I agree as such. Given the the choice between someone who's really smart and can solve hard problems, but a mediocre programmer or someone who's a good programmer, but sucks at problem solving, I'd chose the first one. That being said, if you hire someone in the first category, don't expect them to be happy and competent at writing an iPhone CRUD app in two weeks after seeing Objective-C for the first time.
I think that was the thrust of the article. If you're hiring based solely on someones math and algorithm skills with zero concern about their coding skills, you cannot turn around and be angry when their coding skills aren't what you needed. It's not math vs programming as such, but more generally about tuning your interviews to finding the skills you actually need (as opposed the skills you think you need).
I think you're still suffering from same mindset that led to the problem described in the article though. He wasn't a mediocre programmer - he was a bad programmer. If you make a lot of assumptions about people in an interview you will probably be disappointed.
I think that was the thrust of the article. If you're hiring based solely on someones math and algorithm skills with zero concern about their coding skills, you cannot turn around and be angry when their coding skills aren't what you needed. It's not math vs programming as such, but more generally about tuning your interviews to finding the skills you actually need (as opposed the skills you think you need).