# One year later, I write for myself first

> The thinking matters more than the finished article.

- **Published:** 2026-07-29
- **Tags:** writing, ai
- **Canonical:** https://nbarwicki.com/blog/25-i-write-for-myself/

---

My blog turned one on June 10. I noticed in the middle of July.

I missed the anniversary by five weeks. The delay gave me time to think about what I got from a year of writing.

I benefit most. **Writing has become one of the main ways I figure out what I think.**

## An opinion is easy inside my head

I can carry an opinion for months and feel certain that it makes sense. My brain skips the missing steps because I already know where I want to end up.

_Then I try to explain it._

I have to define the vague words. I have to connect one claim to the next. I have to explain why my experience supports the conclusion. The gaps become harder to ignore once I see them on the page.

Sometimes I find a better argument. Sometimes I discover that I disagree with the idea I planned to defend. **Both outcomes help me.** I would rather change my mind while drafting than publish a polished defense of a weak opinion.

Writing also makes me stay with a topic. A passing thought can feel smart for ten seconds. A short article asks for more. I compare the thought with what I have seen at work and what I claimed in earlier posts. That effort helps me connect ideas I would have left scattered.

## I ask AI to argue with me

My current process starts with a raw dump of thoughts. I write down the opinion and the observations behind it. Then I ask AI to attack it.

> Find the hidden assumptions. Give me the strongest opposing view. Show me where I stretched a personal experience into a general rule. Point out the question I avoided because the answer might damage the whole argument.

I decide what I believe. AI can produce a shallow objection or miss important context. I still have to check the evidence and decide whether my argument survives.

That division matters. **I supply the opinion and remain responsible for it. AI supplies pressure.**

The pressure can strengthen the idea. It can also break it. I have dropped claims and changed conclusions because I could no longer defend the first version.

This article went through the same loop. I started with a broad claim that writing helps people learn. The challenge exposed the overreach: someone can write without learning anything, and polished prose can hide weak reasoning. **The claim I can defend is smaller and personal.** This way of writing helps me see the gaps in my own thinking.

I get more value from that than I would from asking AI to make my first thought sound clever.

## The published post is a record

Each article captures what I understood at a specific point. I can return six months later and see where I was right and which parts no longer match how I work.

That record has value even when few people read it. I have given an idea enough attention to challenge it and reach a decision. Publishing adds useful pressure because another person needs to follow the reasoning, but the work has already changed how I understand the topic.

After one year, I still enjoy sharing what I learn. _I enjoy finding out that my first take was incomplete even more._

Take one opinion that has lived in your head for too long and try to explain it in writing. You may strengthen it. You may lose it. **Either result leaves you with a clearer view than the one you started with.**