I agree with the basic point that a hierarchy is too sparse of a format to represent human knowledge. For a single category there are many different ways you mi

I agree with the basic point that a hierarchy is too sparse of a format to represent human knowledge. For a single category there are many different ways you might want to break it up. Like if I am trying to organize the category of Trees... if I am trying to build a fire I am going to mentally organize trees by "distance to me" and "burnability". Whereas if I am going to build a tire swing I am going to mentally organize trees by "is it in my backyard" and "does it have sturdy branches". But a scientific approach would be entirely different from either of these, and useless for either of my tasks. But he seems to jump from "a regular hierarchy is too sparse" straight to "therefore we need a special super dense hierarchy". Maybe the hierarchy metaphor is the wrong one for this task. Here's the logic that seems questionable to me: But I hated the idea of having to search thru all ten zillion complex items, looking for matches to some pattern. I wanted an indexing scheme that allowed you to step right to exactly those right items. And what gradually came clear to me was that the general topology to allow this, has to place a self-similar sub-hierarchy under each node on the simple hierarchy, trees within trees within trees. Why focus on the problem of "how to handle ten zillion items"? There are plenty of ways to get only a relevant subset of data out of a sea of irrelevant data. The real problem is what sort of AI data structure can help with some useful task, and this totally ignores the question of "what could you do with a fractal thicket".