Nah. I think you inferred too far. I read the sentiment simply as a call for balance in the relationship between first principles and opaque but predictive methods. I have to fight this fight all the time in my work. Even though my reputation (shambles that it is) is in mechanistic modeling (a type of first principles), I have to consistently argue that the opaque (ML) or largely meaningless (over-fit) models can be just as useful, depending on the context.
At my first co-op gig, building trailers to house the engineers launching rockets in the desert, a colleague offered to help me build some speakers for my truck. He pulled out these charts and tables that mapped the properties of the acoustic space, power available, speaker sizes, box sizes/shapes, frequencies (mostly lower to resonate in the shelled bed), etc. Bunches of numbers that he claimed to understand, but with my EE classes under my belt I could tell he didn't *really* understand them. >8^D So, I derived most of the properties of the speakers from (what I thought were) first principles. I got nearly the same answers he got, which was very satisfying for both of us. I think had either of us bailed on our role, the project would have been much less satisfying ... asymmetry between the two types of understanding is unhealthy. On 5/14/20 5:28 AM, Steven A Smith wrote: > I *think* this discussion (or this subthread) has devolved to suggesting that > predictive power is the only use of modeling (and simulation) whilst > explanatory power is not (it is just drama?). -- ☣ uǝlƃ .-. .- -. -.. --- -- -..-. -.. --- - ... -..-. .- -. -.. -..-. -.. .- ... .... . ... FRIAM Applied Complexity Group listserv Zoom Fridays 9:30a-12p Mtn GMT-6 bit.ly/virtualfriam unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com archives: http://friam.471366.n2.nabble.com/ FRIAM-COMIC http://friam-comic.blogspot.com/
