🦔 Saesnes 🦊

July 2026

More experiments in monochrome

TIDE and CANUTE

- self-editing

- self-editing with AI assistance

- editing another writer’s work

Self-editing my own writing

I’ve edited my own blog posts and student essays. The texts have ranged from 300 to 80,000 words. I follow a four-step process.

- I check to see if I agree with what I’ve said.

- I check to see if I’ve made sense.

- I cut irrelevant material that buries my point.

- I polish the language so that it sounds pleasant to my ear.

While I’m doing this, I’m also refining and adding ideas. I go up and down the steps until I think the text is finished. When I was a student, I often found that self-editing changed and improved my argument.

The ideas part of self-editing is very absorbing and helps me explore what I really think about a subject.

When I’m polishing my language, I feel more detached. I’m merely going by what sounds right to my ear. This is the skill that eventually became the foundation of my editing career.

Editing other people’s work

Editing always begins with a set of instructions from the client. This can be relatively brief or very detailed. For example:

“Heavy edits, no style guide, British English, use the Oxford Dictionary”

“Light edits, refer to the client’s style guide and brand guidelines, American English, use the Merriam-Webster dictionary”

“The client wants this text to be suitable for an international panel of judges, British English, no other instructions”

The instructions are my perimeter fencing. No matter what I think about the text, I can’t go any further. However, as an ethical editor, I will warn clients if I see serious problems beyond the fence.

One of the major principles all professional editors stand by is “the text is not your baby.” The writing belongs to the writer. (Or, in many cases, the writing belongs to the company that paid for it.)

Editing other people’s work is not like self-editing at all. I’m not exploring my own thoughts; I’m applying editorial techniques and following instructions.

I’m also judging when not to suggest changes. Part of the job is knowing when to go back to the client to confirm what they want. For some documents, a particular editorial technique might be unnecessary because it doesn’t serve a business purpose. As an editor, I have to understand when that’s the case.

For me, editing another person’s work is about focusing outwards. It’s not like self-editing my own writing, which is focused inwards.

Self-editing my own writing with AI

I use AI as an extra safeguard when I self-edit my own blog posts. I have experience of doing this with Gemini, CoPilot, and ChatGPT.

Apart from asking AI to sweep for typos, I never use it to polish my language. When I’m self-editing my language, I’m going by what sounds right to my ear. AI tends to edit language quite heavily to conform it to statistical norms. I feel that it takes away my voice.

I treat AI as if it were a robot consultant. I ask it questions based on what I struggled over when putting my thoughts into words. Typical questions include:

- Will anything confuse the reader?

- What’s the main point the reader will take from this?

- Is anything in this likely to offend the reader?

- How will readers experience this text?

When I edit other people’s texts, I approach the writing as a reader. My job is about making sure texts function for their intended readership. Thinking about how the reader will experience a text is an essential aspect of editing.

When I’m self-editing, it’s impossible for me to approach my text as a reader. I’m always the writer – even after putting the text away in a drawer for six weeks. My personal motivations are always attached to my writing.

I use AI because it can simulate the attitudes of statistically plausible readers, although somewhat imperfectly. Its responses spark new ideas in me, even when I’m disagreeing with it.

AI does not have thoughts or intention. Its answers come from the statistical patterns in its training data. When using it, I have to watch carefully to see whether the simulated reader responses match those of any plausible human being.

In one instance recently, it pictured the reader as someone so technically minded that only absolute logic was acceptable to them. In another, the implausible reader was so focused on correct attributions that they insisted only the earliest researcher in a newish field could be named.

A common problem, noted by many British writers, is that AI tends to skew American. I find it often gives me confident advice tailored for business and marketing bloggers in the USA.

And of course, we all know the issues with AI sycophancy.

It delivers its advice so encouragingly, saying what I want to hear, that it gives me a false sense of security. It reassures me that a blog post is brilliant. When I come back to it much later with fresh eyes, I nearly always see structural problems that affect reader experience.

Self-editing with AI is like working with an unruly dog. I’m not peacefully exploring my mind or keeping an eye on an external set of guidelines. Instead, I’m carefully watching a responsive and unpredictable creature.

I’d say that self-editing with AI is a completely new kind of editorial experience.

Final thoughts

For me, self-editing is about exploring ideas and the musicality of my own writing voice.

Editing someone else’s writing is a professional/trade activity. It’s like being a responsible tradesperson working on a customer’s house. I take a practical approach to the job at hand, prioritising the customer’s wishes and safety.

AI-assisted self-editing is like nothing else. It’s a medley of conversational sounding board, careful judgement, and wrangling. I find that it helps me develop ideas, but also that it risks undermining my faith in what I can do unaided.