On 13 Dec. 2016, at 03:28, Toralf Förster toralf.foerster@gmx.de wrote: ...
I do wonder if the measurement of the former is somehow affected by the later ?
... Toralf
Yes, the bandwidth authorities measure the bandwidth, then calculate the ratio between the advertised and measured bandwidths. This becomes the consensus weight (after averaging, and then taking the low-median of votes).
Clients choose relays based on consensus weight, which then affects the bandwidth they carry, which affects the observed bandwidth, which is one of the factors that feeds into the measured bandwidth.
Part of the calculation process for the consensus weight makes sure that this feedback converges, rather than diverging.
T