WYSIWYG style detection is a much harder problem than it appears at first. We are aware of the “turd” problem and we’re not happy about it, but at the moment it’s the only way we’re able to do WYSIWYG style detection. I do regret that we don’t warn the user what’s about to happen during the setup wizard.

I’d be interested in finding out why the style detection is failing for your blog. Again, feel free to contact me.

Good question about the random-looking numbers. You’ll notice that when you insert tags, it’s not just a one-shot Insert operation. After inserting tags, if you select the inserted content, you’ll see that the sidebar changes to allow you to modify the tags or tag provider. A similar thing happens with maps, you can modify maps after they are inserted. Maps and tags are examples of “smart content” objects. Each smart content object has metadata that’s not stored directly in the HTML, and each type of smart content has code that is associated with it on the client side.

The first part of the number (the one that never changes) identifies the type of smart content (maps, tags, etc.) and the second part is a truly random number that uniquely identifies this particular smart content object.

If we were really interested in tracing you, there are much more direct ways to do it–after all, we are executing code with full permissions on your machine.

Hope that helps clear things up.