Comment by beckhamc

The issue is the obsession with benchmark datasets and their flaky evaluation