[tonyarkles]

I realize I'm going against the grain a little bit here, and I realize that reversing a linked list is a pretty contrived problem. But...

One of the things I look for is people who have a decent understanding of how the implementations underlying abstractions might be implemented. I've found that that can dramatically reduce the amount of unnecessary complexity and work (either by leaning on the properties of the implementation, or by using a different abstraction that is more appropriate).

A prime example from my work history was with Google App Engine and the data store (this was in 2008ish, before they had a solid SQL abstraction). I read the documentation for the API and kind of "got it", but it wasn't until I read the BigTable paper that I really understood how it worked and how to design for it. Each query has two parts: an index look up, and a bulk read from disk. If you're trying to do something that doesn't fit into that model, it's going to be SLOOWWWW. Once you understand that, you can be more creative in designing your indices and structuring your data to fit into that underlying model. Way better than having your "simple query" explode into 1000 subqueries.