The engineer that wrote the memo needed people for his team. To get people to join a team from any of the others at Google, you need to convince them that your team is awesome to work with. At Google, you can't offer more money/benefits, so it has to be a "join us and you'll change the world!" message. That's going to lead to a lot of overblown statements that may or may not end up being true, and may or may not even be believed by the author.
Google is not in the business of doing things that are bad for the web (and when they have done, there's been an about-face, see: Google Video shutdown), even if internal machinations might make it seem that way.
I don't know what the timeline is for Dart but I will say this:
1. If anyone is capable of the long sort of time frame that something like this can benefit from--event requires--it's Google;
2. If there is anyone who's qualified as a domain expert in Javascript generation it is, by virtue of GWT, Google; and
3. If there is anyone who's qualified to speak to the limits of what you can do with Javascript it is, by virtue of Chrome and V8, Google.
If anything, I believe the error here (if you can call it that) is failing to properly set expectations and communicate the goals of the language (as witnessed by all the comments on this thread from people who were expecting something more and/or different).
One of the biggest pain point in web dev is the inconsistent DOM implementations. I was imagining some sort of DOM-less, HTML5 Canvas-based UI controls. And something about the "Web", Semantic Web/URIs or a new approach to programming on Web. This is NOT a break away language in any sense - its more re-packaging.