If "success" is a coin flip, that means that EVERYTHING inside a company failed--interviewing, hiring, training, management, everything. If even one of those things was working, the success rate should be better than a coin flip.
50% success rate doesn't indicate that recruiting is like throwing a coin flip. If there are only 10% of people that can do the job across interviewee, then throwing a coin would get you just 10% of people able to do the job.
However if there was a pool were 80% of people are able to do the job, getting 50% would be worse than picking anyone randomly.
Bayesian logic :)
1) if you chose randomly from the pool of job applications, does OP really think you have 50% chance to fail / not fail? (probability doesn't work this way)
2) what does it mean that the hire "failed"?
EDIT: I do agree with the general sentiment of the article though.