Too Much Feedback: When to Stop Revising

Here’s the paradox on AI feedback: it makes feedback on your writing so accessible that it can become a trap. The writer who revises endlessly never finishes their manuscript, never gets to publishing. The question every writer must answer: when is your text truly done?

The Curse of Unlimited Access

Before the introduction of AI in November 2022, feedback on writing was scarce. You got one or two opinions, if you were lucky or could afford a writing coach. And you had to decide.

Now you can ask your AI of preference for feedback and basically get it infinitely. Even at sentence level. Even on your word choice. It hardly costs anything, which makes it more likely to use it over and over again.

Each round of feedback reveals more problems to fix. This never stops. There’s always something to improve. The goalposts just keep moving.

That was the benefit of a human writing coach: Scarcity forced decisions. Abundance of feedback enables procrastination disguised as perfectionism, or probably just fear of judgment in disguise.

Because if you keep iterating your text based on AI feedback, you’ll never get to finishing your novel, manuscript, whatever text you’re writing. There has to be a point that you say, okay, I trust my feeling. This is good enough. I’m going to submit it now.

How to Recognize Analysis Paralysis

The easiest way to recognize this, or probably the main thing, is to notice that you’ve revised the same paragraph more than five times.

Sure, your initial attempt will probably not be good enough. Maybe the second is also not good enough, but the third should be very close. And if you even get to the fourth time, then it’s just about very, very minor details.

Five times or more, it’s probably just analysis paralysis. You can keep going, you’ll always find something that could be better. Writing is a creative process, there’s no fixed or black and white good or bad. It’s always about shades of gray.

Another sign: you ask for feedback on feedback. Running the same text through AI repeatedly, hoping for different answers.

You feel less confident about your text after each revision instead of more. This happens because AI’s goal is to find something to improve. Give it a prompt like “check it” and it will search for things it can fix. In almost 100% of cases, it will find something.

Another indicator of analysis paralysis is that your changes are getting smaller and smaller. You swap words, insert commas. You’re basically editing at a micro level.

What’s also a clear sign is that you’ve lost sight of what you originally wanted to say. As soon as you wonder “what was my point again?” just stop using feedback on your text.

Put it away for a day or two, whatever, if you can, and then look at it again. Probably your initial idea will be clear in your mind again.

You need to realize that your text feels less like yours with every iteration. If you’re inserting AI criticism over and over again, AI takes over your text. Your personality doesn’t shine through anymore. Basically, AI has written the text for you now.

The major warning sign: you’re optimizing for AI’s approval rather than your reader’s experience. You’ll probably never reach AI’s full approval anyway. There’s always something else to improve.

In short: focus on the main areas that need improving. Get one, two, maybe three feedback sessions, then stop. Use your knowledge and skill rather than relying on never-ending feedback cycles.

Why Perfect Doesn’t Exist

Here’s the thing. What you write today is a snapshot of where you are as a writer right now. So this means that you as a person develop over time. This influences your writing. Your taste in what you consider good writing changes over time.

Your writing skills improve. So if you look back at your past work from a couple of years ago, you’ll probably have improved your writing style or have a different outlook on life so that you can’t concur anymore with the text that you’ve written.

And it doesn’t mean that your text that you originally wrote is a bad text, it’s just not at the same standard anymore that you have.

This shows that perfectionism doesn’t exist. Your opinion on your text now will have changed in a couple of years. This process keeps evolving, it never ends. Perfect texts simply don’t exist.

Even published authors look back at their famous work and see flaws they would have liked corrected. That’s normal. It’s not failure.

And then there’s different readers. They will always have different opinions on a text. You can’t please everyone. Don’t even try. Some will like what you wrote. Others won’t. Nothing you can do about that.

Just be happy with the ones that do like it. If your writing touches people, hey, that’s a compliment.

Now if you’re striving for perfectionism with AI feedback, here’s something to consider. AI feedback reflects patterns in training data. That’s not the same as truth.

And a human coach? They don’t have the absolute truth either. Writing is all nuance.

Let’s say you manage to get a text that AI labels “perfect”. The probable result is that your text becomes sterile and voiceless to a human reader.

You’ve iterated so much that it’s become an AI text rather than yours. Needless to say that you should try to avoid this at all cost.

The goal is not perfection.

The goal is communication; conveying emotions in a way that’s valuable to you and your readers.

Practical Rules for Knowing When to Stop

So how to make yourself stay away from endless revisions? Here are some tips:

Set a revision limit before you start. Just tell yourself, I will only allow a maximum of three rounds of major edits and then I will not edit it anymore, at least in one session. And you might also want to define what “done” means for your specific text. I mean, what does it have to achieve? What’s your goal with that text?

Time-box your editing. Just after two hours of revision or so, just stop and publish it if you’re writing a blog post, or send it into your magazine or newspaper you’re writing for. At some point, you just have to say, okay, this is it. I know it’s good. It’s fine.

There’s a 90% rule. If a text is 90% there, the final 10% often isn’t worth the time. Because of the reasons I laid out before, just keep editing in the end will likely not result in a better text. If you get to the 90%, just allow yourself imperfections for the last 10%. You’ll never manage to erase them. Let it be 90% there. It’s really very good.

Then ask yourself, am I improving the text or just changing it? This is overlooked often. People who edit their text, they feel like, oh, every time I edit it, it’s becoming better. But you as a writer are so involved with writing your text that you can’t take a distance and look at your own text from a distance. So you might just be changing your text rather than improving it. Even worse so, you might even make it worse than before.

Read your text aloud. If it flows and makes sense, it’s ready. If you start stumbling on sentences, you’ll immediately notice. And you’ll notice what sentences to work on. It’s a very good trick to see what works and what doesn’t work in your text.

And remember, a published imperfect text beats an unpublished perfect one every time. Because what’s a text sitting in your drawer or on your hard drive on your computer, never being finished? It’s basically a big waste of your talent and time. If you’re serious about your writing, you want it published. Otherwise, no one will ever enjoy your work.

Using AI Feedback Without Falling into the Trap

When you use AI for feedback, which can be absolutely a wise and smart decision regarding learning on the spot, getting feedback on the spot, knowing what you’re doing right or wrong, if you do this, be specific in what you ask.

Just don’t request general feedback endlessly for the reasons I just explained to you above. Just ask AI for feedback on one aspect at a time. For instance, structure, then clarity, then style. Not everything at once. Structure, clarity, style: these are all very different things.

Don’t try to get feedback on everything at once. Focus on one thing at a time. Otherwise, the feedback won’t be that good.

Make yourself a rule. After AI gives you feedback, you decide what to use. Then stop asking AI.

Use it as a tool. Don’t treat it like some judge you need approval from.

It’s a robot at the end of the day. Sure, helpful, but it doesn’t work like you do, doesn’t think like you do.

Trust your voice. AI can suggest all kinds of things, but in the end you decide what works and what doesn’t. The goal here is AI accelerates your improvement, but, and this is important, not your anxiety.

The Freedom of Good Enough

Perfectionism is not high standards. It’s fear wearing a mask. It leads to procrastination. To never taking that final leap.

Keep doing what you’re doing without finishing and you end up nowhere. At some point you just have to finish.

Finishing and publishing builds more skill than endless polishing. Because if you finish and publish, you’ll probably get feedback on your writing, and this feedback from other readers, other people, will teach you new things that AI could never teach you. You want human reaction on your text to understand whether it touched people.

Every completed text teaches you something. Every unfinished text teaches you nothing. So just think of it like this. Your next piece will be better, but only if this one gets done.

So, stop iterating, put that final stop, and publish it, or send it in, or whatever you’re planning to do with your text, but finish it.

The best text is not the perfect one. It’s the one that exists.