Mountains of research shows that interviewing job candidates doesn't tell you much about how they'll perform. Depending on whose research you believe, 45% to 75% of new hires don't deliver what employers originally wanted. Assessing performance through exercises doesn't tell you anything better.
This means that tossing a coin has as good a chance of finding you the right person as the more lugubrious process of weeding through résumés and sitting through hours of painful interviews. You could just hire anyone. It would save you time and trouble, and you'd get to learn about the candidate the right way: on the job.
One of the big problems with the interview process is that we invest so much time in it that, when the candidate turns out not to be the ideal we expected, we cannot accept that we got it wrong. So we all spend a great deal of time trying to make things work. If we invested less in a broken system at the beginning, perhaps we'd have the courage to bow out faster.
I recognize that, on some level, this is an outrageous proposition.
But it's spurred by the recognition that interviewing people is pretty outrageous, too: time-consuming, expensive, and emotionally taxing. That might be OK--if it worked. But it doesn't. We aren't the great judges of character that we think we are. The process is hopelessly artificial, and neither boss nor candidate enjoys a very honest exchange.
As I see it, you have two choices: Invest in the system more (by bringing in personality and effectiveness assessments, tracking down informal references, and doing more interviews); or, simply, invest in the system less.
Having spent years trying to make the first option work, I'm willing to give the second a chance.