There's a lot we're trying out here, but there's also some things that we're not focusing on for this test. Here's a couple that are "out of scope" - which is to say, we know that they're issues and we'll want to address them in the future, but we also want to experiment quickly to find what works and what doesn't.
Performance optimization for extremely data-heavy sites
If your site's pagecount reaches into the thousands, then you're gonna hit the limits of the system pretty quickly. It's not to say you shouldn't play around with it, just that it'll be slow to the point of being broken in some cases until the experiment stabilizes enough to start worrying about optimization in a real way.
Customization of look-and-feel without needing to learn to code is planned for the future.
Similar to Source Mode, above, we'd like to add standardized widgets that offer extended functionality where possible. This will help prevent the scourge of a million slightly changed copy/pastes of the same <insert widget here> that currently makes it impossible to fix bugs across multiple communities at the same time.
That said, an SDK and easier path for creating and contributing widgets is planned for the future.