These all seem directionally correct and broadly unobjectionable (their reverse definitely sounds less correct), so I’m pretty surprised by the −7 agreement karma from 6 votes. I don’t know whether people are disagreeing with you being a good AI forecaster, or with your advice (and with which advice), etc. Like, (1) is this, (2) is this, (3) is “all models are wrong but some are useful” and also Chapman’s point that finding a good problem formulation is often most of the work of solving it and explicit models enable better problem formulations by making explicit the vocabulary you’re choosing for describing relevant factors at the right level of description, (5) relates to this, etc.
These all seem directionally correct and broadly unobjectionable (their reverse definitely sounds less correct), so I’m pretty surprised by the −7 agreement karma from 6 votes. I don’t know whether people are disagreeing with you being a good AI forecaster, or with your advice (and with which advice), etc. Like, (1) is this, (2) is this, (3) is “all models are wrong but some are useful” and also Chapman’s point that finding a good problem formulation is often most of the work of solving it and explicit models enable better problem formulations by making explicit the vocabulary you’re choosing for describing relevant factors at the right level of description, (5) relates to this, etc.