1. 3
  1.  

  2. 6

    I have some reservations about their analysis.

    1. I don’t find averages at all useful for discussing latency across the internet - even more so with quic where most of the benefit is in high percentiles.
    2. BBR vs CUBIC is a significant difference. I’d expect that to dominate the comparison, not http 2/3.

    Since this is just an”here’s where we are now” PR post, that’s fine, but I hope people don’t read too much into the numbers.