The Importance of Documented Understanding of the Finicky Features of Machine Learning Models

I think this is actually very important for everyone to understand and internalize, because like I'm sure you've seen the Dylan Patel, the semi analysis leak where a senior Google engineer was talking about not really having a technical moat. Having a kind of documented understanding of a lot of the finicky finicky features of machine learning models can matter a lot even if they're independently discoverable by your kind of open source organization. TSM's advantage is they're buying many machines that outside of China anyoneAnyone can buy a budget. Yeah, so, but they also have many decades of process knowledge and not just process knowledge on doing the process but also process knowledge on training other people to do