitches is in persuasive writing; docs, design docs and commit messages are in technical writing. Reader-first thinking is the same idea as in UX/UI design.
The core principle
Clear writing is clear thinking made visible, aimed at a specific reader. Almost every rule below is a corollary of three ideas.
| Idea | Meaning | Test |
|---|---|---|
| Know what you think | you cannot write clearly about something you have not decided | can you state the point in one sentence, out loud, without notes? |
| One main point | every document, section and paragraph has one job | could a reader summarize it in a line, and would it match yours? |
| Reader first | the reader has less context, less time and less interest than you | what does this reader already know, and what do they need to do next? |
Zinsser puts the first one bluntly in On Writing Well (chapter "Simplicity"): "Clear thinking becomes clear writing: one can't exist without the other." Paul Graham makes the same point from the other side in "Putting Ideas into Words" (2022): writing about something usually shows you that you did not know it as well as you thought.
Practical consequences:
- Write the point first, then the document. If you cannot write the one-sentence version, you are still thinking, not writing. That is fine, but label the draft as thinking.
- Name the reader. "The CFO who will approve this" produces a different document from "people interested in our infrastructure". Write for one real person; others will cope.
- Decide what the reader should do or believe afterward. Informing, deciding and persuading need different shapes (see persuasive writing).
- The reader cannot interrupt. In conversation a confused listener asks. On the page, every gap you leave is filled by a guess or a closed tab.
The canonical rules
Orwell's six rules
George Orwell, "Politics and the English Language" (Horizon, 1946). Before the rules he lists the questions a "scrupulous writer" asks of every sentence: "What am I trying to say? What words will express it? What image or idiom will make it clearer? Is this image fresh enough to have an effect?" Then:
I think the following rules will cover most cases:
i. Never use a metaphor, simile or other figure of speech which you are used to seeing in print.
ii. Never use a long word where a short one will do.
iii. If it is possible to cut a word out, always cut it out.
iv. Never use the passive where you can use the active.
v. Never use a foreign phrase, a scientific word or a jargon word if you can think of an everyday English equivalent.
vi. Break any of these rules sooner than say anything outright barbarous.
(Text as published by the Orwell Foundation; some printings add serial commas in rules i and v.)
| Rule | What it targets | Modern caveat |
|---|---|---|
| i. stale figures | clichés that substitute for thought ("move the needle", "low-hanging fruit") | a fresh, precise metaphor is still one of the best explanation tools |
| ii. long words | inflation: utilize, facilitate, commence | the precise technical word beats a vague short one (idempotent is fine for engineers) |
| iii. cut words | padding, throat-clearing, redundancy | cutting can remove the connectives that make logic followable |
| iv. active voice | hidden agents, evasions ("errors were made") | the passive is often right (see sentences) |
| v. jargon | in-group words used on outsiders | jargon is efficient inside the group; the error is using it on the wrong reader |
| vi. break the rules | rules are tools, not law | the most ignored and most important rule |
Orwell's own essay contains passives and long words; he says himself that "one could keep all of them and still write bad English". The rules are a checklist against habits, not a definition of good prose.
Strunk and White
William Strunk Jr privately printed The Elements of Style for his Cornell students in 1918; E. B. White revised and expanded it for Macmillan in 1959. The useful core is in Strunk's original "Elementary Principles of Composition" (rule numbers from the 1918 text):
| # | Rule (1918 wording) | Keep it? |
|---|---|---|
| 8 | Make the paragraph the unit of composition: one paragraph to each topic | yes |
| 9 | As a rule, begin each paragraph with a topic sentence; end it in conformity with the beginning | yes, for expository prose |
| 10 | Use the active voice | as a default, not a ban |
| 11 | Put statements in positive form | yes: "late" beats "not on time" |
| 12 | Use definite, specific, concrete language | yes, the most valuable rule in the book |
| 13 | Omit needless words | yes, but it only helps if you can tell which words are needless |
| 15 | Express co-ordinate ideas in similar form | yes (parallelism) |
| 16 | Keep related words together | yes |
| 18 | Place the emphatic words of a sentence at the end | yes (end focus) |
Strunk's rule 13 in full: "Vigorous writing is concise. A sentence should contain no unnecessary words, a paragraph no unnecessary sentences, for the same reason that a drawing should have no unnecessary lines and a machine no unnecessary parts. This requires not that the writer make all his sentences short, or that he avoid all detail and treat his subjects only in outline, but that he make every word tell." The second sentence is the part everyone forgets: concision is about every word earning its place, not about brevity.
The critique. Linguist Geoffrey Pullum's "50 Years of Stupid Grammar Advice" (Chronicle of Higher Education, April 2009) argues that the book's "advice ranges from limp platitudes to inconsistent nonsense". His main points:
- Much of the style advice is harmless but empty: "Omit needless words" is "useless" because "The students who know which words are needless don't need the instruction."
- The grammar is wrong. Of the four examples offered to show the passive, Pullum shows three contain no passive at all ("There were a great number of dead leaves lying on the ground").
- The anti-passive bias ignores information flow: a passive sounds bad when its subject is newer than the agent, and natural otherwise ("The bill was paid by an anonymous benefactor").
- Rules against starting with however, against which in restrictive clauses and for singular none are contradicted by the usage of respected writers.
Verdict: use Strunk's composition rules as prompts; ignore its grammar pronouncements; never enforce "no passives" mechanically.
Zinsser: clutter and simplicity
William Zinsser, On Writing Well (first edition 1976). The chapter "Simplicity" opens: "Clutter is the disease of American writing." His fix: "the secret of good writing is to strip every sentence to its cleanest components. Every word that serves no function, every long word that could be a short word, every adverb that carries the same meaning that is already in the verb, every passive construction that leaves the reader unsure of who is doing what—these are the thousand and one adulterants that weaken the strength of a sentence." (Wording from the 1980 edition.)
| Clutter type | Example | Clean |
|---|---|---|
| inflated phrase | "at this point in time" | "now" |
| redundant adverb | "totally unique", "smiled happily" | "unique", "smiled" |
| qualifier | "in a funny sort of way", "rather", "quite" | delete |
| euphemism | "negative patient care outcome" | "the patient died" |
| buzzword | "leverage our learnings going forward" | "use what we learned" |
Zinsser also insists that "Writing is hard work. A clear sentence is no accident." Most clear sentences are rewrites.
Pinker: The Sense of Style
Steven Pinker, The Sense of Style (2014), is the best modern guide because it explains why prose is hard to read, using cognitive science rather than decree.
Classic style. Borrowed from Thomas and Turner's Clear and Simple as the Truth (1994). Pinker: "The guiding metaphor of classic style is seeing the world. The writer can see something that the reader has not yet noticed, and he orients the reader's gaze so that she can see it for herself." Writing is a window onto the world; writer and reader are equals in a conversation; the writer shows rather than argues about how hard the topic is.
| Classic style does | Classic style avoids |
|---|---|
| points at concrete things, examples, numbers | talking about the writing itself ("in this section we will…") |
| states claims with appropriate confidence | reflexive hedging to avoid being wrong |
| treats the reader as intelligent and uninformed | condescension or in-group shorthand |
| uses the concrete noun and the active verb | abstractions stacked on abstractions |
| presents the conclusion, not the writer's struggle | the literature-review tour before the point |
The curse of knowledge. Pinker calls it "the single best explanation I know of why good people write bad prose": once you know something, you cannot easily imagine not knowing it. The term comes from economists Camerer, Loewenstein and Weber (1989). In Elizabeth Newton's 1990 Stanford study, people tapped the rhythm of well-known songs and predicted listeners would name about half; listeners named 2.5%. Symptoms in writing: undefined abbreviations, skipped steps, "obviously", examples that only make sense if you already understand.
| Fix | How |
|---|---|
| show a draft to a real outsider | someone from the target audience, not a teammate |
| read your own draft after a gap | days, not minutes; you forget enough to see the gaps |
| define on first use | every abbreviation and every term the reader might not share |
| add the example | an abstraction plus one concrete instance beats three abstractions |
| replace labels with descriptions | "the Q3 thing" → "the pricing change we shipped in September" |
Zombie nouns. Helen Sword's term ("Zombie Nouns", New York Times Opinionator, 23 July 2012) for nominalisations: verbs and adjectives turned into abstract nouns (investigate → investigation, decide → decision, able → ability). They drain the agent and the action from a sentence. Pinker describes the move as taking "a perfectly spry verb" and embalming it into a noun.
| Zombie | Alive |
|---|---|
| "The implementation of the migration resulted in a reduction of latency." | "Migrating cut latency." |
| "We had a discussion about the prioritization of the roadmap." | "We discussed which roadmap items come first." |
| "Failure to provide confirmation will result in cancellation." | "If you don't confirm, we'll cancel your booking." |
| "There was an agreement among the team on the adoption of TypeScript." | "The team agreed to adopt TypeScript." |
Metadiscourse is writing about the writing: "In this section I will discuss…", "As noted above…", "It is worth mentioning that…". A little signposting helps in long documents; most of it is throat-clearing. Pinker's alternatives: open with the question the section answers, or just state the point.
Hedging. Compulsive qualifiers (somewhat, arguably, to some extent, it could be said that) protect the writer, not the reader. Hedge once, precisely, where there is real uncertainty, and say how uncertain: "probably (we have three weeks of data)" beats "it may potentially be the case".
Plain language
The Plain Writing Act of 2010 (Public Law 111-274, signed 13 October 2010) requires US federal agencies to use "plain writing" in public documents, defined as "writing that is clear, concise, well-organized, and follows other best practices appropriate to the subject or field and intended audience." The Federal Plain Language Guidelines (PLAIN, March 2011, revision 1 May 2011) turn that into practice; the material formerly on plainlanguage.gov now lives in digital.gov's plain language guides.
| Guideline theme | Practice |
|---|---|
| write for your audience | identify readers and their top tasks; answer their questions in their order |
| organize | most important first; useful headings (often questions); short sections |
| words | short, everyday words; the same term for the same thing; define unavoidable jargon |
| sentences | active voice; "you" for the reader; "must" for obligations; short sentences |
| layout | lists for steps and options; tables for comparisons; white space |
| test | check comprehension with real readers, not with formulas alone |
Plain language is not dumbing down. Experts read plain prose faster too, and nobody has ever complained that a specification was too easy to understand.
Structure
Structure matters more than sentences. A reader who knows where the argument is going forgives clumsy sentences; perfect sentences in the wrong order still fail.
| Pattern | Rule | Use for |
|---|---|---|
| BLUF (bottom line up front) | put the conclusion, request or decision in the first line or two | emails, status updates, memos, anything someone might stop reading |
| Inverted pyramid | most important information first, then supporting detail in descending importance | news-style updates, incident notices, announcements |
| Pyramid Principle | answer first, then grouped supporting arguments (Minto) | recommendations; see persuasive writing |
| Topic sentence | the first sentence of a paragraph states what the paragraph shows | expository paragraphs |
| One idea per paragraph | a new idea starts a new paragraph | everything |
| Chronological | events in order | incident timelines, histories, tutorials |
| Problem → solution | describe the pain before the fix | proposals, design docs |
BLUF comes from US military correspondence; the inverted pyramid from newspaper practice. Both exist because readers stop early.
Topic sentences and skimming
If a reader reads only the first sentence of each paragraph, they should get the argument. This "skim test" is the single most useful structural check.
| Before | After | Why |
|---|---|---|
| "Over the last quarter the team has been looking at several aspects of the onboarding flow, including drop-off, time to activate and support volume." | "Onboarding loses 40% of sign-ups at the email-verification step." | the topic sentence states the finding, not the activity |
| "There are a number of considerations regarding the database choice." | "Postgres fits our workload better than MongoDB for three reasons." | announces the claim and the shape of what follows |
| "Security is important." | "Every endpoint that changes data must check the session's role." | specific, testable, actually says something |
Parallelism
Items in a list, a comparison or a series take the same grammatical form (Strunk's rule 15).
| Before | After |
|---|---|
| "The plan covers hiring, how we will budget, and a timeline for launch." | "The plan covers hiring, budget and launch timing." |
| "Steps: 1. Install the CLI 2. Configuration of the token 3. You should now run the tests" | "1. Install the CLI. 2. Configure the token. 3. Run the tests." |
| "She is fast, accurate and has experience." | "She is fast, accurate and experienced." |
Lists vs prose
| Use a list when | Use prose when |
|---|---|
| items are parallel and independent (options, requirements, steps) | ideas depend on each other (because, therefore, unless) |
| order matters (numbered steps) | you are making an argument |
| the reader will scan or check items off | nuance and trade-offs matter |
| there are 3–9 items | there are 2 items (a sentence is shorter) |
Bullets hide missing logic: three bullets can sit side by side without anyone stating how they relate. Bezos's 2004 email banning slides at Amazon's senior meetings made this point: "If someone builds a list of bullet points in word, that would be just as bad as powerpoint." Use lists for data, prose for reasoning.
Sentences
Active and passive voice
Active: subject does the action ("The deploy script deleted the table"). Passive: subject receives it ("The table was deleted", optionally "by the deploy script"). A passive needs a form of be (or get) plus a past participle; "There were leaves on the ground" and "The server is down" are not passives.
| Prefer active when | Passive is right when |
|---|---|
| the agent matters and is known | the agent is unknown or irrelevant: "The server was restarted at 02:00." |
| the passive would hide responsibility: "mistakes were made" | the thing acted on is the topic you are continuing: "The parser has a bug. It was introduced in v2.3." |
| instructions: "Run the migration" | you want the new or heavy element at the end: "The prize was won by a team of three first-year students from Leeds." |
| the sentence gets shorter | the convention requires it (methods sections, some legal text) |
Strong verbs, fewer nouns
The action of a sentence belongs in its verb. Hunt for weak verbs (make, do, have, give, provide, perform, conduct, is) attached to action nouns.
| Before | After | Why |
|---|---|---|
| "We performed an analysis of the logs." | "We analyzed the logs." | nominalisation → verb |
| "The report provides a description of the outage." | "The report describes the outage." | same |
| "There is a need for us to reduce costs." | "We need to cut costs." | expletive "there is" plus zombie noun |
| "It is recommended that users update." | "Update to 4.2." | hidden agent; instruction as command |
| "The team made the decision to postpone." | "The team postponed." | the decision is implied |
| "Our goal is the facilitation of faster onboarding." | "We want new users productive in a day." | abstraction → concrete target |
Given-new and end focus
Readers process a sentence by linking its start to what they already know, then adding the new part (the "given-new" principle from Haviland and Clark's psycholinguistics research in the 1970s). English also puts emphasis at the end of a clause (Strunk's rule 18). Together: start with the familiar, end with the new or important.
| Before | After | Why |
|---|---|---|
| "Moving to Postgres in March was what let us drop the document store, because JSONB support is included in Postgres." | "We moved to Postgres in March. Its JSONB support let us drop the document store." | each sentence starts with something already mentioned |
| "A 40% cost reduction is expected from the new caching layer." | "The new caching layer should cut costs by 40%." | the number is the news; put it last |
| "We should ship on Friday, in my view, given the risks." | "Given the risks, we should ship on Friday." | the recommendation lands last; qualifiers move to the front |
Chains of given → new across sentences produce cohesion: each sentence picks up where the last one ended. Topic-hopping (every sentence starting with a new subject) feels choppy even when each sentence is fine.
Sentence length
There is no correct length; there is a correct variation. Uniform short sentences read like a telegram; uniform long ones exhaust working memory.
| Guideline | Detail |
|---|---|
| average 15–20 words | a common plain-language target; not a hard cap |
| vary deliberately | a short sentence after long ones lands a point |
| one main clause per idea | split when a sentence has two "and"-joined main points |
| keep subject and verb close | long insertions between them force the reader to hold the subject in memory |
| front-load conditions only when short | "If the build fails, rerun it." fine; a four-line "if" clause first is not |
| avoid noun stacks | "user account password reset email template" → "the template for password-reset emails" |
Positive, concrete, specific
| Before | After |
|---|---|
| "not unlikely" | "likely" |
| "did not remember" | "forgot" |
| "The results were not insignificant." | "Conversion rose from 2.1% to 2.9%." |
| "a period of unfavourable weather" | "a week of rain" |
| "improve performance" | "cut p95 latency from 800 ms to 200 ms" |
| "soon" | "by Friday 3 October" |
Word choice
The table below is a working list. "Instead of" items are not banned; they are defaults to question.
| Type | Instead of | Write | Note |
|---|---|---|---|
| wordy | in order to | to | "in order to" occasionally helps avoid ambiguity |
| wordy | due to the fact that / owing to the fact that | because | |
| wordy | in spite of the fact that | although | |
| wordy | at this point in time / at the present time | now | |
| wordy | in the event that | if | |
| wordy | with regard to / in relation to | about, on | |
| wordy | for the purpose of | to, for | |
| wordy | has the ability to / is able to | can | |
| wordy | a large number of / a majority of | many / most | |
| wordy | prior to / subsequent to | before / after | |
| wordy | in close proximity to | near | |
| wordy | on a daily basis | daily | |
| wordy | the question as to whether | whether | Strunk's own example |
| wordy | it is important to note that | (delete) | if it is important, the reader will notice |
| wordy | in terms of | (rewrite) | usually hides a vague relation |
| inflated | utilize / utilization | use | |
| inflated | facilitate | help, enable, run | |
| inflated | commence / terminate | start / end | |
| inflated | endeavor | try | |
| inflated | sufficient / numerous | enough / many | |
| inflated | approximately | about | keep "approximately" in technical specs if house style wants it |
| nominalisation | make a decision | decide | |
| nominalisation | conduct an investigation | investigate | |
| nominalisation | provide assistance | help | |
| nominalisation | give consideration to | consider | |
| nominalisation | achieve a reduction in | reduce | |
| redundant pair | each and every | each, every | |
| redundant pair | first and foremost | first | |
| redundant pair | null and void | void | |
| redundancy | end result / final outcome | result / outcome | |
| redundancy | past history / future plans | history / plans | |
| redundancy | basic fundamentals | fundamentals | |
| redundancy | completely finished / very unique | finished / unique | |
| redundancy | close proximity / advance planning | near / planning | |
| hedge | I think / I believe / it seems that | (delete) or give the confidence level | the byline already says it is your view |
| hedge | quite, rather, somewhat, fairly, relatively | (delete) or quantify | "relatively fast" → "twice as fast as v1" |
| hedge | may potentially / might possibly | may / might | double hedge |
| weasel | some people say / it is widely believed | name who, cite where | |
| weasel | up to 50% faster | typically 30% faster (median of N runs) | "up to" hides the typical case |
| weasel | significantly (not statistical) | the number | "significant" implies a test in technical writing |
| weasel | industry-leading, best-in-class, world-class | the evidence | readers discount unsupported superlatives |
| weasel | virtually, essentially, basically | (delete) or be precise | |
| jargon | leverage (verb) | use | |
| jargon | synergy | what exactly combines, and the gain | |
| jargon | bandwidth (for time) | time, capacity | |
| jargon | circle back / touch base | "I'll reply by Tuesday" | give a time |
| jargon | move the needle | raise X by Y | |
| jargon | going forward | from now on, or delete | |
| jargon | impactful | effective, or state the effect | |
| jargon | learnings | lessons, findings | |
| jargon | robust / scalable / seamless | the actual property: "handles 10× traffic" | adjectives without numbers |
Punctuation essentials
| Mark | Rule | Example |
|---|---|---|
| comma, non-restrictive clause | commas around information that could be removed without changing which thing you mean | "My brother, who lives in Leeds, is a doctor." (one brother) |
| no comma, restrictive clause | no commas when the clause identifies which one | "My brother who lives in Leeds is a doctor." (implies other brothers) |
| that vs which | US style reserves which for non-restrictive clauses; British usage allows restrictive which | "the bug that broke login" / "the bug, which broke login, …" |
| comma after long introductory element | helps parsing | "After the migration finished, we reran the tests." |
| comma splice | don't join two independent clauses with only a comma | "The build failed, we rolled back." → full stop or semicolon |
| semicolon | joins closely related independent clauses; separates list items that contain commas | "The build failed; we rolled back." / "Leeds, UK; Lyon, France; Porto, Portugal" |
| colon | introduces what the first clause promises: a list, an explanation, an example | "One thing blocks launch: the payment provider." |
| em dash (—) | a strong break for an aside or an amplification; use sparingly | "The fix — one line — took three days to find." |
| en dash (–) | ranges and connections; British style often uses a spaced en dash where US uses an em dash | "pages 10–20", "London–Paris route" |
| hyphen | compound modifiers before a noun | "a well-known issue", "an open-source tool", but "the tool is open source" |
| apostrophe | possession and contraction, never plurals | "the team's plan", "it's" (= it is), "its" (possessive), "APIs" |
| Oxford (serial) comma | comma before the final and/or in a list of three or more | "Rust, Go, and Python" |
The Oxford comma is a style choice, not a rule. Chicago, the Google developer style guide and the Microsoft style guide use it; AP and many British newspapers generally don't. Pick one per document and use it consistently; add one anyway wherever omitting it creates ambiguity ("I thanked my parents, Ayn Rand and God"). The comma is not always trivial: in O'Connor v. Oakhurst Dairy (US Court of Appeals, First Circuit, 2017) a missing serial comma in a Maine overtime exemption decided the case in the drivers' favor.
Readability formulas
Formulas estimate difficulty from surface features: sentence length and word length (syllables). With = words, = sentences and = syllables:
Flesch Reading Ease (Rudolf Flesch, 1948; higher is easier, roughly 0–100):
Flesch–Kincaid Grade Level (Kincaid and colleagues for the US Navy, 1975; approximate US school grade):
| FRE score | US grade | Description |
|---|---|---|
| 90–100 | 5th | very easy |
| 80–90 | 6th | easy; conversational |
| 70–80 | 7th | fairly easy |
| 60–70 | 8th–9th | plain English |
| 50–60 | 10th–12th | fairly difficult |
| 30–50 | college | difficult |
| 10–30 | college graduate | very difficult |
| 0–10 | professional | extremely difficult |
Worked example. A 100-word paragraph with 5 sentences and 150 syllables: , , so and . Rewritten as 8 sentences with 135 syllables (, ): FRE , FKGL .
Other formulas (Gunning fog, SMOG, Coleman–Liau, Automated Readability Index) use similar inputs and give similar rankings.
Limits. Formulas were built on school texts and Navy manuals and measure proxies, not comprehension.
| Formula blind spot | Consequence |
|---|---|
| meaning and logic | a shuffled paragraph scores the same as the original |
| word familiarity | "Kubernetes" counts as 4 syllables of difficulty; "ontic" counts as easy |
| reader knowledge | a specialist reads dense jargon faster than a layperson reads "easy" text |
| structure and layout | headings, lists and examples don't count |
| gaming | chopping sentences raises the score and can reduce cohesion |
Use a formula as a smoke alarm (a grade of 16 on a customer email means something is wrong), never as a target. The real test is a reader: ask someone from the audience what the text says and what they would do next.
Editing process
Drafting and editing are different jobs. Draft fast without judging; edit slowly in passes, from large to small, because fixing words in a paragraph you will delete is wasted work.
Structure
- Is the main point in the first paragraph (or the first line of an email)?
- Does the order match the reader's questions?
- Skim test: read only headings and first sentences. Does the argument survive?
- Cut or move whole sections that don't serve the point.
Paragraphs
- One idea each, stated in the topic sentence.
- Each paragraph follows from the last; add the missing link where it doesn't.
- Split paragraphs longer than about 6–8 lines on screen.
Sentences
- Agent as subject, action as verb; convert zombie nouns.
- Given before new; the important element last.
- Vary length; split sentences with two main clauses joined by "and".
Words
- Run the word-choice table: wordy phrases, hedges, weasels, jargon.
- Replace vague adjectives with numbers or examples.
- Same term for the same thing throughout.
Proofread
- Names, numbers, dates, links, attachments.
- Read aloud or with text-to-speech; stumbles mark bad sentences.
- Last: spelling and punctuation (a spell-checker, then your eyes).
| Technique | Why it works |
|---|---|
| Read aloud | the ear catches rhythm problems, missing words and sentences you run out of breath in; Paul Graham recommends it in "Writing, Briefly" (2005) |
| Cut 10% | Stephen King, in On Writing (2000), describes adopting a rejection slip's formula: second draft = first draft minus 10% |
| Wait | a day's gap reduces the curse of knowledge; you read what is there, not what you meant |
| Change the medium | print it, change the font, read on your phone; familiarity hides errors |
| Reverse outline | write one line per paragraph of the finished draft; gaps and repeats become obvious |
| Outsider read | the only direct measure of clarity; ask "what is this asking you to do?" |
| Search for tics | grep your own crutch words (just, very, really, actually, basically, in order to) |
Common mistakes
| Mistake | Symptom | Fix |
|---|---|---|
| writing before thinking | the conclusion appears in the last paragraph | move it to the top; rewrite from there |
| burying the lede | background first, point on page 2 | BLUF |
| curse of knowledge | undefined terms, skipped steps | outsider read; define on first use |
| throat-clearing | "I wanted to reach out to touch base about…" | start with the second sentence |
| zombie nouns | "-tion", "-ment", "-ance" everywhere; weak verbs | put the action back into the verb |
| hedging everything | "perhaps", "arguably", "somewhat" in every paragraph | hedge once, precisely |
| wall of text | 15-line paragraphs, no headings | one idea per paragraph; descriptive headings |
| bullet soup | lists of fragments with no logic between them | prose for reasoning |
| inconsistent terms | "user", "customer", "account holder" for the same thing | pick one; add a glossary |
| vague quantities | "significant growth", "soon", "many users" | numbers, dates, names |
| elegant variation | synonyms for the same concept to avoid repetition | repeat the term; readers assume a new word means a new thing |
| over-applying rules | no passives ever, no sentence over 20 words | rules serve the reader, not the other way round |
| one-pass editing | typos fixed in a structurally broken draft | edit in passes, largest first |
Templates
Editing checklist
STRUCTURE
[ ] Main point stated in the first paragraph / first line
[ ] I can say the point in one sentence
[ ] Reader named; I know what they should do next
[ ] Headings + first sentences alone carry the argument
[ ] Every section serves the main point (else cut)
PARAGRAPHS
[ ] One idea per paragraph, stated first
[ ] Each paragraph follows from the previous one
[ ] No paragraph longer than ~8 lines on screen
SENTENCES
[ ] Agent is the subject, action is the verb
[ ] Zombie nouns turned back into verbs
[ ] Given before new; important words at the end
[ ] Passives only where they help
[ ] Sentence length varies; average around 15-20 words
WORDS
[ ] Wordy phrases, hedges, weasel words removed
[ ] Jargon defined or replaced for this reader
[ ] Vague adjectives replaced by numbers or examples
[ ] One term per concept, used consistently
PROOF
[ ] Read aloud once
[ ] Names, numbers, dates, links checked
[ ] Cut ~10% from the first draftOne-point paragraph
[Topic sentence: the claim, specific and testable.]
[Evidence: a number, an example, or a source.]
[Explanation: why the evidence supports the claim.]
[Consequence: what it means for the reader / what next.]
Example:
Onboarding loses 40% of sign-ups at email verification.
In September, 6,200 of 15,500 new accounts never clicked
the link. Most were on mobile, where the link opens a
different browser and loses the session. Letting users
verify with a six-digit code should recover about half.BLUF status update
Subject: [Project] status: [on track / at risk / blocked]
- [one-line headline]
BLUF: [One or two sentences: state, and what you need.]
Progress since [date]:
- [Outcome, not activity]
- [Outcome]
Risks / blockers:
- [Risk] - [impact] - [what we're doing / what we need]
Next: [what happens by when]
Ask: [specific request, owner, deadline] (or "none")Explain-it-to-a-newcomer paragraph
[Term] is [plain one-line definition].
It matters because [consequence the reader cares about].
For example, [concrete instance with real names/numbers].
It is not [common confusion]; that is [the other thing].References
- George Orwell, "Politics and the English Language" (1946) (opens in a new tab): the six rules and the four questions, text from the Orwell Foundation
- William Strunk Jr, The Elements of Style (1918), Project Gutenberg (opens in a new tab): Strunk's original rules, including rule 13, "Omit needless words"
- Geoffrey K. Pullum, "50 Years of Stupid Grammar Advice" (Chronicle of Higher Education, 2009) (opens in a new tab): the linguist's critique of Strunk and White, especially on the passive
- William Zinsser, On Writing Well (Harper & Row, 1976; later editions HarperCollins): clutter, simplicity, the nonfiction classic
- Steven Pinker, The Sense of Style (Viking, 2014): classic style, the curse of knowledge, zombie nouns, given-new
- Francis-Noël Thomas and Mark Turner, Clear and Simple as the Truth (Princeton University Press, 1994): the source of "classic style"
- Helen Sword, "Zombie Nouns" (New York Times Opinionator, 23 July 2012) (opens in a new tab): the term for nominalisations
- Joseph M. Williams, Style: Lessons in Clarity and Grace (Pearson; many editions): characters as subjects, actions as verbs, old-before-new
- Plain Writing Act of 2010, Public Law 111-274 (opens in a new tab): the statutory definition of plain writing
- Digital.gov plain language guides (opens in a new tab): successor to plainlanguage.gov and the Federal Plain Language Guidelines (2011)
- Flesch–Kincaid readability tests (Wikipedia) (opens in a new tab): formulas, score bands and history
- Kincaid et al. (1975), Derivation of New Readability Formulas for Navy Enlisted Personnel (ERIC) (opens in a new tab): origin of the grade-level formula
- Paul Graham, "Writing, Briefly" (2005) (opens in a new tab): a one-paragraph writing method
- Paul Graham, "Write Simply" (2021) (opens in a new tab): the case for ordinary words and simple sentences
- Paul Graham, "Putting Ideas into Words" (2022) (opens in a new tab): writing as a test of what you know
- Stephen King, On Writing: A Memoir of the Craft (Scribner, 2000): the "minus 10%" revision rule
- Persuasive writing and technical writing: the companion sheets