From their article it looks like what they are doing is pseudo-science at best. Hopefully not harmful pseudo-science, but it could be actually harmful. If one i

From their article it looks like what they are doing is pseudo-science at best. Hopefully not harmful pseudo-science, but it could be actually harmful. If one is not careful enough it is rather easy to replace more-or-less working common sense with some horrible pseudo-scientific approach. For example, optimizing for 'individual employee performance numbers'? Without consideration that, say, adding yet another employee with 'a good number' may actually diminish the total sum? Regarding prediction of 'employee performance numbers' based on 'interview scores'. It is not surprising that the result is completely random. As far as I understand it, in the big companies performance numbers don't show actual contribution of an employee. At best they are showing "how this person was able to use the resources around them to achieve goals", note, mostly via using human resources of informal social networks. But usually performance numbers are just random. And at worst they could be negatively correlated with actual contributions.