At first we thought the high error rate could be the result of the fact that we always run Safari on Windows, while it’s made by Apple. That’s easy to dismiss, because earlier versions of Safari were fine.
The earlier data points on the graph show low error rates for Safari browsers on Windows.
The article may have benefited from including a larger caveat about only testing one OS, but this seems like an intellectually honest elimination of a confounding variable.
Perhaps intellectually honest, but a weak conclusion nonetheless: in my understanding, getting Safari 5 to run on Windows requires porting parts of Apple's huge, complicated Cocoa framework to Windows. That framework is always changing, and so is the browser-- and they both primarily target OS X. With OS X, Windows, Cocoa, and Safari all moving quite a bit between Safari versions, I would be surprised if it didn't become harder to support Windows as the product grows.
I'm sorry, what? That's a pretty weak dismissal.