The Editorial Mindset

I've always appreciated Stephen King's advice on writing. So much so that I once bought a jacket with a tiger on the back because he suggested that it was a good way to build confidence as a writer¹ (see George Stark in The Dark Half). Another piece of advice from King is this handy rubric for editing: "cut 10%"... or in the words of his high-school era editor:

“Not bad, but PUFFY. You need to revise for length. Formula: 2nd Draft = 1st Draft — 10%. Good luck.”

This is solid advice, but as you might guess there is a lot more to editing than making things shorter.

Editing in Everyday Life

During the course of my career as a writer and technologist I have often found myself in the role of editor. Sometimes formally, like when I was the Publishing Editor of a bilingual newspaper in China, or when I was serving as the Technical Editor of Twitter's engineering blog. But often it's less formal. Someone asks me to read a draft that they are working on, or in the technical world, review a pull request. Or my kids ask me for help with their homework. Being an editor is more of a mindset than it is a job title, and that mindset is something that has been on my mind a lot recently.

In the modern world, we have an increasing need to adopt an editor's mindset. We are barraged with misinformation, sometimes maliciously, but often just accidentally. "Don't believe everything you read on the internet" is a modern mantra, but sometimes you have to apply that pragmatic disbelief to things like established news media outlets who are slinging poorly informed takes out on social media, almost as fast as they end up retracting them. Often it's just small conversations with your family about something they heard about that you know is wrong, and that you find yourself having to debunk. There's a constant need to maintain vigilance and to hone your skills of discernment. To not just know what information to trust, but how to trust it. This is also being an editor.

In the modern world of AI-generated content, we find ourselves uncomfortably in the editor's chair even just as casual consumers. The AI generates something plausible, but is it right? How do you trust the output? It's our job to do the due diligence to look up citations, read the original research, and fact-check claims. Even better if you can recreate an experiment or do some original research to test a hypothesis. The best way to trust a piece of information is by challenging it. Part of the process of being an editor involves becoming an arbiter of truth. And for better or worse, determining what information to trust is a skill that we are forced to practice in every aspect and every moment of our lives.

The Editor in the Age of AI

From our vantage point in 2025, we are seeing generative AI tools such as ChatGPT, Claude, and Gemini increasingly take on the task of writing, with humans shifting seats from the typewriter to the editor's desk. As editor to the AI, you are the guiding natural intelligence that shapes and refines the raw output into a finished work. The editorial mindset is the discipline that defines the quality, integrity, and impact of our collaboration with intelligent machines. Every software developer, marketer, analyst, and technologist suddenly finds themselves thrust into an editorial role, shaping AI-generated content for clarity, correctness, and perhaps most importantly, attempting to build trust with the reader.

So what makes for a good editor? Good judgement. Editorial judgment is rooted in the ability to evaluate writing as a work in progress. It's about building empathy for both the writer and for the reader. It's about recognizing the potential for the piece, and having faith in the writer that they can achieve that potential. A good editor sees a writer's weaknesses and devises strategies to either sidestep them or strengthen them. Editors seek to preserve the author's voice while also imagining how the piece might become clearer, more persuasive, and more aligned with its audience and with the writer's purpose. A skilled editor observes the piece through multiple lenses at once: the author's intent, the reader's expectations, and the broader conversation the writing participates in. The editorial mindset engages our interpretive skills, has sensitivity to tone, and brings a deep familiarity with context and how it shapes meaning. It's this layered awareness that transforms editing from a mechanical task into a creative and intellectual practice.

Editing as Interpretation: Theory and Practice

From a theoretical standpoint, editing draws on multiple academic traditions. Literary theory reminds us that all texts are situated. They are produced in context, interpreted in a different context, and are rich with implicit meaning. Reader-response theory teaches us that the meaning of a text is co-created by its audience. Editors must therefore consider not only what a text says, but how it will be read… and misread.

Hermeneutics, the theory of interpretation, offers tools for unpacking ambiguity, decoding reference, and preserving nuance. It doesn't do this by eliminating ambiguity, but rather by recognizing it as an integral feature of interpretation. Rather than seeking a single, final meaning, hermeneutical methods navigate multiple possible interpretations by considering the historical context, authorial intent, and the interpreter's own perspective. While hermeneutics is primarily a tool for understanding historical literary and religious texts, it is just as useful as a way of understanding how a new work will be received and interpreted in the culture of the now.

As Scott Noegel puts it in “Wordplay” in Ancient Near Eastern Texts:

"One of the fundamental principles informing all types of polysemy is ambiguity. It is important to emphasize that ambiguous signs, words, and lines do not leave a text impenetrable to understanding, and thus incapable of conveying meaning. Rather, they pack the text with interpretive options, contingencies, and points of view—they overload their contexts with meanings."

From these disciplines, we learn that editing is not merely a technical activity, but a deeply intellectual one.

Editorial judgment develops through close reading, a term for the slow, attentive, and often recursive encounter with a text, where you may read and re-read the same words over and over, diving deeper into your interpretation and understanding with each pass. This is a practice that resists automation, because it relies on human intuition, taste, and the lived experience of language, culture, history, and meaning.

An editor may ask themselves a variety of questions as they read the work:

• Does the voice maintain integrity throughout?
• Does the structure serve the thesis?
• Do transitions guide the reader's attention, or do they jolt and confuse?

A skilled editor will look for opportunities to restructure arguments, tighten the prose, and call out inconsistencies. In the age of AI, where content may be voluminous but undifferentiated, developing an editorial mindset will strengthen your role as a guardian of quality, defender of voice, and the clarifying conscience that is lacking within the machine mind.

This is all pretty highfalutin language though. Let's talk about some of the practical strategies that support this mindset.

Editorial Tactics

One foundational tactic is the "big picture first" approach. Rather than focusing immediately on grammar or word choice, which is more copy-editing than it is editing editing, begin by asking "What is the purpose of this piece?" or "What were they trying to accomplish here?" and "Who is this for?".

Once you have that big picture established, ask yourself "Is the structure sound?", "Are the ideas in the right order?", "Is the intended audience clear, and is the voice appropriate for them?". It can be helpful to visually map out the structure and flow of the narrative in a simple outline or flow chart. Similarly, make notes about the audiences it seems to be addressing, and the dominant aspects of voice. These notes can help you see when there are conflicts or inconsistencies.

Reading aloud is another enduring editorial technique that becomes very powerful, especially when working with AI-generated content. It is so elementally human to see the words as symbols, interpret them into concepts, and then speak them out in your own voice and at your own pace. It allows the editor to catch clumsy transitions, rhythm issues, or unintentional tone shifts that can slip past during silent reading. Reading aloud can expose the cognitive gaps between what we think the text says and what it actually communicates.

For technical content, there are some additional concerns. I like to use a review checklist similar to this one:

• Accuracy: Are the technical details correct?
• Completeness: Are there missing steps or assumptions?
• Usability: Can someone follow these instructions successfully?
• Coherence: Are the claims logically consistent and supported with sufficient evidence?

This structured approach helps ensure consistent quality across technical content. For example, if you're reviewing software documentation, but haven't actually run the code examples, then you must acknowledge that the docs are NOT ready to publish. Was running the code an intuitive experience? Is there room for improvement of the developer experience at the code or API level? This is part of the job of editing.

Speaking of code, another technique I often use is re-transcribing the code. Similar to reading aloud, this is an intentionally inefficient practice. Instead of copy/pasting the code, just look at it in the browser window, and re-type it into a code editor in a neighboring window… character for character, line for line. Similar to the way that taking handwritten notes can, somewhat ironically, help you remember things so well that you don't actually need the notes. Re-transcribing the code will help you see aspects that might have blown right past you when copy/pasting.

Giving Feedback and Building Rapport

Another common piece of writing wisdom:

"You should never be your own editor"

As authors, we vitally depend on the external perspective and voice of an editor to show us things we would have missed on our own. Editors view our work from a different vantage point, one that is less likely to treat the work as precious and is more open to change. They can see sides of it we are blind to from our vantage point. While you may be stuck inside of your context as a writer, a good editor will understand a piece of work in multiple contexts at once.

The skill of giving feedback cannot be overstated. A good editor knows how to give careful and precise feedback to help the author improve their writing. Once you've figured out where the piece can be improved, you now have to deliver that news in a way that will be well received by the author. With human authors, we have to think about their emotional condition and how they might react to feedback. It requires a certain amount of diplomacy and empathy.

However, whether you're giving feedback to a squishy emotional human, or a sycophantic unfeeling AI model, the same techniques apply. Make sure that your feedback is actionable, specific, and framed in terms of the goals of the piece. Good editors ask precise questions like "Is this section serving the argument?" or "Would this idea land better earlier in the piece?", rather than issuing vague pronouncements or imperious commands. The goal is for the writer to synthesize the feedback and understand the issues themselves. That said, sometimes a direct suggestion can be very helpful, but don't be surprised when someone hears your suggestion, rejects it and then does something entirely different and wholly unexpected with that feedback. The back and forth between editor and writer (or AI chatbot) can be exhilarating and at times frustrating. But isn't that always the case when two minds meet in the earnest pursuit of a shared creative vision?

One place where the feedback process between humans diverges from the interaction with machine actors is in the area of memory. Over time, two people will learn each other, build a rapport, and remember their interactions. An editor will come to KNOW the writer, and a writer will KNOW the editor, and the quality of the interactions will deepen and become more nuanced. But with machine authors this doesn't happen as readily. ChatGPT will not remember what you talked about last time. Claude will not remember that you told it not to capitalize the term "newYork" to "New York", because it's a term in a piece of code that creates a new "York" object, not a reference to the iconic North American city.

Therefore, it is up to you as the editor to maintain that context. You must track the various iterations of prompts that you wrote and what they got right and what they got wrong. You must document your editorial decisions, and the language needed to guide the AI tooling to your desired outcomes. Most tools include a way to give that vital context, such as an AGENTS.md file in the root of a code repository or well-tested prompts offered up from a MCP server. This infrastructure creates a knowledge base that improves future AI interactions and helps teams learn from successful editorial patterns.

Editorial Ethics

Finally, consider your values. I've talked about judgement a lot in this article. To edit is to intervene, and intervention always involves judgment. But judgment is not value-neutral. It shapes meaning, directs opinion, and prioritizes certain perspectives. That will always have a socio-political impact. A responsible editor acknowledges this and remains attuned to the ethical dimensions of both their role and the content of the writing. This is all part of understanding a piece in context. Those simple questions from earlier like "What is the purpose of this?" and "Who is this for?" are themselves ambiguous and of multi-dimensional purpose, and they can and SHOULD be interpreted many ways.

For example, an article on how to use AI more effectively can have many different outcomes. Will it strengthen the rich and powerful who run the AI data centers, or will it democratize access to technology, delivering much needed power into the hands of the disenfranchised? Will it help artists create better art, or will it destroy their motivation to create? What impact will that have on the world? Further, will a certain metaphor or example be understood by a global population, or only by a specific subset of people who have the same cultural context as the author? Who is being left out by this? Is the language used in the article too complicated, preventing younger readers and second-language readers from easily understanding it? Could more simple terms be used to make it more accessible? Is that a goal or a non-goal?

One thing we can be sure of, as more and more content originates from AI, these questions of voice, authenticity, and bias are becoming more urgent, because AI doesn't ask those questions if you don't. The editorial mindset demands a willingness to confront these questions directly. The buck stops with you.

The Importance of Continuing to Write and Be Human

And now that I've spent more than 2,500 words talking about how to be an editor, I will leave you with a different call to action: don't stop being an author. When you put on your new hat that says "Editor in Chief" across the front, you are working a muscle that might not have gotten a lot of attention in the past. But over-focusing on one muscle can lead to atrophy in muscles that were previously much stronger. Please don't stop scrawling bad poetry in Moleskine notebooks. Please don't worry about whether or not your writing is GOOD, just worry that it's HONEST. Ensuring that the writing is good is the editor's job, but only the writer can ensure that it's honest. Also, I frequently have to remind myself that just because I am a good editor, doesn't mean that I should think like an editor when I'm writing. Instead, we need to let the words just flow out. Use voice-to-text to get past the fear of a blank page. Or, as another timeless and sage piece of writing advice implores us to do… "Write drunk, edit sober." Be human in the way that ONLY YOU can be. You are unique and your perspective matters, and cannot be replaced by AI. We are not all Hemingways, but we all have something distinct and irreplaceable inside of us that we can share with the world.

Consider occasionally letting ChatGPT wear the editor's hat instead. Ask it to ask those questions I listed above and to give you feedback about your writing. Or better yet, go find your local writers group (there definitely is one in your town, I promise), and meet another squishy awkward human and be vulnerable with them. Ask them to read your bad poetry and thank them for giving you their thoughts afterwards. If there is anything precious within the minds of humanity it is our ability to connect and share, and to be honest and gentle with each other. And that should be core to both your writing and your editing.