On top of that they want you to come up with the optimal solution on the first try, which is unrealistic and a poor judgement of programmer competency. Why? Because the answers can found anywhere on the internet.
Even if you succeed with just memorising and having some understanding of these solutions, the interviewers will still accuse you of cheating. To be frank, these challenges are best suited for coding competitions and using them for interviews is pointless. It can be the most efficient solution created in less than a minute but if its unreadable to the team, then its a 'write-only magic solution' which is useless to everyone.
Instead, a showcase of selected open-source projects or a link to some contributions is more of a realistic way of hiring potential candidates and will give a better sense of some choices on design patterns, code quality, code review comments etc, which any 'developer' can do. For example, the author is the creator of NetNewsWire [0] which is open source and can be freely reviewed on GitHub. I have no doubt that showing that alone is a reason to bypass 'coding challenges' to go straight into the onsite interview. You can't cheat with this if either you're not on the AUTHORS file or your name/email isn't in commits or file header in the original repository.
So, these coding challenges are hardly useful for assessing code quality in a team but only serve as a filter for competitive programmers who risk sending unreadable excellent solutions that only they can read but would be difficult to refactor for others; hence write-only code.
[0] https://github.com/Ranchero-Software/NetNewsWire/blob/master...
