I think it is a bad idea to tell developers, “if your package is good enough to be curated and you push an update we don’t like , then we reserve the right to usurp your package”. For example, let’s say I write a package that ends up in Curated, and I decide to change the license from MIT to GPLv3. Now, the curation committee wants to keep things MIT for shareable, so they fork my package, and remove mine from curated. Who is going to maintain the MIT version? Not me, I continue updating my GPL version, and now my code is GPL so you can’t keep the fork up-to-date with my new code. Now, the “curated” package has no dedicated maintainer and will eventually get stale. Meanwhile my package is still going strong in General, while a nearly identical copy is slowly eroding in Curated, which is the opposite of what was intended.
This is just one hypothetical to be sure, but there are any number of reasons that could fall under “disagree” so commitee control seems like a can of worms best left unopened. Unmaintained/abandoned is a completely different thing from disagreements with actively maintained packages. So, I don’t think a curated registry would have helped the current situation anyway, especially if Kaimon development continues outside the General Registry.
(to stray a bit from the OP) Another option to that doesn’t require a curated registry is just to give people more info on packages when searching for them. When I need to pick a new package the signs of a good one I look for are: initial documentation/readme, code examples, download popularity, and github stars. People generally flock to the best/most useful packages, so getting usage statistics like in Stefan’s new proposal AND including them in search results should be enough direction to give users an idea of what to install. I proposed this in another thread as well, but including a sortable search feature into Pkg.jl would go a long way towards giving more confidence about packages. At a minimum, more usage stats could be included in the JuliaHub package search to give an idea of the most widely used – and by extenstion – most widely trusted packages.