If you look at the chart in the second link, where does the wait time leave the origin? Around 60%. The first one is even worse; by 80% you're already seeing

If you look at the chart in the second link, where does the wait time leave the origin? Around 60%. The first one is even worse; by 80% you're already seeing twice the delay of 70%. If I were to describe the second chart I'd say 80% is when you start to get into trouble, not just noticing a slowdown. I said minimize latency, not optimize latency.