Hacker News Reader

Ah that's right. So there's only one or two people sampled per subcategory.

That would be very small indeed, I interpreted it "favorably" as as 6 interviews but each did yield more than 1 datapoint per category.