So he starts off with stuff about how he's feeling old and the surest sign of it is bitching about "kids these days".
You need to have read more of Joel's writing. That's just his irreverent style.
I didn't have a problem with that part. I felt that his age DOES show, but that's not why.
His premise is that, in order to be a good programmer, you need the right kind of metal aptitude which is a you-either-have-it-or-you-don't thing and not a skill that can be learned. While there may be other ways to test for that aptitude, his claim is that one sure-fire way to test for it is the ability to understand pointers.
I get his premise. I just think he's wrong.
I worked with a guy who understood pointers. He was a brilliant guy. He was also a terrible programmer. His code was universally unintelligble -- and before anyone claims the fault was on my end, it's not. I was the guy in the office who understood pointers better than he did. He would write shell scripts and awk, and they were just as unintelligble. They weren't a case of being so clever that lesser minds struggled with them. They were just complicated in needless ways. Other guys on staff could modify his code and make it both more efficient and more readable in one shot.
When interviewing potential hires, I'm more concerned with how they break down a problem than anything else. I've hired guys to do C, Java, perl, and ruby among others. I'm not perfect, but better than 90% of the time I give a guy the green light he turns out to be solid.