[jcranmer]

Correction: a set of microbenchmarks whose own documentation emphatically states the difficulty of drawing any conclusions from those results.

Comparing different programming languages is notoriously difficult, particularly since the metric is underdefined. The best programmers might write faster code in language A compared to language B, but if average programmers write code that is no faster, does that mean language A is better than language B?

[igouy]

> the difficulty of drawing any conclusions from those results <

The difficulty of drawing general conclusions from program performance measurements -- not just "those results".