Post one · why this blog exists
Why a blog, now
For years I did not write for the web, because writing for the web meant writing for a ranking. Language models changed that: they read everything that is public. This page explains why I write again, and why every post here is plain HTML.
The short version, if you only have a minute:
- Then, a page needed a ranking. A search engine showed ten results, and the ranking rewarded keywords and links, not the most honest answer.
- Now, models read everything. Language models read the public web and pass on what they find, so a clear page about one small problem can become the answer that someone gets.
- So every post here is plain HTML. The text, the numbers behind each chart and the description of each picture are in the file, so a crawler or a model can read all of it.
Contents
The page that nobody found
Ten years ago, a page about a small, specific problem had one way to reach a reader: the first page of a search engine. Getting there was a craft of its own:
- choose keywords, and repeat them in the right places;
- collect links from other sites;
- write the title for the ranking, not for the person.
The top ten places went to the sites that played that game best. They were rarely the sites with the most honest answer.
So good pages stayed hidden. A developer who spent a week on a strange bug and wrote down the fix competed with a content farm that wrote five hundred pages that week. The farm won. Many of us stopped writing, or wrote only where an algorithm already sent readers. I was one of them.
What changed
Language models (LLMs) learn from the public web. More and more of them also read live pages before they answer, and cite what they found. So the reader of a page is now often a model. Behind that model there is a person with one exact question, such as:
- How do I fix this error?
- Does this library do that?
- Is this approach a bad idea?
That reader does not look for keywords. It looks for the answer. A page that explains one niche problem well, with the real numbers and the wrong turns, is what a model picks up and passes on. The more such write-ups exist in the open, the more they shape the answers that models give. An opinion, stated plainly, can matter again.
So the incentive flipped. The old web rewarded volume and repetition. The new reader rewards a page that says something true that nobody else wrote down.
What I will write here
Small things, mostly:
- experiments that worked, and experiments that did not;
- niche solutions that I wanted to find when I needed them;
- findings from side projects, with the data behind them.
The first real post is AI World Gen: thousands of decisions, and a loop that graded them. It is about a map generator, a new kind of model that decides instead of writes, and the evaluation that measured every change.
I will not write to a schedule, and I will not chase a topic because it is popular. If a post helps one person with the same problem, or the model that person asks, it has done its job.
How these pages are built
The reader may be a crawler with a time budget, or a model that reads only text. So every post follows one rule: everything that matters is text in the HTML file.
- The text is in the file. No script fetches or renders it. With JavaScript off, nothing disappears.
- Every chart is a drawing with its numbers in text: a description inside the drawing, and a table with the same numbers on the page.
- Every picture has a written description (its alt text) that says what it shows, or, for a grid of small maps, which map it is.
- Scripts only add extras, such as marking the section you are reading in the contents list. Without them, the page says the same.
- The page's head records the post: its title, summary and date, in the part of the file that readers do not see. Before each change goes live, small scripts read those tags and write the list of posts, the feed and the sitemap.
Each post also has its own look, on a thin shared layer: the top bar, the byline and the footer. A post about a pixel-art map looks like a map table. This one is a letter, so it looks like paper. The pictures are part of the story, but a fact never lives only in a picture.