[dtaht99]

Nice to see you keith. I agree that the network latency part of the equation tends to get too much play, where glass to glass is the better metric. On the other hand the bufferbloat effort started where network latencies were often the dominant contributor (seconds!) and for those that have actually implemented things like your e2e algos and/or installed stuff like fq_codel/fq_pie/sch_cake - the encoding step now dominates.

'Course, I still kind of miss scan-lines, and had hoped we'd finally all have enough bandwidth to just blast raw frames over the network by now.

/me hides