[tor-talk] Why is 'Wgm' (middle-relay-for-guard weight) not zero?

[starlight at binnacle.cx]

Finally came up with a search that yielded
helpful information.  Searching

   "tor" "Wgd=0"

pulled a mathematically intense post written
by Mike Perry:

https://lists.torproject.org/pipermail/tor-dev/2010-January/001039.html

Which discusses validation of the statistics
with Mathematica and provides insight into
the design.  No mention of Wgm but I will
spend a few days working to understand
the math and perhaps will see how Wgm
fits into the picture.




At 12:01 12/13/2015 -0500, you wrote:
>The recent major fix for #17772 inspired
>reflection on the practical effect
>considering relay weighting.  Knowing
>nothing about it aside from the graph
>lines shown by Atlas and Globe, I thought
>the change might make little difference
>as Guard Weight would mostly prevent
>non-guards from being considered.
>
>However, per dir-spec and the current
>consensus, it seems that middle relays
>have a weight equal to guard relays
>when guard selection occurs:
>
>Wgd=0     - exit as guard
>Wgg=6065  - guard as guard
>Wgm=6065  - no-flag as guard
>
>I would like to understand the purpose
>behind this.  Can anyone comment?
>
>Thanks