The term “understanding” does a lot of work in thinking about code and data science. All code is an abstraction. And, abstractions embody a balance of tradeoffs between competing concerns like readability, performance, and maintenance costs. This post lays that out really thoughtfully!

I appreciate the technical content being put out by Lander Analytics. Consistently great stuff, on the cutting edge, thoughtfully presented. I use DuckDB all the time and read the official post about the Quack protocol, but this article made some practical recommendations that clicked for me.