On: Researchers, please replace SQLite with DuckDB now

[rcarmo]

The SQLite table has no indexes, and that alone would ensure instant response in count(*). The entire article follows some pretty skewed reasoning.

Don’t researchers know how to create database indexes? That would explain a lot of what I see in day to day pseudo-benchmarking like this.

(And yes, DuckDB is pretty cool. But trying to compare things without understanding how they should work is… sub-optimal)