On the flip side, as an interviewer, this
all incredibly true.
In particular, I frequently end up asking algorithmic questions during interviews. For all the hate that they seem to get, I find them to actually be quite predictive; in my experience, the people who are extended offers despite just barely getting by in that portion of the interview are also the ones that struggle when those types of issues come up during actual development- and yes, despite claims to the contrary, basic fundamental algorithmic knowledge really does turn out to be useful in a surprising number of circumstances.
However, I will say that the utility is entirely based on how the interviewer handles things. I've passed many candidates who have failed to complete the problem at hand because throughout their attempt they demonstrated an ability to communicate, an ability to reason about a problem, and an ability to respond to feedback during the course of trying to arrive at a solution. In particular, one of my favorite questions has a tendency to make candidates try an approach that ultimately ends up being a dead-end as their first attempt. That's fine! It's completely natural! The really interesting bit is how they they respond when I produce counterexamples that start showing the limitations of that approach (and, ultimately, its infeasibility.) Do they keep digging the whole? Do they rethink their approach? Do they say "Yep, that makes sense, but let's see if we can get this done for the easy case and then we can approach the more general problem?" Those sorts of questions are the ones that ultimately end up being the most valuable.