[tor-talk] single entity running >36% exit probability with 5 relays? scary.

[nusenu]

> After a bit of investiguation turns out 3 new relays were advertising a
> HUGE bandwidth.

The advertised bw should not really matter when we have bwauths
providing measurements, no?

So lets have a look at what measurements they provided here.

Example: consensus 2015-12-10 14:00:00 [1]

relay PhantomTrain6 had a consensus weigh of: 2150000

(that implies that the median of bwauth measurements is 2150000 or
~2GByte/s)

longclaw measured: 2150000
maatuska measured: 1280000
moria1 measured:   2150000
gabelmoo no measurment
Faravahar no measurement

tor's definition of the median of 1280000 2150000 2150000
is: 2150000 (correct)

so the actual question is:

Why do bwauth come to such measurement results?




(these URL have a limited lifetime)
[1]
https://collector.torproject.org/recent/relay-descriptors/consensuses/2015-12-10-14-00-00-consensus

measurement votes:

https://collector.torproject.org/recent/relay-descriptors/votes/2015-12-10-14-00-00-vote-23D15D965BC35114467363C165C4F724B64B4F66-B1ACC50C2312FDBF79A865CF02C84768B0A0DB42

https://collector.torproject.org/recent/relay-descriptors/votes/2015-12-10-14-00-00-vote-49015F787433103580E3B66A1707A00E60F2D15B-864EC3E54278C0712395FABB8AA0693CF134DFB7

https://collector.torproject.org/recent/relay-descriptors/votes/2015-12-10-14-00-00-vote-D586D18309DED4CD6D57C18FDB97EFA96D330566-E4B0896C35645E9F77FA480D376A09E2F0D0D335