Notes on writing well
Working notes, by Michael Nielsen: These are rough notes. I'm writing them to help me improve my polished writing, not as models of prose in themselves.
It's hard to write the truth: When I say it's hard to write the truth, it sounds as though I'm saying that I lie a lot, or I'm accusing other writers of lying. That's not my intent. If your writing has novelty, then early drafts will likely be, at best, rough approximations to the truth. Those drafts will contain sentences and paragraphs which are the written equivalent of those conversational fumbles where we finally burst out with "You know what I mean". Unfortunately, when writing it's easy not to notice when we make such fumbles, and to settle for writing which is a poor approximation to the truth. The only way I know to address this problem is to revise over and over, asking repeatedly of each sentence whether it can be sharper, and if we really, truly, believe the sentence.
Good writing is good psychology: Advice on writing is sometimes presented in a tough-sounding, apparently objective style. "Avoid adverbs". "Avoid needless words". And so on. Justifications for such rules are either omitted entirely, or couched in generalities about what makes good writing. But good writing isn't an intrinsic property of the text. It's a quality of the relationship between the reader and the text. And so useful advice on writing is about how to change that relationship. Good writers build up a theory of that relationship, a psychology of reading. The better a writer's psychology of reading, the better their writing. Good writing is really an exercise in applied psychology.
Beware words that you wouldn't use while speaking: Using such words in your writing isn't always a mistake, but it must be done consciously, and for good reasons. I'm often tempted to use such words to appear authoritative, writing in what I imagine is expert language. Unfortunately, lacking practice with that language I'm likely to use it poorly, unless I'm exceptionally careful. This desire to appear authoritative can easily become a self-serving agenda, not an agenda in service to the reader. With that said, if you're confident you're using the word in the reader's service, go for it!
One purpose: A good piece of writing has a single, sharp, overriding purpose. Every part of the writing --- even the digressions! --- should serve that purpose. Put another way, clarity of overall purpose is an absolute requirement in a good piece of writing.
A common pattern is to begin a piece with a problem or a mystery or a promise. Resolving the problem or mystery or fulfilling the promise is what defines the purpose of the piece. Of course, your writing may take the reader on a journey, gradually refining (and perhaps redefining) the purpose. But it nonetheless remains with the reader and the author. "Destroy the One Ring" is our obsession in The Lord of the Rings. The problem confronted in Where Do Good Ideas Come From? needs no elaboration. The Language Instinct promises to explore the idea that much of the structure of language is innate to our brains, not learned. Each book focuses relentlessly on its purpose.
I'm not sure I really understand why clarity of purpose matters so much to good writing. Yet writing which violates it nearly always suffers. Lois Bujold begins one of her books with a mystery, which she resolves partway through the book. The book then changes to be about something else. It's telling that I don't remember which of her books it is, or more plot details.
Perhaps an explanation lies in the nature of our emotional bond to a book. That bond is formed early, and if changes too much, the bond is severed. The author may try to establish a new bond, but we-the-reader become confused. Are we supposed to commit again? What is the book really about?
There are examples which successfully evade this rule. I'm not sure what Douglas Adams' books are about, but I enjoy them for the sheer sentence-by-sentence pleasure. Perhaps that is what Adams' books are about. I'm also not sure what Robert Pirsig's Zen and the Art of Motorcycle Maintenance is about, but it's a terrific book. Such examples aside, most successful writing has a single overriding purpose.
This observation matters because it's often tempting to let your purpose expand and become vague. Writing a piece about gardens? Hey, why not include that important related thought you had about rainforests? Now you have a piece that's sort of about gardens and sort of about rainforests, and not really about anything. The reader can no longer bond to it. This expansion in purpose is done with the best of intentions --- you want to tell the reader all your good ideas. But it breaks the clarity of your communication with the reader. It's a type of discursive disease.
Discursive disease is difficult to avoid. The only surefire solution is to rewrite, culling and editing ruthlessly until you have a single purpose. But a partial solution is to make sure you always know your purpose. I'm currently experimenting with writing the purpose of pieces I'm working on into the head of the file for that piece. I'll then review it each time I sit down to write.
This is not to say that the purpose should be fixed. If you're writing to discover then the purpose will evolve over time, and perhaps change radically. That's part of why you're writing. But your final piece should reflect only the purpose that you discovered during the writing. All else, no matter how good, must be removed. File it somewhere where you can find it again later, for use in future writing projects. When Aaron Sorkin wrote the television movie The American President he gathered far more material than he could use in the final script. He put the extra material aside, and later it helped him write his television series The West Wing.
A complicating factor is that sometimes you need to explore beyond the boundaries of your current purpose. You're writing for purpose A, but your instinct says that you need to explore subject B. Unfortunately, you're not yet sure how subject B fits in. If that's the case then you must take time to explore, and to understand how, if at all, subject B fits in, and whether you need to revise your purpose. This is emotionally difficult. It creates uncertainty, and you may feel as though your work on subject B is wasted effort. These doubts must be resisted.
"What is your book (or essay) about?" You must have a compelling answer to this. If you don't, then your book or essay almost certainly doesn't have a single overriding purpose.
Occam's razor for writing: Occam's razor applies not just to science but also to writing: given two written passages which convey almost the same meaning, the shorter passage is usually better. The reason is that clearer, more vivid explanations are preferable. And shortening usually makes writing clearer and more vivid. At least in part this is a quirk of our brains: it is usually better to say something is "big" than "humungous"; that a day is "cloudy" rather than
"Write one true sentence:" This was Hemingway's method for getting started when he was blocked:
Sometimes when I was starting a new story and I could not get it going, I would sit in front of the fire and squeeze the peel of the little oranges into the edge of the flame and watch the sputter of blue that they made. I would stand and look out over the roofs of Paris and think, "Do not worry. You have always written before and you will write now. All you have to do is write one true sentence. Write the truest sentence that you know." So finally I would write one true sentence, and then go on from there. It was easy then because there was always one true sentence that I knew or had seen or had heard someone say. If I started to write elaborately, or like someone introducing or presenting something, I found that I could cut that scrollwork or ornament out and throw it away and start with the first true simple declarative sentence I had written.
This is not always as easy as Hemingway suggests. The mind may be blank, unstimulated. One must be provoked by some external mental stimulus --- preferably either to irritation or to excitement. Either provocation will, eventually, produce a true sentence or three, as you engage with whatever is irritating or exciting you. And so you must also cultivate methods for stimulating your mind in this way.
Stop when you know what comes next:
The best way is always to stop when you are going good and when you know what will happen next. If you do that every day when you are writing a novel you will never be stuck. That is the most valuable thing I can tell you so try to remember it.
This works best when also combined with a word quota. Meet one's word quota for the day, and after that, stop where you know what happens next.
Finding what's new: This may be finding a new result, it may be digging out a narrative. It's about what's going on when you're having trouble, when a difficulty is presenting itself. It's about patience. It's about finding new sources of stimuli. It's about trying things which you know won't work, as a way of learning new things.
On prefaces: (1) Most prefaces are awful. They drone on, with little idea of what they should accomplish. (2) An introduction or preface is a literary form in its own right. (3) A good preface explains what a book is about, who it is for, and what that person can expect to learn. It omits everything obvious, and only explains what is strikingly different about the book's answers to those questions. A preface is good very nearly in proportion to the extent that has a striking of view. (4) It takes patience to write a good preface. Doing so requires a superb understanding of what your book is trying to achieve, and it may require many attempts to figure that out. (5) Steven Pinker's preface to "The Language Instinct" is very good, as prefaces go. Zed Shaw's for "Learn Python the Hard Way" is also pretty good. But for each good example, I've found ten that are bad, or mediocre. And, frankly, even the best prefaces aren't all that good. The reason, I think, is people don't want to be reading about the book any more than absolutely necessary, they want to be enjoying the actual subject at hand. For that reason, we have: (6) If material for the preface can be delayed until later in the book, it should be. (7) A good preface is, for all the reasons just enumerated, short. How short? In some ways, it'd be good to omit it entirely. The problem is that people do expect to be able to find out what the book is about, who it's for, and what they can expect to learn. And so a preface is quite conventional. In a sense, the preface is metadata. As an author, one shouldn't have the expectation that the preface is amazing. Readers don't have this expectation, either. It's about the book, not the subject at hand, and thus is intrinsically less interesting. (8) With that said, the more opinionated and unusual the preface, the better. (9) Many genres get away without a preface. Much fiction tells you what it's about in the first or second chapter, setting up some big problem which needs to be solved. That can also sometimes work in non-fiction. The opening then becomes an introduction: it sets up some big problem to be solved, with only a few minor prefatory comments. A Brief History of Nearly Everything uses this structure extremely effectively. When one can get away with it --- when the problem addressed can be made approachable --- I think this is preferable to writing a preface.
On note-taking: Something I wish I'd understood earlier is how important it is to get good at note-taking. It is not trivial to do well. Rather, it is something that one can get better and better at. The payoff of improving is that it makes writing far easier. The better your notes, the easier the writing will be. There are several elements to it: (1) Primary note-taking; (2) Queueing; (3) Refactoring and organization. You need to get good at all.
A Brief History of Time (Hawking): Amusing, in multiple ways: the self-reference of having a history of time; the self-reference that it's brief; and finally the fact that the title is a good literal description of the book.
The Future of Ideas (Lessig): I'm not sure this is a good title. It's ambitious and provocative, a big-picture title. It almost forces the reader to confront the question: what is the future of ideas? But it's huge scope also means that it's vague. And, in the end, it's not what the book is about. The book is really about the future of expression, i.e., the future of how ideas and culture are expressed. This makes the title elusive, the kind which, after reading the book, you may not recall.
The Society of Mind (Minsky): This title really is exactly what the book is about. After you've read the book you won't have any difficulty recalling the title. A problem with the title is that it's not obvious to prospective readers what it means. It does, at least, sound pregnant with meaning, and it's also an idea that is relatively easy to explain. It might, perhaps, have been helped by a subtitle: "how the mind arises from many simple agents", or something in that vein. That's not quite the right subtitle, but something along those lines might have worked.
Trimalchio in West Egg: F. Scott Fitzgerald's alternate (and sometimes preferred) title for "The Great Gatsby". Despite being advised against the title by both his editor and his wife, Fitzgerald sometimes preferred "Trimalchio"-themed titles right up to publication. A reminder that even very good writers can have and hold onto very bad ideas.
Juxtaposition: Many titles juxtapose two words (or concepts) that are not usually together. This is particularly effective when the juxtaposition is surprising, but in retrospect meaningful. "The Selfish Gene" is, for example, more surprising than "The Future of Ideas". "Future Shock" is an excellent juxtaposition.
Three types of titles: (1) Books that describe a subject or field. These titles name the subject or field, perhaps with a simple modifier: Introduction to algorithms; Principles of neuroscience; Quantum Computation and Quantum Information. (2) Journalistic books that are principally reporting. These may follow the pattern of (1), or perhaps have a quirkier title that reflects some aspect of the subject or field under investigation. In the plex is an example of the quirky approach. Complexity is an example that follows (1). (3) Idea books, i.e., books which explore a new concept or propose a theory. Here, the title ideally names the central new concept or theory. The Selfish Gene. The Ingenuity Gap. Everything is Miscellaneous. On the Origin of Species by Means of Natural Selection. Biology is Technology. Failing that, the title describes the problem to be addressed: The Trouble With Physics. Reinventing Discovery.
Adjective-noun title forms: A common form for the big idea in big-idea books. The Selfish Gene. Free Culture (free has a double meaning as a verb here). The Long Tail. Collective Intelligence. Global Brain. Hey, this can be turned into a fun mad libs game! I'll now generate two lists, drawn from the books on the bookcase immediately beside me. A few of these words aren't usually used as adjectives or nouns, but they are in the books titles I'm quoting, so I'll run with it. I've taken the first twenty adjectives, and the last twenty nouns (i.e., they're being drawn from different books on my shelf):
selfish, free, long, collective, global, public, parallel, distributed, critical, digital, big, open source, fab, economic, wiki, lifelong, information, administrative, artificial, formal
principle, wikinomics, 2.0, surplus, smog, genius, optimist, mobs, bubble, intelligence, manifesto, culture, innovation, chronicles, crunchers, tail, University, brain, science, cyborg
These lists suggest many amusing or stimulating titles. Some I'd certainly consider reading. Here are a few that tickled me: The Selfish Brain. Administrative Genius. The Lifelong Optimist. The Free Cyborg. The Distributed University. And they suggest derivatives as well: The Information Deficit. A Manifesto for Ignorance. And so on. Not bad for ten minutes work. It's also notable that this process is easily scaled, and won't hit the point of diminishing returns very quickly. Spend ten times as long, get ten times as many titles.
A cynic might say that all this proves is that our public intellectual life is a narrow game. I think that's wrong (though The Narrow Game is a good title, hmm...) Instead, I think the message is optimistic: it's easy to generate good, novel ideas and questions!
Noun-is-noun title forms: Everything bad is good for you. Biology is technology. Everything is illuminated. This is not a novel. Again, acquires power through contrast in juxtaposition. I love the title Everything is illuminated: it's simple, yet suggestive of so much. While it doesn't produce quite as many hits as adjective-noun forms, it's still fun to play mad libs style.
In "Structure and Interpretation of Computer Programs":
- "Building Abstractions with Procedures"
- "Building Abstractions with Data"
- "Modularity, Objects, and State"
- "Metalinguistic Abstraction"
- "Computing with Register Machines"
Chapter titles 1, 2 and 5 are verb phrases, while 3 and 4 are noun phrases. I prefer the more active titles, perhaps for the obvious reason that they create more of a sense of going someplace!
In "The Language Instinct": Here are the two best of the chapter titles:
"An Instinct to Acquire an Art": I love this title. It is just very slightly false: language is not really an art. But what would be better? "Craft"? "Habit"? "Ability"? None quite works, and I can think of nothing else, either.
What I really like about this title is that it actually makes us smarter, by succinctly summarizing a complex set of ideas.
"The Language Mavens": I don't know if the coinage is original, but this is a beautiful term for a class of people that we all recognize at once. It also conveys Pinker's point of view --- he has some time for these people, but he is also dismissive of some of their values and self-righteousness. "The Language Police" would have ascribed to them more power than Pinker wants to acknowledge. "The Language Gurus" wouldn't have been bad, but perhaps overpraises them slightly. No, "maven" is almost perfect.
What I like about this title is that it instantly gives us a name for a phenomenon, and even an attitude (albeit, uninformed --- attitude without knowledge) to take toward that phenomenon. Beautiful!
Is the sentence striking enough to make the reader want to read more?
Does the sentence start as late as possible?
When in trouble, quickly brainstorm ten or more opening sentences: On occasion, one comes to a piece of writing with the perfect opening sentence. Take the good fortune and run!
When one does not have that sentence, things become difficult. It's tempting to compromise, and to accept a mediocre opening sentence. This is a bad mistake. The opening sentence is the reader's introduction to your writing. A mediocre opening is a statement that you're not going to make good use of their time. It's also a turn-off for you as an author, making you less excited about your project. By contrast, a good opening sentence establishes trust with the reader, and helps excite you about your own work.
So what to do when you're having trouble finding a good opening sentence?
In my opinion: just start brainstorming. Quickly generate 10 or more opening sentences. Don't worry about whether they're good or not. Instead, try to connect with many different ways of thinking about the subject at hand. Then, once you've got a stock of sentences, start to think analytically about the sentences. What makes each one good or bad? Can you improve any of the sentences? Do they suggest more ideas for sentences?
This is a time-intensive process. It can feel like wasted time -- you might end up spending an hour (or even more!) on this. But it's not wasted time. The opening sentence is the beginning of your relationship with the reader, and with your own piece of work. On both counts you want things to start off on the right foot.
Many of these comments apply also to the title, subtitle, and to the opening paragraph. Note that a great opening paragraph can, to a considerable extent, compensate for a less-than-great opening sentence. Of course, it's better to have both.
Opening sentence: "Fahrenheit 451" (Bradbury)
It was a pleasure to burn.
A beautiful evocation of feeling. It has both internal content -- the feeling a specific person gets -- as well as a dramatic visual.
Opening sentence: "How Buildings Learn" (Brand)
Year after year, the cultural elite of San Francisco is treated to the sight of its pre-eminent ladies, resplendently gowned, lined up in public waiting to pee.
A terrific opening. The observation is striking, true, interesting, funny, and, it turns out, a good introduction to the topic at hand.
I especially like the humorous contrast. Brand starts with lofty language: "cultural elite", "resplendently gowned", etc. But then we have the crass "waiting to pee". If you toned down the lofty language, or toned up the crass language, the sentence would not be nearly as funny. Furthermore, while the observation would remain true, it would be a less interesting truth.
Opening sentence: "The Selfish Gene" (Dawkins)
Intelligent life on a planet comes of age when it first works out the reason for its own existence
It's a provocative assertion. Possibly wrong, certainly contentious. But interesting, in a let's-stay-up-talking-to-3am kind of way, a kicker of a conversation starter. The phrase "on a planet" could arguably be omitted. It might be better to say something like: "An intelligent species comes of age when...".
Opening sentence: "Neuromancer" (Gibson)
The sky above the port was the color of television, tuned to a dead channel.
Gibson is drawing our attention to a place. And what a wonderful evocation of place it is! We can see half the scene. I worry that it's a bit too clever: prose drawing attention to itself. That can perhaps only be determined by what follows.
"Dead channel", incidentally, evokes the macabre. It says subtly that "this is not a good place". So the sentence works both literally, and by indirect meaning.
Opening sentence: "How to do what you love" (Graham)
To do something well you have to like it.
A simple assertion about reality. It has power because: (a) if true, it has profound consequences for how we live; (b) it's almost but not quite true (creating tension); and especially (c) many of us are in the uncomfortable position of both believing it's largely correct, but not always acting on it.
The result is that there's a lot immediately on the line, and considerable tension in the air. Graham has started with the kind of near-truism that we should think about and act upon, but often don't, because we're made uncomfortable by the gap between the truth of the assertion and the reality of our lives.
Opening sentence: "What Technology Wants" (Kelly)
For most of my life I owned very little.
Probably a surprise for most readers. Kelly is the author of books on technology -- not usually the kind of person we think of as owning little. So the surprise invites us in. Also, many of us have a complicated relationship to what we own, so it promises to provide an unusual perspective on a topic of broad interest.
Opening sentence: "A Brief History of Time" (Hawking)
A well-known scientist (some say it was Bertrand Russell) once gave a public lecture on astronomy.
This is scene-setting, an introduction to a story. It has the advantage that the tone is light. The "some say it was Bertrand Russell" even makes the point that it is determinedly light and informal!
Opening sentence: "The Language Instinct" (Pinker)
As you are reading these words, you are taking part in one of the wonders of the natural world.
A beautiful observation, reminding us of the extraordinary in everyday life.
Opening sentence: "Cosmos" (Sagan)
In ancient times, in everyday speech and custom, the most mundane happenings were connected with the grandest cosmic events.
This sentence describes a striking phenomenon. But it loses something because it leaves vague exactly what is being connected to what.
Opening sentence: "A Planet of Viruses" (Zimmer)
Fifty miles southeast of the Mexican city of Chihuahua is a dry, bare mountain range called Sierra de Naica.
Interesting: A bad tic is to announce interestingness. This is a tic to which many academics and ex-academics (including myself) are susceptible:
"It's interesting to consider..."
"An interesting example..."
There are several problem with these common uses of "interesting". First, you're instructing the reader how they should feel, when you should instead be making them feel that way by your interesting prose. Second, these uses are vague. What exactly is the interesting aspect of the subjects under discussion?
The first example above can be rewritten as "Consider...". "Interesting" can be elided completely from the second example.
The problem with "interesting" is when it's used because you lack confidence. It's announcing "hey, this next topic might sound boring, but I promise you, it's interesting". It subtly signals the opposite. It's telling the reader that the subject sounds boring. The solution is to make it interesting, right from the start. Often, this means deleting your sentence about how interesting the topic to come is, and instead just getting on with it. But make damn sure your discussion is interesting!
Quite: The second-worst of the adverbs. Much of what is said below about "very" applies also to "quite", with obvious changes.
Very: The worst of the adverbs. Many uses of "very" can be omitted. It is often used to intensify descriptions:
He was a very important person.
She looked very angry.
Jake would very much like to eat lunch.
Unfortunately, "very" is so generic that any such gains in intensity are often lost due to the dilution caused by using an extra word. It's rarely worth the tradeoff. The best that can be said is that if used very sparingly, the rareness of the occurrence may serve as a sign to the reader that the writer means business.
In the first two sentences above the word "very" can be omitted with little loss. If "very" is essential to the sentence, then its work should be done by the surrounding context. Show us his great importance; show her great anger. Merely using "very" will not convince us!
The third sentence is more complex. Omission is unsatisfactory:
Jake would like to eat lunch.
Something is lost, since the intensity in "very much" is the point of the original sentence. Unfortunately, "very much" remains weak because of its generic nature. It's better to replace the phrase with something specific:
Jake was ravenous for lunch.
There are times when "very" can be used well. Here's an acceptable use, in dialogue:
"Was she angry?"
In this example "very" is being used with a specific purpose, not as a generic intensifer. It works well because it makes its point with such economy. In general, "very" is okay when being used for a specific purpose, not as a generic intensifier.
Is scientific publishing about to be disrupted: This is an essay I wrote in 2009. In outline, the opening four paragraphs of this essay do the following: (1) tells the story of Kongo Gumi, a 1400 year old company that had gone bankrupt shortly before I wrote the essay; (2) poses the question: "Why do big companies fail?", with several examples; (3) recaps the standard explanations: stupidity or malevolence; (4) explains what my essay does: proposes a different (and scarier) explanation, and explains how it relates to scientific publishing.
Abstracting further, this structure tells a micro-story that leads one to ask a big, broad question. It then explains the standard answer to that question, and that the point of the essay is to offer an alternate answer, and to illustrate it by an application. It's likely a good opening structure whenever you're dealing with a big question with a standard answer and your point is to explain an alternate answer.
On Classic Style
Notes on classic writing style, as explained in Francis-Noel Thomas and Mark Turner's book Clear and Simple as the Truth.
Beware classic style if you've not yet mastered a subject: Sometimes, you wish to write about a subject, even though you haven't yet mastered that subject. It is difficult to write on such a subject in classic style, because the assumptions of classic style very nearly imply mastery. It is easier to instead slip occasionally into a personal style, explaining your exploration of the subject, and the limits of your knowledge. In such cases it may work to use a hybrid style, in which you alternate the personal with the classic. Feynman occasionally uses this style, either to explain his own thinking, or to explain the hypothetical thinking of others.
The war against the conventional story
Martin Amis has written a beautifully titled book, "The War Against Cliche". In that book he shows what can be achieved if one avoids the cliche, and instead finds both striking new ways of expressing ideas and --- a process that is hard to distinguish --- also striking ways of expressing new ideas.
Finding such expressions is difficult work. Should we undertake that work in technical writing, or should we leave it to literary authors? To put it another way, when should we work hard to find striking expressions, and when should we aim mainly for clarity?
Before answering those questions, let me introduce a slightly different idea, the idea of cliche in explanation. It's often easy to reach for the easy or conventional explanation of something, to aspire toward clarity as the main goal. This is especially true when there is a conventional story about some subject or question.
It's true that clarity is important. But for your writing to matter, to actually be saying something new, you must go beyond clarity. You want to find really striking and original explanations, explanations that reach beyond previous discussions and go somewhere new.
I believe it is worthy to engage in a war against the conventional in writing. Even when --- especially when --- discussing the most humdrum questions I should challenge myself to produce something striking and original. I should be aware of the existence of a conventional story (or stories) in part so I can avoid that story. While ultimately I will want to absorb and incorporate the wisdom in the conventional stories, that wisdom should not form the spine of my writing. Instead I should ask early whether there is some unusual perspective I can take? Is there some unusual insight that can be used? As a byproduct, such perspectives and insights are especially likely to produce a fortuitous turn of phrase.
To return to the original question, the goal therefore should not be the literary goal of avoiding cliche in favour of striking prose. Rather, it should be to find new perspectives on the underlying questions and phenomena. Striking prose will merely be a happy byproduct.
F. Scott Fitzgerald (in Gatsby): "There was dancing now on the canvas in the garden; old men pushing young girls backwards in eternal graceless circles [...]"
In three words, Fitzgerald conveys many things:
(1) "eternal" refers to the obvious fact that the dance can continue as long as the music holds out, but simultaneously conveys a sense of all dances continuing in this way, for all time, and perhaps even to a sense that there is just one dance, continuing in circles forever.
(2) "graceless" is descriptive, but we are no sure whether it is cynical, or realistic. No matter which, it is presumably a parochial comment, since not all dancers are graceless.
(3) This is all possible because Fitzgerald starts with a very primary, raw observaton about what is going on: people are moving in circles. It's the rawness and simpicity and concreteness of that observation that enables the beautiful modifiers.
Let me attempt something similar. Suppose I wanted to describe the passage of cars along my street. Some simple observations: the cars move in lines; they stop and start; they move in different ways at different times of the day; they move in opposite directions; they for the most part act as barriers to one another. An interesting kind of barrier: a barrier to intention, since sometimes one will clearly wish to pass another, but it is difficult to do. The cars sometimes seem sad to me, especially when people are commuting to work. I can't quite say how, but there is somehow a dejected quality I associate to the cars' motion. Candidate phrases:
"Capacious machines moving in opposing lines". There's something wrong with it --- I would never use "capacious machines" --- but it still gets at the sense of raw observation.
"Large machines moving in opposing lines, sometimes halting, but moving often enough with speed to please their inhabitants".
Neither attempt has elegance. I should continue until I succeed.
Advertising billboards: "Beautiful people looking down upon us with eternal smiles". Not bad. What other quality do those smiles have? The obvious cliche is "frozen". Better is "artificial". Better still is "hired": "Beautiful people looking down upon us with their hired smiles" (or perhaps "smiles-for-hire"). As actors, their smile is a professional asset. Another cliche: "appealing" --- that's the intent. Another cliche: "forced". Better is "slightly manic". Can I do better than "Beautiful people"? "Two-dimensional people looking down upon us with [slightly?] manic smiles" is certainly better thban "Beautiful people looking down upon us with frozen smiles".
Note that in Fitzgerald's example what is happening is that we are getting a fresh look at a phenomenon which would otherwise be familiar. There are two ways of looking at this: (1) Purely as technique, for giving prose life; and (2) As a way of seeing anew, and more deeply, something which has intrinsic purpose. I believe (2) should be held primary; to the extent this is a technique, it is one that should be used to reveal only that which one believes to be of significance, not the breath life into the dull or unimportant. Fitzgerald saw the dance in his mind's eye, and was convinced that it mattered, and he wanted to convey it truly. So he looked at the dance closely, and described what he saw, as well as he could. They truly were dancing in "eternal graceless circles".