Saying this method of interview is not 100% effective so it's useless is not a great argument. Reality is that bad hires are really expensive (as argued above) and that current interview methods are somewhat - admittedly not 100% - effective at avoiding those while mostly - again not 100% - effective at admitting the great hires. Maybe this "convert to discussion" method is more effective (I have my doubts), and I'm sure there are more effective methods out there and we should look for them, but let's recognize the reasons for what we have while we look for something better.