Stubbing: Strong Norm: Please hook up an AI for changelog generation at least, and start using the i
Strong Norm: Please hook up an AI for changelog generation at least, and start using the ignyr directive.
At the very least, use a Jev-class super cheap classifier to post category-based changelogs like “minor”, “major”, “typo” etc. Without this signal, rollups and feeds in other people’s readers will present your feed in a noisy, low-quality way. I’ve made a pace-layered feed the default in Studio, and currently it moves old items up by 1 for every change. Once we have richer changelog data, the swimming upstream can be smarter.
I’m going to make “ignyr” tagging automated and on by default in Studio.
I’m also going to work on a scheme for AI assisted auto-pinning too, which will take changelog data as input in deciding whether to pin a new version (or prompt you to). Pinning is going to be a particularly strong signal once we close that loop properly, since an important class of high-signal feed heuristics will be “pinned only.”
So the intent of Blygger is starting to sink in for me here. I had to run this by ChatGPT to make sure I understood it correctly, and ChatGPT confirmed that it understood it as I did: Robots will help behind-the-scense in publishing to help us humans make sense of the stuff coming out in the blyggerverse.
I think it's natural for us to see robot-aided-writing as something that deals with the content, but now I'm starting to see it. The robots will help organize what we read and how we manage the flow. I guess this has been obvious all along, looking back, but it can take a while for you to see a thing when it's new.
OK, this is going to be a new challenge for my nascent vibecoding skills, but let's go. ChatGPT is recommending I install a tiny version of Qwen on my Mac to start writing metadata to go out with my posts. Not a development I foresaw as I undertook Blynger, but LFG.