[rcfox]

> Poems about trees are indeed lovely, as Joyce Kilmer promised us, but data of any kind represented as a tree … isn't.

If you don't see tree data structures as "lovely", then I have to wonder if you even understand them. Putting data into trees can make difficult problems easier to solve in an elegant way: searches, file systems, space partitioning, and so on.

> Rigid hierarchy is generally not how the human mind works...

That's why we keep creating ontologies to explain the world? Ask any 10-year-old, and they'll tell you that tigers are cats, which are mammals, which are animals, etc.

[aamar]

> Rigid hierarchy is generally not how the human mind works...

Jeff's sneaking a lot into the qualifier "rigid" here. The human mind loves hierarchies of many sorts, but most when the hierarchy is based on the context. For example: house cats could be mammals (near tigers) or pets (near fish).

A concept or object that is relevant to different contexts needs to show up in differently-organized hierarchies, or failing that, probably ought to be organized non-hierarchically (e.g. flat or "tagged"), so as to reduce cognitive dissonance.

Over-simplifying considerably, comments on Reddit/HN tend to be interesting mainly in just one context, so a hierarchy is probably helpful.

[thisishugo]

Well, a b-tree might make a good backend for indexes or storage engines in a (large) number of databases, but mostly that data isn't looked at in the same way as it's stored. So Jeff has half a point.

However I agree with you that for hierarchal data such as, for example, family trees, biological classifications, and, dare I say it, threads of conversation, a hierarchal representation does seem to make sense.