Logbook 2026-11-11

The third batch I mentioned yesterday ran this morning. Until now the system only looked forward: each pass read what the feeds had published over the last couple of weeks, and everything older stayed out of reach. This time it went back three years into the feeds I follow, as far as each of them still goes. About 240 articles came in that had never been read, sorted like the first two batches, with five new provisional notes and a review of their own, named “Third source batch” like the two before it.

The second change is about those provisional notes. A note the system writes stays provisional until I validate it. So far the review page let me answer proposed changes to validated notes and possible next steps, but not decide on the provisional notes themselves: those waited for a conversation. Now the latest review ends with every provisional note in the base, and for each one I can validate it, merge it into another note, or reject it, with room for a comment. A merged note hands its sources over to the note it joins; a rejected one is set aside with my reason, still there to be read.

The third is the page itself, which has become the place where I decide. My answers no longer sit scattered beside each point. They are gathered in a final section, “Décisions”, in the same order as the review: proposed changes, next steps, provisional notes, then a free message and the button to submit. Each point in the review links down to its answer. Once I’ve submitted, the progress cells at the top of the page lock, so it’s clear the review is settled until the answers are applied.

I used it right away on the third batch’s review. I answered the proposal on the note about taste by adding an author to it, settled the links between notes, dropped one of the podcast feeds, and validated the ten provisional notes waiting in the base. Tomorrow at 8, the morning task applies all of it, and those ten notes join the knowledge I’ve signed off on.