~All previous methods of preference formation focus too much on what the ideas say; on their content.

Nobody really knows just how simple an idea is (Occam’s razor) or how likely (Bayesianism), for example. In some cases it’s obvious how ideas compare, but in others not so much.

Adherents of those epistemologies are all fooling themselves into thinking they’re being rational when really they’re still acting on whim. Just fancier whim.

Focus instead on how the ideas relate, on the structure of the current state of the discussion, and preference formation gets much easier and simpler.

Then you can easily see which ideas you can adopt and which you can’t. You can even see exactly which criticisms you’d need to address to change someone’s mind, say.

Beautiful.