> we should stop supporting X browser because some kid doesn't know how to support it too?
No, we should look seriously at whether we need to keep doing things the cheap labor can't perform anymore.
Supporting IE6 is a Black Art, and keeping people who can perform Black Arts reliably is expensive. I suppose the traditional comparison is to COBOL codebases and greenscreen IBM development. If thinking about that convinces you IE6 isn't the mainstream anymore, go for it. The point is that increasing the number of different, and I mean really different, browsers your team has to know about is not free, as the cheap labor only knows the relatively tight cluster of post-IE6 graphical browsers. The article's the evidence of that.
... and keeping people who can perform Black Arts reliably is expensive.
I have nothing to add ("me, too!"), but would like to point out that this sounds like a line from a fantasy novel written by an economist. If such a thing actually exists, I would love to hear of it.
For the majority of websites it's not so much a Black Art as something anyone who designed websites more than 5 years ago should manage competently (assuming there aren't unreasonable requirements such as pixel perfection or reproducing every hover effect). The comparison is more finding someone that knows how to program Python instead of copy/pasting PHP scripts.
Html and css were not designed to be an arcane programming language, but a simple layout markup. If well designed then copy paste should work fine. Arguably we didnt quite make that simplicity, but I think your Python PHP comment is way off as far as html css is concerned...
No, we should look seriously at whether we need to keep doing things the cheap labor can't perform anymore.
Supporting IE6 is a Black Art, and keeping people who can perform Black Arts reliably is expensive. I suppose the traditional comparison is to COBOL codebases and greenscreen IBM development. If thinking about that convinces you IE6 isn't the mainstream anymore, go for it. The point is that increasing the number of different, and I mean really different, browsers your team has to know about is not free, as the cheap labor only knows the relatively tight cluster of post-IE6 graphical browsers. The article's the evidence of that.