I'm not sure that applies to everything, though. I'm sure we agree there's a limit.
Joel Spolsky is right when he writes about the 'Law of Leaky Abstractions', but I'm not sure the average CRUD developer needs to know the gory nuts and bolts of his DBMS. Of course they need to know enough to reason effectively about good design, performance, security, etc, and more low-level knowledge is always a good thing.
As an extreme example, PHP programmers don't need to understand branch prediction and caches.
You are very correct for anyone who would consider themselves a programmer. I was thinking in a bigger scope; with so many non-programmers working technical roles (and especially with data/databases), I think this is a marvellous way for e.g. project managers to have a brief mental model of what a db index is.
I think 'Law of Leaky Abstractions' is overrated. Abstractions always leak, the question is whether it is part of the API and explained or not. Another example that is not indices: async/await is a leak of the underlying implementation and inner working.
> the question is whether it is part of the API and explained or not
More broadly it's a question of pretending things can be wrapped up in tidy self-contained entities, and the idea that we can fully design away the messy details of reality. See also the Three Big Lies of C++ https://youtu.be/rX0ItVEVjHc?t=17m15s
All told I'd say it's an successful but imperfect abstraction. I've been very impressed by it all in all. Concurrent programming isn't easy, and it's not a silver bullet, but it's a great language feature.
In an ideal world you'd just take existing synchronous code and throw async/await keywords at it to make it async. The abstraction isn't quite that successful, of course.
If people get lost in the details, they are not prepared to fully understand the big picture.