Tadd, when you said, “I don't think you can't expect subjective interviewing practices or simply data science by themselves to give you the best solution here. It has to be a mixture of both." I thought - Bingo!
Sadly, while I think we'd both agree this is common sense, I was disturbed by one recruiter in the article that said his managers want to hire based on scores. I do realize that this is N=1 scenario, but I do think we need to be careful in understanding the algorithm's output presents an incomplete picture of a candidate.
That said, I'm not convinced that data science can weed out 60-80% of candidates not worth interviewing. In fact, I’m on the fence as to whether this is just too complex a problem for algos. Too many attributes, too many assumptions, and too little understanding of the weighting factors that make up an ideal candidate (of course, my statement depends on the job role). Plus, as one person on Twitter wrote to me; “How do you calculate character?”
I suppose time will tell if people analytics are more help than hindrance. Thanks for commenting! I appreciate your input/feedback!