Typos are the new Trust Signal
I use tools like Claude Code, Codex, or other AI tools for my workflows to help me speed up certain tasks or gain a new perspective on a project that I am working on.
But writing is different.
Writing is the most direct window I have into how someone thinks about a certain topic. When I read something, I’m not just consuming information - I’m watching someone wrestle an idea into words and sentences. The work of taking the chaos in your head and forcing it into meaningful sentences is where the value for me lives. Even if I might not agree with you at all. Once you outsource that to an LLM or any other tool that automates this process for you, I’m not sure what’s left. Why should I bother reading something someone else couldn’t be bothered to write?
Broken is Now Better
As a reader, typos and grammatical errors sometimes were a negative signal. Careless writing meant someone didn’t care enough to fix the last mistakes (sometimes you just miss a typo here and there, but you get my point, hopefully).
Funnily enough, that’s completely flipped for me. The less polished and “artificially” well-written something seems, the more attention I give it. Imperfect writing somewhat has become a credibility marker - a sign that a person wrote it, not a pipeline made up of a bullet list and a model publishing it. To throw in another random thought is that the point is and always has been the process. And it doesn’t matter whether we’re talking about writing, code, or anything else that is a craftsmanship in any way.
The irony is that this won’t hold for long either. A model can fake broken sentences or some weird wording just as easily as polished perfect sentences. The dead internet theory gets harder to dismiss every month.
A Small Hope
Till now it was some sort of a random internet rant, but I won’t just finish the post here without providing something that might be useful to you. Introducing the Kagi Small Web, which is a project made by Kagi to humanize the web with content made from real people who care about what they post to the web. In general, the rules are pretty simple. You can submit your online resource only with two other sites that you don’t own. Auto-generated content is forbidden and the blog must have recent posts.
You can try it out for yourself if you want: https://kagi.com/smallweb
Yes, you will see stuff you are not interested in. But every now and then, there is a small blog which sparks your interest or encourages you to dig in deeper.
Sources
Dead Internet Theory - https://en.wikipedia.org/wiki/Dead_Internet_theory
Kagi Small Web GitHub - https://github.com/kagisearch/smallweb
Kagi Website - https://kagi.com/