Which raises the question of why do the other kind of interview at all. I mean the leetcode nonsense whose end state is often a theatrical performance: “I’m pretending to invent this clever algorithm on the fly although I’ve practiced it for weeks”.

Is there any evidence that candidates who do well on design interviews but fumble on leetcode would be any worse at the actual job?

Leetcode is simply testing whether you studied Computer Science, took Data Structures & Algorithms course, and read the CLR book on algos and practices it. Nothing else. This is really freshman level cs work. very simple.

people who cannot leetcode - a simple data structures & algorithms inteview - cannot understand runtime nuances of what they write.

This is how you end up with N+1 algorithms and exponential runtime.

For example look at all modern front-end in javascript - usually leetcode is more relaxed for front end, and you end up with ungodly gobbles of unnecessary loops inside loops wrapped in loops that traverse DOM for no good reason and freeze the browser.

it is the javascript people who import node libraries that contain 3 lines of code, instead of writing it properly

Maybe. My experience is that a lot of the unprimed recall gets weak very quickly without practice. There's irony in testing for skills that a fresh graduate may do better at than someone with substantially more experience.

Primed recall can remain strong for many years, but leetcode often penalizes googling around to refresh your memory, even though that is what almost everyone should do before taking any further steps toward an implementation. Depending, it's often also the correct thing to do before choosing a library, the parameters to a function, code base organization, and lots of other details.

Of course, domain experts may be very strong in their specialty, but that's a special case with narrower scope.

I can hypothesize that testing for a "fast study" may have more predictive power. I have seen some interviews that are designed for this, leetcode adjacent but less antagonistic.

But anyway, I am spitballing, to be real with you.

You don't need to be good at leetcode to make a performant and well written frontend. You don't even need to know data structures, algorithms, or Big O notation.

what do you mean by N+1 algorithms?

IME it's a test in which it is easy to grade many people in a way that it is really unlikely that two people end up with the same grade. That is what leetcode achieves and that is really the majority of the reason it is used. So it is just very convenient for hiring. In India, we have companies coming to hire students from colleges, and as the first filter, they give you three leetcode questions, and average the number of test cases you passed in each of them and rank you based on that. I spoke to one of the companies, they said that they had _very_ few people that had the same score, and it was easy to choose within those small sets by reading their resume/whatever.