SQL has, by modern standards, a shitty API. The DSL is an ugly mess and it's extremely difficult to reason about performance, especially when joins are involved

SQL has, by modern standards, a shitty API. The DSL is an ugly mess and it's extremely difficult to reason about performance, especially when joins are involved. Ergo, some not-by-choice SQL users only scratch the surface and use common features: INSERT, UPDATE, DELETE, SELECT, also known as CRUD. Ergo, it becomes easy (although wrong) to conclude that SQL isn't doing very complicated work behind the scenes and that it's just another overcomplicated dinosaur POS like the Windows operating system, one that remains popular only because it's a standard. Thus, a lot of undue hate gets directed at SQL with little attention paid to the subtleties of what it does extremely well and where it behaves poorly. I think "SQL isn't scalable" is in the same league as "Java isn't concurrent". It can be, if you have learned the necessary skills and are willing to deal with a bit of pain. Is Clojure astronomically better for concurrency? Sure, but people can and do scale with SQL databases, and they do this in large part because there are problems for which SQL is the appropriate solution.