[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: [tor-talk] some confirmations about Tor Metrics research

Hash: SHA1

On 25/03/15 16:06, Anonymous Kyoto wrote:
> Dear Karsten,


> I am not sure whether it is right to ask you some questions about
> the Tor metrics via personal email like this, or should I ask
> through Tor-talk mailing list. So please forgive, and tell me if I
> should ask you via the public mailing list.

Well, since you asked, let's move this discussion to tor-talk@.

> âAt the footer of the paper :"Measuring the Tor Network, Evaluation
> of Client requests to the Directories". link 
> <https://research.torproject.org/techreports/directory-requests-2009-06-25.pdf>,
you wrote that "*This report is mostly superseded by report 2010-10-001. *"
> Is it a mistake because I could only find the  Report 2010-11-001
> with the title: "Privacy-preserving ways to estimate the number of
> Tor users" instead of "*report 2010-10-001*"

Oops, you're right.  Too many ones and zeros.  Fixed.

> âWhat do you think "when will be the appropriate time for
> introducing the counting-cell approach"? Because the proposal has
> been there for almost 5 years, but still has not been implemented.

Nobody is actively working on it, AFAIK.  I don't expect it to happen.

> âIs it correct to separately estimate and then take the sum of both
> "new or returning users" and "recurring users", because as far as I
> am aware to become a "recurring user", a user needs to be a "new
> user" at first?
> By summing up the two values, do you mean that the number of daily
> users consists of "new or returning users" which are those newly
> joining users, and  "recurring users" which are those users keep
> using Tor since the days before? If it is the case, then please
> pass this question.

We stopped distinguishing the two user groups.  Please take a look at
this more recent report:


> âIn the paper, you wrote:"Our current approach for estimating
> (recurring) daily users is based on the fact that every client
> needs to refresh its network information *every few hours* in order
> to make indistinguishable path selection decisions."
> So my question is that: "why *every few hours* but not *every one
> hour*? because to the best of my knowledge, Tor directory
> authorities publish a new consensus every one hour, and all clients
> has to follow this part in the "Tor directory protocol"
>> *Clients try to have a live consensus network-status document at
>> all times.*
> *   A network-status document is "live" if the time in its
> valid-until field *
> *   has not passed. *

A consensus is *fresh* for one hour but *valid* for three hours.  Look
at these three fields in the most recent consensus:

valid-after 2015-03-26 10:00:00
fresh-until 2015-03-26 11:00:00
valid-until 2015-03-26 13:00:00

> Thank you so much for reading my email, I know that you are very
> busy, so I would be really grateful if you can just give me a short
> answer or even a hint where I can find the correct answers.

Questions are fine.  And in this case they led to fixing a broken
reference, so thanks for asking!

All the best,

Version: GnuPG v1
Comment: GPGTools - http://gpgtools.org

tor-talk mailing list - tor-talk@xxxxxxxxxxxxxxxxxxxx
To unsubscribe or change other settings go to