On Tue, Sep 17, 2013 at 08:27:57PM +0200, Moritz Bartl wrote:
The recipient share is calculated from the throughput per relay * country factor
It might be worthwhile to make it clearer what "throughput" is here.
I hope it's not consensus weight, since that's not really a measure of how much use the relay sees.
It could be the bandwidth listed in the descriptor, though that could be gamed.
The bandwidth history in the extra-info descriptor might actually be most accurate, since it reflects bytes actually used. But it seems hard to accurately count that up for a long time period without double- or under-counting. Unless maybe stem already does exactly this for us? (These numbers are still gameable too, to be fair.)
I guess just about any answer is a fine start, but I expect you'll have people wanting to know which answer you picked.
Thanks for keeping this going!
--Roger