The phrase "I Notice That I Am Confused"

In LessWrong-speak, “I notice that I am confused” is a pretty standard phrase. I’ve seen Zvi in particular use that phrase in a great many contexts where my understanding of it doesn’t match. So here’s my understanding of it.

“I Notice That I Am Confused” is the phrase that accompanies/triggers the motion “pause, attach the debugger to your world modelling system, and replay the last explanation you accepted”.

Here are a couple of possible reasons I might want to attach the debugger:

  • I can feel my world model contorting itself around something. (Maybe I can only feel the fact of a contortion and not the nature of it.)
  • Some part of my world modelling system appears to be actively working on something, but I can see no reason for it to do so. An idle process should not be consuming 20% CPU. Maybe it’s still shaking out the implications of something (which means by definition I’m still confused).
  • Or maybe what it’s working on is “make sure I never see this flaw in the model”! That actually happens! The modelling system wants to feel that its model is correct, and it’s only of secondary importance that the model actually be correct. I am very familiar with the mental motion “there’s no need to look over here, and there’s definitely nothing over here that would invalidate the explanation”.

Those all feel quite adversarial; here’s one which is more of a cooperative process:

  • Maybe there’s just a nagging sense that not all the data is explained. An explanation feels complete, but the model is highlighting that some piece of data doesn’t quite fit (possibly I haven’t yet consciously identified what doesn’t fit or why).

The point of the phrase “I notice that I am confused” is to indicate “this explanation feels compelling, but something is off and I need to promote this to consciousness to discover why”.