I interview candidates at a company with a relatively high volume of interviews and I use DSA criteria in my interviews. But here's what I do:
- I phrase the question in a way that has some semblance of day-to-day relevance. That is to say that at some point in the process of coming up with a solution, the ability to apply a relevant data structure will come up, but it will be in service of an end goal that looks like the deliverable of a sprint task.
- I come into the interview aware of multiple solutions and I am open to any of them.
- I pace feedback so that the candidate actually solves the problem by the end of the interview, no matter their level (which does mean, in some cases, literally spelling out the step to unblock themselves).
The rationale is that solving a hard DSA question doesn't give me all that much signal in and of itself. Watching a candidate bang out something with a level of complexity a little higher than fizz buzz is usually sufficient to evaluate whether the candidate has familiarity with the language. The choice of idioms and APIs can tell me things about their relative level of expertise with the stack (i.e. it can generally be safely assumed that an already employed candidate can hold their candle, and the question for me is more along the lines of "to which extent").
During the course of an interview, I can usually pick up a distinct and noticeable difference in focus between candidates, especially surrounding topics related to proactiveness/curiosity (e.g. does the candidate have understanding of aspects one abstraction level lower than the API they usually use, are they aware of well known pros and cons of some specific idiom, does their argumentation seem derived from personal experience vs parroted from a hivemind, etc). This tends to correlate surprisingly accurately to how much autonomy and growth they demonstrate on the job.
"Hardcore" DSA evaluation only really comes in as a criteria to determine whether the candidate is of very high quality when most other criteria have already been evaluated as acceptable/desirable. These nice-to-have criteria come into play in some cases where I want to advocate for the candidate when the evaluation panel is split due to one seemingly bad session (possibly due to factors such as nervousness or mixed signals), or inversely when the role logically demands a higher bar but the panel is situationally incentivized to hire down to meet a quota.
I've been told by several candidates that they appreciate my interviewing style, and conversely, I feel like I get a much better feel for the candidate than strictly evaluating DSA skills and nothing else.