The key problem with Google Now is that you have to remember to look at it, and so, most people don't get the benefit as much as they should. The act of having to physically remove the phone, unlock it, to see contextual Now card notifications IMHO is a transaction cost that makes it less likely you'll use it effectively.
You need to be able to see and dismiss a contextual notification within a second, like a warning light in a cockpit. It should be effortless and almost second nature.
That's one reason why the glass, or going further, direct-mind-machine interface, would be better - speed. Latency matters. I am often deterred by the time it takes to unlock a phone to perform an action. For example, I once witnessed a car accident and wanted to snap a picture, but it took too long to get my phone out, and I missed the moment. Other times, I want to translate something I'm looking at, but bringing up an app to do it is just too annoying. Look at the way the characters in Star Trek the Next Generation call on the computer to answer questions as if it is always just there for the asking. That's the kind of responsiveness you want.
I think a lot of people assume that a device that sits on your face would be "always on", constantly distracting you, but I'd argue that the ideal device is one that is out of your way, but activates extremely quickly on demand, or contextually as appropriate, like a subconscious subroutine or personal assistant.
Too much effort I think is focused on immediate mass mainstream adoption. It's as if, you can't have an iPhone level sales volume, it's not even worth doing. But there are lots of things worth doing that the masses don't necessarily adopt out of the gate. Someone needs to lay the foundation that the future is built on, that other people can build on. Google Glass doesn't need to "kill" the iPhone to succeed.