I think it’s great that the CSS Working Group does these. It’s like planting a flag in the ground saying this is what CSS looks like at this specific point in time. They do specifically say it’s not for us CSS authors though…
This document collects together into one definition all the specs that together form the current state of Cascading Style Sheets (CSS) as of 2020. The primary audience is CSS implementers, not CSS authors, as this definition includes modules by specification stability, not Web browser adoption rate.
Remember “CSS3”? That was the closest thing we had to a “snapshot” that was designed for CSS authors (and learners). Because CSS3 was so wildly successful, we saw a short round of enthusiasm for CSS4, me included. There is zero marketing panache on that snapshot page, which is exactly what CSS4 would need to succeed. Remember, HTML5 and friends (including CSS3) even had fancy logos!

If someone were to say to me “Chris, when CSS3 came around, I boned up on all that, but I haven’t kept up with CSS since, what should I learn?” I’d say “That’s a damn fine question, developer that has a normal healthy relationship with technology.” But honestly, I might struggle to answer cohesively.
I’d say: Uhm, CSS grid for sure. Custom properties. Clipping and Offset paths I suppose. prefers-reduced-motion
. I dunno. There are probably like 100 things, but there is no great single reference point to see them all together.
I’ll work on putting a list together. I don’t think I’ll have the gumption to call it CSS4, but at least I’ll be able to answer that question. Feel free to suggest ideas in the comments.
Obviously I would include there the CSS Houdini thing, too, if possible.
Not sure if I am allowed to post a link here however I would remind everyone the great speech of Una Kravets at JSConf EU 2019 (you can find it on YouTube)
The Webdev had this discussion many times in the past, where two camps emerged: the people wanting version numbers and the people thinking in modules.
Browser vendors check a proposal and may or may not bring it into their browser, or maybe at different times. For this reason alone a fixed version number makes no sense to me.
Also should the version number increase when a single, highly requested function makes it into a browser, like aspect ratio? If we say CSS right now is CSS 4, is it 4.2 then with aspect-ratio in? Or CSS5? Semver makes no sense for CSS, since we do not have patches or bugfixes, just releases.
As a CSS author, here’s a naive versioning scheme I’d like to see:
January 2021, declare a new “version” CSS2021 with all the (new) features that ALL modern browsers support, saying: These features are save to use!
I think the poor marketing of newer CSS features has really hurt the developer view of CSS. Leaving us with the current craze of style in JS :-(
HTML might be done at version 5, but CSS 3 still had room for improvements!
I typically use the “State of CSS” or “State of JS” apart from other self-curated sources to answer those questions. I’d start there for suggestions on what to include in your list.