1. 3

  2. 2

    Rigid hierarchy is generally not how the human mind works, and the strict parent-child relationship it enforces is particularly terrible for fluid human group discussion. Browsing a tree is complicated […]

    I’m wary of essentialist claims about “how the human mind works”. People generally learn to conform to all kinds of torturous social constructs, including poorly thought-out UI design idioms. We may get confused, but we rarely complain. There’s a more basic conceptual problem with modeling group discussions as trees, though – one that has nothing to do with visual layout. That is the arbitrary one-comment-per-reply restriction. You can’t know in advance how many replies a comment will receive, but there’s no good reason not to be able to reply to multiple comments at once. In practice of course people come up with ad-hoc workarounds, and may never even notice the structural mismatch. Nonetheless, conceptually, a DAG is a better structure.

    Spoelsky doesn’t touch on what I consider to be the biggest problem (anti-feature?) of threaded forums: one that applies equally to flat models, though it’s less pernicious there. That is, reordering the event timeline based on popularity. This breaks (or rather, deviously manipulates) very basic expectations about causality, and encourages a lot of bad behaviors. It only makes sense for StackOverflow because, as he notes, SO is like the opposite of a discussion forum.