I disagree. I've been programming for over 20 years, and SQL has been one of the only constants used at every job. I'm surprised how many other programmers I run into who are don't know or are even scared of learning SQL. Which is a shame because data will almost always outlive the program(s).

I used to think this was true, back in the days when I used 'Group by' mindlessly and fiddled with the syntax dumbly until I got the right answer. Once you spend some time thinking about the concepts of it all you should do much better at remembering and if you totally forget the SQL syntax that doesn't really matter (Modulo the hideously annoying random differences between the databases themselves, but that's true of anything.)

I completely disagree. A good full-stack dev knows SQL pretty well: at least well enough to be able to work through the general questions in the document. Good SQL skills are essential, and the lack of good SQL developers I would attribute mostly to ORMs that abstract it away so mediocre developers can mostly ignore it (which makes the job a whole lot harder for people that actually DO know how a relational database works, and give it the respect it deserves)

I really don't believe that to be true. For our application(s) at work, the database is always the hardest part to scale and tehrefore also the biggest performance bottleneck, and the difference between good and bad SQL can be magnitudes of order of performance impact (on the query itself and locking up the DB for other queries).

Only in a "Every database has 2 TB of memory and 64 cores"-world is SQL (and database design) a negligible skill.

I'm sure data scientists and such probably use SQL at least somewhat regularly.

What I forget about SQL is the little inconsistencies between platforms. It's a standard with a dozen carve outs. Do I use TOP or LIMIT? What's the wildcard syntax?

Many data analysts use exclusively SQL and R/Python