Sometimes the best ideas are nothing more than reformulations of existing ideas in a convenient package. The iPad springs to mind, but that’s not what I came here to talk about. Another fine example is this hierarchy of disagreements by Paul Graham (via Marbury). There’s nothing there that hasn’t been observed countless times in human discourse, but the innovation lies in arranging the types of disagreement into a hierarchy according to how convincing they are, and (especially clever) giving them short labels. Thus:
- DH0. Name-calling.
- DH1. Ad Hominem.
- DH2. Responding to Tone.
- DH3. Contradiction.
- DH4. Counterargument.
- DH5. Refutation.
- DH6. Refuting the Central Point.
Graham notes that there’s a distinct shift in quality after DH3. Up to that point you have rhetorical techniques that will always unconvincing. DH4 is the first level that has any capacity to convince.
I believe there’s a simple reason for this shift in quality. DH0 to DH3 do not require that the speaker make any attempt to understand what their opponent is actually saying. You can do DH4 while missing the point entirely, but it at least involves staying on roughly the same topic.
I am particularly interested in what happens when one participant in a discussion is operating in the range from DH0-3, and the other is operating at DH4-5 (DH6 is, I think, seldom attainable). That is, one participant is discussing the topic and the other is just playing games. It’s tempting to just assume that this is a fixed state: no discussion is possible because one person is a troll, and they’re not going to change. But this isn’t the case in practice: sometimes people do shift from DH0-3 to DH4-5. The trick is to work out whether there’s any potential for this in any given conversation.



