I really don't mind that sort of phrase when it's pointing out that the cutting-edge features that they are demonstrating, while they are on the standards track, currently work best in a particular browser.
On an actual production web page? Sure, that's awful. Blocking people on other browsers from viewing (like Apple did with their HTML5 showcase)? Yeah, that's bad. Demonstrating a technique that still requires vendor prefixes, and not applying all of the vendor prefixes that work? Well, that's just lazy.
But informing people that the effects you are demonstrating work best in a particular browser because the other browsers don't happen to implement those features yet, or aren't fully compatible yet, isn't bad. It's just preventing confusion for people who come there and don't see the given effect.
In order for web standards to progress, we need experimentation. We need early implementations, that test out the space, and we need designers who try out these implementations and see what you can do with them. And yeah, when that happens, they will be experimenting with features that aren't supported everywhere, and will create examples that only work on one or two browsers.
Complain when you see this in production. Complain when you see people recommending user agent sniffing. But for early stage experimentation? It's not a problem.
For some weird reason his javascript only sets -webkit- prefixed CSS properties; it's quite likely this would work in firefox (and possibly eventually IE10), were it not for this.
Sadly doesn't obviously transfer well to anything other than a flat glossy object though.