The biggest reason is that when you have one implementation, the implementation becomes standard irrespective of what the standard itself says. When IE6 was dominant, this meant a lot of people had to reverse engineer the details of IE6's weird CSS implementation. Chief among its faults was an incorrect box model (although W3C later added an attribute to use IE's old box model) and the infamous hasLayout bug, which Wikipedia at one point had a page on (it appears to be gone now). Sure, Webkit may be open-source, but that doesn't really lend itself to finding issues related to buggy implementations.