> you might as well pick one at random

Indeed. Is there any scientific evidence that tech interviews are better than random selection?