nice handwavey dismissal. to call these microbenchmarks (especially when viewed collectively) for the purpose of dismissing the results is dishonest at best.

nice handwavey dismissal. to call these microbenchmarks (especially when viewed collectively) for the purpose of dismissing the results is dishonest at best. each datapoint on the graph represents a lot of internal metrics (GC, JS, DOM, layout/css). these aren't some one-liner JS loops. when you have a diverse set of "microbenchmarks", and they all show the same trend, you can be pretty sure the results are not "contrived".