[quaquaqua1]

Many of the problems that the median software developer is working on can be acceptably solved with o(n^2) ideas which are the types of implementations that the median developer can reasonably think of "on the spot" with little prior training. Or the job is HTML CSS and basic js.

However many interviews demand many rounds of demonstrating o(logn) ideas quickly, which I would argue takes x months of preparation, not x weeks or x days.

I think its still very important to learn these structures, but the way they are applied in brain teaser format is not effective at finding the best all around candidate in terms of salary, general skills, sociability, iterative ability, etc.

And that's just taking the process at face value. Some interviewers will take this exercise to its extreme as a reason to deny you and make themselves feel good by giving you genuinely obscure and difficult problems that even Linus Torvalds and Guido admit they can't solve today in a reasonable time frame.