To improve this faultiness they moved the page rendering back to the server, and laid they architecture on the AMD modules specification. Such that, every modules declares its dependencies and a loader is responsible for handling the loading of the dependencies and making sure everything is in place.
But doesn’t this the old model of the web? and basically the JS just renders here and there small amount of data internally in the user page, same as the good old days before the JS wild wild west.
While I understand the need and why they are doing it, I think that they could have come with a better solution. Like the first page rendered from the server and continue from there. It’s not clear from the blog post if that’s the way they are going. The other thing is the network bandwidth of loading many files async or lazy – what about that twitter?
The good thing here, is that they are getting rid of the hashbangs.