Yeah, the claims appears to be the usual ”-We do the desktop thing, but it's actually completely a different thing because we implement it in a browser, and it's natural confines".

However, imho one of the major problems with software related patents is allowing an extremely ambiguous language in a field where exact language is the native form of expression.

It simultaneously makes it excessively expensive to challenge a patent, and almost impossible to divine if a particular implementation is covered.

There is code I've written that a client wanted to patent, and knowing the nature of the code, it's not a snowball's chance in hell I would recognise that (my own) algorithm if it was written in patentese, at least without spending days on it.

I don't even want to think about how many hours of legal fees it would cost to challenge even such a minor algorithm.