Do you believe interviews that focus more on data structures and algorithms than the technical knowledge for a particular position (I.e Android, .Net) are an effective way to screen for talent?
I try to do a mixture of both. I just know that more than once I have hired someone who can do a great job coding tree traversals on a whiteboard and do every sorting algorithm from memory...who can't tell me the basics of memory management in Java.
What if they used some other platform. What if they are specialists in a platform you don't know anything about. Learning about java memory management is not hard if you have some brains and general engineering aptitude, which is what algorithmic questions show
True, but interviews aren't aboug aptitude. If I'm hiring for Android, I'm looking for domain experience and an understanding of garbage collection is a better signal of being prepared for the role than the algorithms.
At a FANG interview I solved a problem in 10 minutes using Java Collections. The interviewer was not impressed because he wanted me to write hand coded Stack, Tree and Queues. He asked me to explain the asymptotic runtime of my code (which I had done 15 years ago). The problem was super easy but I failed. However according to me no one should ever write D'S by hand because we have libraries which provide all DS. Performance is done using load testing rather than asymptotic analysis.
Interesting that this is split almost completely evenly
At the same time there isn't a scalable alternative