I wouldn't say Dart is worthless, it is actually a really interesting technology. Dart is definitely geared towards creating larger, more complex web applications.
Lots of developers are trying to create Single Page Applications with a very interactive experience. With this comes more client side code. Dart gives you the ability to better organize your application through packages (check out their package manager: http://pub.dartlang.org/). You get optional strong typing, which is always a plus for larger applications that need to be maintained over a long period of time.
Because Dart is it's own VM it could narrow the performance gap between native applications and web applications.
Lastly, Dart has an IO library that can be used to create the server side of web applications.
I'm going to wait a little longer before jumping onto the band wagon myself (have only fooled around with it a little), but I am looking forward to seeing how it matures over the coming months.
By worthless, I was alluding to the fact that all the performance related benefits are still hindered by browser incompatibilities and lack of support by other browser vendors. As an ecosystem, I think it is a pretty interesting piece of technology. I just have my doubts that it actually solves any "huge" problem. But I can imagine a good use-case, the enterprise. I believe something like Dart is more appealing to hermits who are still working with Java and .Net Stacks for the web (intranets and stuff like that). However, I just don't see it succeed with open-source, modular and startup-type tech. Again, I can be completely wrong about it, but if I'm allowed to use something like Node.js as a benchmark, there seems to be a huge difference in terms of adoption and growth speed between both, obviously in Node's favor. Anyway, you make some good points and I hope to hear from more people about it.