I totally agree, with your point that we should use more real life problems like you mention. The issue here is that this often requires a lot of preparation from the interviewer. You can't always share the code you are working on with a candidate. I would have to find something similar, maybe setup a backend, etc.
Would you ask this problem to be solved on the spot or let them try at home with the help of friends? Should I restrict time e.g. 2 hours to create a login screen for an iOS app. Or let them take as much time as needed?
I'm interested to know if I could scale this process. We have sometimes to hire batches of 6 to 10 engineers at the time for very different projects and that's when these code challenges seems attractive. Thoughts?
Coding challenges are getting very popular now because they save time during hiring processes. Recruiters use them to filter out candidates. Usually having a bad answer is better than an incomplete one. For that reason it is important to start with a simple solution and iterate over it.
One frustrating thing in my opinion is that coding challenges only measure CS skills like data structures, DP, graph. While important those are not as relevant as more specific knowledge like HTML5, iOS, Android, etc.
Very interesting! And also a great talk. But I understood from it is that once you pass the Google hiring bar only then this inverse correlation happens.
What about to evaluate some candidate from the street?
Very interesting that they use real applications as a code challenge. How did it work initially? Did they just sent you the code repo or used a testing tool like hackerrank?
Quite the opposite: in that scenario swift could be implemented by multiple vendors like microsoft. This would increasing the reach of swift apps (more appstores, devices, etc).