Unfortunately, the skill of being trainable/educable is very unevenly distributed in the population and very hard to assess from afar. Instead, companies hire based on whether you've already been trained/educated in how to do the thing, which amounts to letting someone else take on the risk associated with your failing to learn something.
If I were tasked with solving this, I'd advocate for more frank discussion and use of various pseudo tests, e.g. SAT or GRE scores. This strikes a lot of people as unfair and non-inclusive.
That's a very good point. Over the course of my career, the best people I've seen are those that are really good at educating themselves and keep constantly learning new things. They were a small minority though of the coworkers I've worked with and an even smaller minority of people I've interviewed (the problem is that when trying to hire people who have those qualities, it's hard to avoid false positives).
I think this has less to do with assessment issues in trainability and is rather an economic thought: I don't have to train the employee that has already been trained by my competition. You can see a similar effect in companies pushing more and more training to academic institutions, expecting them to produce full fledged developers from day one (at least it's the case here in europe).
All people I know who have been good at leetcode have been good at coding (of course they may lack other important skills but that you can assess with other questions). You can't become good at leetcode without being good at logical thinking. But not all good coders are good at it because it also requires skills unrelated to the job to be good at leetcode.
I do not advocate for leetcode tests, but I think they have a bit unfair reputation. There are many worse interview questions.
I agree with you, for what it's worth. I think some of the "easy" leetcode questions are probably a decent filter, they're often contrived enough to force you t handle an edge case or two but simple enough that they should be solvable within a few minutes - they're a step above fizz buzz.
They filter for a few super important qualities - can you solve a simple problem, follow some basic instructions, write somewhat comprehensible code. But most importantly, if you get stuck, how do you react. If your response is to lash out and blame the questions or the interview, that's a huge red flag that you're going to react like that when challenged in the work place. I wish I wasn't speaking from experience.
If I were tasked with solving this, I'd advocate for more frank discussion and use of various pseudo tests, e.g. SAT or GRE scores. This strikes a lot of people as unfair and non-inclusive.