The process for successful interviewing at places that use whiteboard testing for interviewing involves digging in and putting in a lot of hard work studying previously unknown concepts-- work that most people don't consider very fun.
To me, this sounds like exactly the kind of thing that a person will need to do to be successful in their first few months in a big company. Learning about their particular systems, authentication and authorization methods, security audit procedures, deployment methods, how they do dependency injection, logging, diagnostics, troubleshooting...
I find this very ironic: Hiring for suitability to handle edge cases is also a brute force approach that doesn't scale.
A good algorithm optimizes for expe ted inputs and farms out cache misses and unexpected behavior to some other system.
Overfitting for the possibility that it might be necessary in a design choice is not the way to gatekeep candidates
Yes the whole interview process is based on the idea of doing stuff for the company that you wouldn't actually do.
Indeed I interview my candidate's ability to find resources, to know what considerations are needed, how to deal with IDEs, collaborative tools, the git protocol. I will fire someone on the spot for implementing an academic algorithm if there was a library and potentially peer reviewed library they should have used instead.