I think the whole idea is summed up by pretty much two moments, the first being the slide which gives the reasons why they think a browser based dev environment is a good idea, which contains just an "everything else is web based" argument. The second being towards the end when extolling the virtues of a browser based tool and he says, without a hint of irony, "if my laptop broke I could just borrow a laptop and show you this demo on that, if the internet was working"
It falls into the classic trap of thinking that to get the benefits of the web you need to be browser based. We really ought to have moved past that stage now, smartphones have demonstrated to everyone that people prefer the smart-client approach pretty much universally. It's just not necessary to lose the benefits of rich local applications in order to gain the main benefits of online tools.
AndyC, I'm a little puzzled here. I would die for a browser-based IDE. Just think of the integration with the web: easily getting libraries and documentation, getting snippets faster and up to date, better search etc. Granted, all these exist today, but I feel they'd be better. Also if the sites he mentioned in the talk, beautifiers, linters, tutorials, and the rest, if they could be integrated to the IDE, it would be even better. Also a multiuser experience like in Google docs with chat etc, would also be a killer. And yet you say yawn. I just don't get it.
BTW, his comment you quoted about if a computer breaks, is on the mark. It seems whenyour laptop breaks, you'd rather install the OS, the IDE, SCCS of some sort, connect to your repository, download the project and dependencies, and i may have forgotten some steps