Why Bad Writing is a Sign of Bad Thinking

George Orwell wrote the most important sentence about writing in 1946. Not a craft tip. Not advice about style. A warning: "The slovenliness of our language makes it easier for us to have foolish thoughts."

Read that again. He is not saying bad thinking causes bad writing, though it does. He is saying bad writing causes bad thinking. The relationship runs both ways. Vague language does not just fail to communicate your ideas. It prevents you from having clear ideas in the first place.

Eighty years later, we have a technology that generates infinite vague language on demand. And the consequences Orwell predicted are playing out at scale.

The Orwell Argument

In "Politics and the English Language", Orwell describes how dying metaphors, verbal false limbs, and pretentious diction let writers avoid committing to a specific claim. You can write a paragraph that sounds like it says something without actually saying anything. The language does the thinking for you. Or rather, it does the not-thinking for you.

"Political language is designed to make lies sound truthful and murder respectable, and to give an appearance of solidity to pure wind." The mechanism is simple. Use abstract enough language and you never have to confront what you actually mean. "Pacification" instead of "bombing villages." "Restructuring" instead of "firing people." The euphemism protects the writer from their own meaning.

This is not just a political problem. It happens in every email, every memo, every report where someone uses phrases like "leverage synergies" or "drive alignment" or "facilitate outcomes." These phrases feel like communication. They are the opposite. They are padding shaped like thought.

Writing as Thinking at McKinsey

McKinsey & Company figured this out decades ago. At the firm, partners require associates to write out their analysis before presenting it. Not as a formality. As a thinking tool.

The logic is brutal in its simplicity. A slide deck lets you hide behind bullet points. A verbal presentation lets you handwave past gaps. But a written document forces you to connect every claim to evidence, every conclusion to the analysis that supports it. You cannot write "the market is attractive" without immediately confronting the question: attractive how? Compared to what? Based on what data?

Barbara Minto, who developed McKinsey's writing methodology, called it the Pyramid Principle. Every written argument must have a single governing thought at the top, supported by logically grouped sub-arguments, each supported by evidence. If you cannot structure your writing this way, you do not understand your own argument.

This is not a McKinsey quirk. It is a universal truth about writing. The act of putting words on a page forces you to confront gaps in your logic that you can gloss over in conversation or in your head. Writing is where fuzzy thinking goes to die.

Writing as Discovery

Verlyn Klinkenborg takes this further. In Several Short Sentences About Writing, he argues that writing is not the expression of thinking. Writing is the thinking itself.

"The piece you're writing is about what you find in the piece you're writing." You do not know what you believe until you write it down. You do not know what your argument is until you build it sentence by sentence. Outlining in advance assumes you already know what you want to say. But the entire point of writing is to discover what you want to say.

Every writer has had this experience. You sit down to write about one thing and end up writing about something else. The sentence you just finished changes the sentence that comes next. Your thinking evolves as you write because the writing is the thinking.

Jasmine Sun puts it this way: "When you become a writer, here is one of the first things you will discover: with about a month of dedicated effort, you can go from knowing nothing about a topic to being able to present fluently." The writing forces the understanding. You research because you need to write. You clarify because the sentence demands it. Remove the writing and you remove the forcing function.

The Simplicity Test

There is a saying often attributed to Einstein or Feynman: "If you can't explain it simply, you don't understand it well enough." Whoever said it first, the observation is correct.

Clear writing requires deep understanding. You cannot write a simple sentence about a complex topic unless you understand the topic well enough to identify what matters and what does not. Jargon is a shortcut that lets you skip this step. Passive voice lets you avoid saying who did what. Long words let you sound sophisticated without being specific. Every one of these is a symptom of incomplete understanding.

Richard Feynman was famous for explaining quantum mechanics to first-year undergraduates. Not because he dumbed it down. Because he understood it so thoroughly that he could find the simple version. The simple version is harder to write. It requires more understanding, not less.

Bad writing is often defended as "complex ideas require complex language." This is almost always wrong. Complex ideas expressed in complex language usually mean the writer has not done the work to understand what they are saying. They are hiding behind the complexity instead of cutting through it.

How AI Breaks the Loop

Here is where it gets urgent. AI lets you generate text without thinking. You type a prompt. The model produces fluent, grammatically correct, structurally sound prose. It reads well. It says nothing specific.

The writing-thinking loop is broken. Normally, writing forces thinking because you have to choose words, build sentences, connect claims. Each choice is a tiny act of understanding. When AI makes those choices for you, the understanding never happens. You get the output of thinking without the thinking.

A joint study by MIT, Wellesley, and Massachusetts College of Art measured brain activity in writers using AI. Writers who used AI freely from the start showed lower brain activity than writers who wrote without AI. The AI was doing the cognitive work. The writer's brain was coasting.

This is Orwell's nightmare realized. A machine that constructs your sentences for you and thinks your thoughts for you. Except the machine does not think. It predicts tokens. The thoughts never happen at all.

When you use AI to draft an email, you skip the step where you figure out what you actually want to say. When you use AI to write a memo, you skip the step where you confront the gaps in your analysis. When you use AI to write a blog post, you skip the step where you discover what you believe. The text exists. The thinking does not.

The Fluency Trap

AI text is dangerous precisely because it sounds good. It is grammatically correct. It uses transitions. It has structure. It reads like someone who knows what they are talking about. But nobody is talking. Nobody knows anything. The fluency is a disguise.

In the old world, bad thinking produced bad writing, and bad writing was easy to spot. The person who did not understand their argument wrote a confusing memo. The confusion was a signal. A useful signal. It told you the thinking was not done.

AI removes that signal. Now you can produce polished prose with zero understanding. The memo reads well. The analysis sounds rigorous. But no one did the analysis. No one confronted the hard questions. The writing looks like thinking happened, but it did not.

This is worse than bad writing. Bad writing at least tells you something is wrong. AI writing tells you nothing. It is a perfect mask.

The Fix

The solution is not to avoid AI. It is to use it in the right order.

Write your first draft yourself. This is the thinking step. The draft does not have to be good. It has to be yours. Every word you choose, every sentence you build, every connection you make between ideas - that is thinking happening. That is you confronting what you know and do not know. You cannot outsource this and call it writing.

Then use AI to edit. Once you have done the thinking, let AI help with the expression. Ask it to tighten a paragraph. Find a better word. Cut unnecessary sentences. Restructure for clarity. These are editorial tasks. The thinking is already done. The AI is polishing, not replacing.

The MIT study confirms this works. Writers who drafted first and then used AI showed increased brain activity. Higher than the no-AI group. Evaluating AI suggestions is cognitively demanding. Accepting some, rejecting others, deciding what stays and what goes. That is judgment. Judgment is the highest-order thinking skill.

AI as editor preserves the thinking. AI as ghostwriter removes it. The distinction matters more than any other choice you make about how to use AI in your writing.

Writing Is the Work

The temptation of AI is that it makes writing easy. But writing is not supposed to be easy. The difficulty is the point. The struggle to find the right word is how you figure out what you mean. The effort to connect two ideas in a sentence is how you discover whether they are actually connected. The frustration of a paragraph that will not come together is your brain telling you the thinking is not done.

Skip the difficulty and you skip the benefit. You get words on a page but not thoughts in your head. You get a document that looks finished but represents nothing. You get communication without comprehension.

Orwell, Klinkenborg, the McKinsey partners, the MIT researchers - they all point to the same truth. Writing is thinking. Not a record of thinking. Not a byproduct of thinking. The thinking itself.

Bad writing is a sign of bad thinking because writing is where thinking happens. Fix your writing and you fix your thinking. Let a machine write for you and you stop thinking altogether.

Athens is built around this principle. Write first. Think through your words. Then use AI to sharpen what you wrote - with inline diffs that show every change, so you stay in control of every decision. The AI proposes. You decide. The thinking stays yours.