Writing & Expression: Clarity in Academic WritingAbstract · Review · Argument · Against Jargon
BigCat's Writing
Academic prose is famously hard to read — but it is almost never hard because the ideas run deep. It is hard because the writing is bad: obscurity as a gatekeeper, nominalization to hide emptiness, citation-dumping posing as a review. This issue borrows four people who actually write well, and hauls academic writing back to what it should be: the clear transmission of a thought. Engineers' RFCs, design docs, and promo packets benefit just as much.
Principle 01
The Abstract: Tell a Whole Story in Four Sentences
The Abstract — Four Sentences
SPJ · Conclusion First
The Principle + A Master's Words
An abstract is not a table of contents; it is your paper's elevator pitch. Most readers decide whether to read on based on the abstract alone — so it must tell a complete story by itself, and the conclusion must appear. No withholding for suspense.
"I usually structure the abstract in four sentences: state the problem; say why it's interesting; say what your solution achieves; say what follows from it."
— Simon Peyton Jones, "How to Write a Great Research Paper"
Why It Works
When a reader decides whether to invest half an hour in the full text, they weigh just three things: Is this relevant to me? Is it credible? What do I take away? Four sentences answer all three. Academic writing's worst habit is addiction to throat-clearing — three paragraphs of background before the point, by which time the reader is gone. The abstract pushes the pyramid principle to its limit: conclusion first, and only one layer deep.
1Problem What real, unsolved problem does this field have?
2Why it matters Once solved, who benefits?
3What you did / the result What is your solution, and what measurable result did it get?
4What it means What larger lesson follows from that result?
SPJ's four-sentence abstract: if you can't write the fourth sentence, the work itself hasn't been thought through
Revision in Action
This paper conducts an in-depth study and discussion of the consistency problem in distributed systems; it first introduces the background, then analyzes shortcomings of existing methods, and finally proposes a new method with experimental verification, achieving good results.Cross-region databases cannot have both strong consistency and low latency, which blocks financial workloads from the cloud. We present X, which cuts cross-region read latency from 200ms to 40ms without violating business invariants. A principled "weak but sufficient" middle ground beats the all-or-nothing debate.The first only says "I studied X"; the second lands all four: problem, importance, result (with a number), implication.
Where to Use It + Common Mistakes
✓ Paper abstracts, the TL;DR of an RFC, the one-paragraph summary opening a design doc, the impact summary in a promo packet, the first line of a PR description
✗ Stating "I studied X" without "here is what I found" — a topic is not a contribution
✗ Filling the space with "fairly good," "to some extent," "explored" — zero-information filler
✗ Cramming method detail into the abstract; the reader wants the conclusion, not the procedure
This Week's Exercise + A Question
Take a recent design doc or PR. Delete its current summary and rewrite it in SPJ's four sentences: problem / why it matters / what you did / what follows — each under 20 words. Question: If you can't write the fourth sentence — "what follows" — has the meaning of the work itself been thought through yet?
Principle 02
The Literature Review: Join a Conversation, Not a Roll Call
Literature Review — Join the Conversation
Burke · Conversation, Not a List
The Principle + A Master's Words
A review is not proof that "I've read a lot." It places your work inside an ongoing conversation. A bad review is a ledger: "A said…; B said…; C said…". A good one has a line running through it: what the field is arguing about, where it is stuck, and which sentence you are answering.
"Imagine that you enter a parlor. You come late. … The discussion had already begun long before any of them got there. … You listen for a while, until you decide that you have caught the tenor of the argument; then you put in your oar."
— Kenneth Burke, "The Philosophy of Literary Form" (1941)
Why It Works
A review's real job is to draw a map of what's known for the reader and to mark one blank spot on it — that blank is the reason your work exists (the gap). A review without a gap gives the reader no "so what." Burke's parlor reminds you: you are neither the first to speak nor the last; your contribution only means something once it is slotted into the conversation. An engineer's Related Work is the same — not a disclaimer, but a set of coordinates.
Revision in Action
On cache consistency, Smith (2019) proposed method A, Jones (2020) proposed method B, and Lee (2021) studied C. These methods each have pros and cons.Work on cache coherence has swung between speed and correctness: early designs (Smith 2019) traded consistency for latency, while later protocol-heavy approaches (Jones 2020) restored it at 3× the cost. What no one has shown is whether "good-enough" consistency is achievable without protocol overhead — the gap this paper addresses.The ledger becomes a line of tension that lands on the gap, so the reader instantly sees why you need to enter.
Where to Use It + Common Mistakes
✓ A paper's Related Work, the "existing options" comparison in a design doc, a tech-selection report, the "why others couldn't do it" in a promo packet
✗ The ledger — one sentence per paper, no through-line; or all praise, or all dismissal
✗ Ending the review without landing on a gap, leaving the reader with no "so what are you going to do"
✗ Piling on unrelated citations to look well-read — a review is a filter, not an inventory count
This Week's Exercise + A Question
Pick a technical field you know. In three sentences, write its conversation: how people used to do it → what problem showed up → where it's stuck now. Then add one sentence for what you'd say next. Question: If you draw that map honestly, is your work filling the blank, or repeating what others have already done?
Principle 03
Argument Structure: Half a Sentence's Meaning Hides in Position
Argument Structure — Reader Expectations
Gopen & Swan · Reader Expectations
The Principle + A Master's Words
Readers have fixed expectations about where information should appear: the start of a sentence (the topic position) holds old information and the known character; the end (the stress position) holds new information, the thing you want emphasized. Violate that expectation and even an accurate sentence reads badly.
"Readers do not simply read; they interpret. … The information a sentence contains is only part of what a reader gets; the other part comes from where in the sentence that information appears."
— Gopen & Swan, "The Science of Scientific Writing," American Scientist (1990)
Why It Works
This holds in English and Chinese alike. A reader treats the sentence's opening as "who this is about" and its ending as "where the emphasis is." Bury the conclusion mid-sentence, or open every sentence with an unfamiliar new term, and the reader's interpretation engine spins in neutral. Academic arguments become unreadable not usually because the logic broke, but because the old-info-to-new-info relay broke: the end of one sentence never handed off to the start of the next.
Topic position (start)old info · known character
…move forward
Stress position (end)new info · what to emphasize
Readers naturally read the start as "who" and the end as "the point" — write with that expectation and the sentence clears itself up
Revision in Action
A causal-order-based conflict detection algorithm was adopted in our system to address the data inconsistency that arises during concurrent writes, which is the core innovation of our method.Concurrent writes cause data inconsistency. Our fix is a causal-order conflict detector that catches conflicts at write time, not after."Core innovation" moves from the buried tail into the stress position; the known actor returns to the topic position; passive becomes active.
Where to Use It + Common Mistakes
✓ Paper bodies, the argument paragraphs of a design doc, decision memos, any long text that must persuade step by step
✗ Key conclusions buried mid-sentence or mid-paragraph, so the reader hits the point only at the end
✗ Every sentence opening with an unfamiliar new term, so it never connects to the one before
✗ Passive voice overused to hide "who did it"; paragraphs with no topic sentence
This Week's Exercise + A Question
Take a paragraph of your own technical argument and check each sentence for two things: does it open with the known character? Does the word most worth remembering land at the end? Rewrite the three that violate this. Question: Chinese drops subjects easily (subjectless sentences). Does that make the topic-position rule harder to use, or actually easier?
Principle 04
Against Jargon: Kill the Zombie Nouns
Against Jargon — Zombie Nouns
Sword · Orwell · Anti-Nominalization
The Principle + A Master's Words
Academic jargon's chief symptom is nominalization — crushing verbs and adjectives into abstract nouns, so the sentence loses its action and its people. Helen Sword calls them "zombie nouns": they eat verbs, drain adjectives, and swap living humans for abstract entities.
"Nominalizations are the zombie nouns of academic prose: they cannibalize active verbs, suck the lifeblood from adjectives and substitute abstract entities for human beings."
— Helen Sword, "Zombie Nouns," New York Times (2012)
"Never use a long word where a short one will do."
— George Orwell, "Politics and the English Language" (1946)
Why It Works
Why do smart people write obscurely? Pinker's answer is the curse of knowledge — you are so steeped in your own material that you forget the reader isn't inside your head, and you drop the bridge that would let them follow. Nominalization and jargon also serve a less honorable purpose: cover. "Optimization was carried out" need not say what was optimized, or by how much; "there exist certain problems" need not say whose. Vagueness is the grammar of dodging responsibility; clarity is a form of honesty — it forces you to commit.
Revision in Action
The implementation of an optimization of system performance was achieved through the modification of the caching strategy.We changed the caching strategy to make the system faster.Three stacked noun-phrases plus "was achieved" compress into one subject and one verb.
Owing to certain objective reasons, a degree of challenge exists in the course of the project's advancement.We're short two engineers, so the launch slips three weeks.Jargon's other use is evasion — translate it back into the concrete and the responsibility and the facts reappear.
Where to Use It + Common Mistakes
✓ All formal writing — papers, RFCs, weekly reports, postmortems, promo packets. The more you want to "look professional," the more to watch out
✗ Using "carry out / undertake / achieve + noun" in place of one verb (carry out an optimization → optimize)
✗ Making an abstract noun the subject, so no human appears ("the occurrence of the phenomenon")
✗ Mistaking long words for expertise; using vague words to dodge concrete commitments ("to some extent," "the relevant aspects")
This Week's Exercise + A Question
Take a recent piece of your formal writing. Circle every "X of Y of Z" noun-stack and every "carry out / undertake + noun," and rewrite each as "who + verb + what." Count the words you deleted. Question: Is there a case where the vague jargon is deliberate — where you don't actually want to commit? Is that a writing problem, or a not-yet-thought-through problem?
Going Deeper
Clarity is demanded, but some ideas are genuinely intricate. Does saying them simply distort them?
A complex idea and a complex sentence are two different things. Einstein: "as simple as possible, but no simpler." Clarity doesn't simplify the content; it removes the obstacles to understanding — so the reader spends their cognitive budget on the idea, not on decoding the sentence. Intricate ideas need the clearest prose precisely so they aren't misread. If a claim only holds when stated obliquely, that's usually not depth — it hasn't been thought through, or obscurity is hiding a void.
Is Chinese "translationese jargon" the same disease as English nominalization?
Same root, different form. English nominalizes through -tion/-ment suffixes; Chinese does it through stacked "de" phrases, "carry out / undertake + noun," and "bei + verb" structures borrowed from translation. The root cause in both is avoiding action and subject. But Chinese has its own strains: subjectless sentences make "who did it" vanish more easily, and the "-ness / -ize / -degree" suffixes can coin words endlessly. The cure is the same — recover the subject, turn nouns back into verbs, unpack "X of Y" into "who did Y."
"Conclusion first" is obvious in papers, but the humanities prize slow buildup to a final reveal. Does clarity have disciplinary borders?
It does, but far narrower than most assume. Suspense structure genuinely has value in literature and philosophical essays — the process is the content. Even so, readers still need signposts to know where they are. Most so-called "disciplinary style" is a pretext for bad habits. The test: is your indirection serving the reader's understanding, or covering the fact that you haven't figured it out yourself? The former is art; the latter is jargon.
Do technical docs (RFCs, design docs) and academic papers hold clarity to the same standard?
Same core — conclusion first, honor reader expectations, avoid jargon — but different weights. A technical doc's reader is "a colleague who has to decide," so front-load the conclusion aggressively and make it scannable; a paper must survive peer review, so it needs a full chain of argument and its place in the literature. What they share: both get pushed toward obscurity by the urge to "seem rigorous," and both should resist. The zombie nouns in your RFC deserve killing just as much as the ones in a paper.
AI can already "translate" jargon-laden papers into clear ones, even write the abstract. Is learning to write clearly still worth it?
More worth it — the center of gravity just shifted. AI can polish sentences, kill nominalizations, apply an abstract template — the mechanical layer, faster than you. But it can't replace three things: judging where the gap actually is, deciding what the work truly means (the abstract's fourth sentence), and telling whether a review has a real through-line or is just fluent emptiness. The bedrock of clear writing is always clear thinking; AI makes clarity of expression cheap, which forces clarity of thought into being the only scarce thing.