I agree about sources, but it is dated March 2025, so I doubt whether the slop comment is accurate. Besides, the actual tattoos (well, some of them) were "beautiful" (scare quotes intended for cultural reasons).
"These are important questions because scientists are increasingly overwhelmed with the volume of new work posted on preprint servers and published in journals. As a result, traditional quality signals used for triaging papers, such as journal, conference venue, and institution, are becoming less reliable."
The diagnosis is right but the treat is dead wrong. Instead of silly scoring systems, please improve recommender systems for papers. In this space, also opt-in and query-based personalization would be okay.
I think it was a fairly good take, including w.r.t. lobbying, if you are into it. I liked especially one of the takeaways:
"As such, we need to ask not only what research can do for policy but also what policy can do for research: What kind of research do we need and what kind of evidence do we want to develop?"
I mean, for a while, I thought something like Substack (and not Fediverse) could disturb things a little, but I suppose it and many others have already been killed by slop. So, if you do verified identity management, which is good for certain purposes but perhaps not for others, I suppose you should also do decentralized trust management, and with an ability to delete nodes from a personal but federated trust chain. (And feel free to adopt the idea also for science; it would be very much needed.)
I forget things all the time, and many of them I regret afterwards (e.g., because they were some good ideas, papers, blog posts, code, whatever). Three things come to mind: old-fashioned README files within directories you are working on (whatever you are working on); having local git repositories and committing frequently (i.e., also beyond code); and, for some things, old-school pen and a notebook.