This article's intended audience is clearly the subset of ML engineers that have a practical rather than a theoretical/academic background in ML. I'd argue it definitely has a use as a practical guide to understanding an original approach to RL, which looks like it has good potential. You could fairly argue that the article could use more mathematical grounding to what's being explained, like AMS blog posts tend to be. However, consider how afraid of mathematics a lot of the CS crowd tends to unreasonably be. The pedagogy of the article is noteworthy, it helps the reader get a hold of the jargon and ideas of this burgeoning approach. It prepares them for further research, kind of like Quanta Magazine tries to do, but it allows itself more technicality, in line with the blog post format. It's not an easy task given how multidisciplinary meta-RL really is, and the author does a rather great job, IMO.