Inside the operations of this site: the raw numbers from the first eight articles to today

Summary

Of the first eight pieces I put out together, only half made it to publication. The cause for most of the four that stalled was not the quality of the writing, but the preparation done before writing: checking whether the material was cleared for publication. So before raising my writing ability, I first built up this before-writing preparation and the mechanism that inspects the text after writing. Through that accumulation, the published articles grew from the first eight to sixteen, and the operations records grew from around twelve lines to fifty-seven. For those of you who subscribe, this note shows the inside of how I have run the site from those first eight pieces to today, in the raw numbers of the published records.

What I show on the paid side puts this growth into a form you can trace in numbers. First, I look at what stalled and where among the first eight. Next, through the number of findings across the six pieces written in the daily job, I follow how I routed the causes of those stalls into confirming the rule documents the writer reads before writing, the before-writing checklist, and the mechanism that hands recently frequent findings to the writer before writing. On top of that, I set out three things: how many rounds the loop of judging and fixing that inspects the text after writing settled into; the check that looks for verbatim overlap with the material by machine alone; and the number of records that decided whether a finding was adopted or set aside. I close with the number of articles now published and the items I will keep counting every time.

Who this is for and what you can take away

This is a note for those who, alone or in a small team, run a setup that writes articles with AI and carries them through to publication. In the raw numbers of this site’s own operations records, you can check how many articles, behind the number I did publish, stalled before publication, whether they stalled on writing quality or on confidentiality or on procedure, how the number of findings moves once the causes are routed into before-writing preparation and pre-publication inspection, and how many rounds the loop of judging and fixing usually settles into. From that you can take away a rough guide to which points of your own operation are worth counting so they pay off later. This note is based on the operations records of this very site, which I run on my own.