I think the fact that many people were willing to try switching to mongodb despite the fact that it doesn’t use sql and had quite poor semantics should suggest

I think the fact that many people were willing to try switching to mongodb despite the fact that it doesn’t use sql and had quite poor semantics should suggest that there would be some way for a better relational database query language (and engine) to catch on. On the other hand, the fact that many new time series databases (and other engines, including Materialize which the author of the article worked on) wanted sql (or fake sql-like languages when they couldn’t manage sql) suggests that not being sql can be a big hindrance.