You're right - it would've been very easy to replace the browser detect with a feature detect once I knew what features I'd have settled on, at the time though compatibility was a distant second to proving the concept, so I lazily didn't bother.

I have spent many years of my life working on a browser that is frequently hurt by agent detection, so I'm well aware of the issues with the approach I took; and while I do think that there are a broad range of circumstances where it is appropriate, this is probably not one of them, so I accept your criticism and and pout into my bowl of ice-cream :)

Edit: I added your suggested code and am running through the tests now, thanks.