This (event taxonomy) is an interesting conundrum. I brought it up in the Thwack Camp presentation because I think it's a common stumbling block, too. In the end, I'm just not sure how/when to introduce the concept. If we put too much conceptual stuff up front it makes the product seem very intimidating and ain't nobody got time for that! But it is really how the "nerve center" of LEM works when it comes to normalized data. I have heard the argument that customers of LEM shouldn't have to care about that stuff, but I don't foundationally know if that's truly possible or even a good idea.
Maybe a blog post on this topic would be helpful, but it still sounds pretty boring when I think about it.