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

Re: RFC: The Standards Game

On Sat, 08 Nov 2003 16:35:50 -0500, Nick Mathewson <nickm@freehaven.net> wrote:


Hi, all.  I'd like your suggestions for the following document.  It's
meant to be a cute introduction to our directory agreement problem to
interest people who don't care about anonymity networks.

I'm not looking for _solutions_ to the directory agreement problem from
the mailing list: we've already discussed it a few times and I don't
think we're any closer to a solution than we were in August.  Instead,
what I want to know is this:
- Do you think this is an interesting example to get people
interested in the problem?

Yes, at least people who are interested in socialism and/or the IETF :-)


- How can I make it more accurate, clear, helpful, or amusing?

In [Assume that users are sometimes confused about expertise .. they get good standards] it says '(but now always)' which probably should be 'not always'.

I myself would like the term 'peers' instead of committee members, but that is just a personal preference.

I liked the [====== General scenario .. chosen by any process you like] very much. It gave me a feeling of being in control or something I think...

I guess that lg means 10log, right? I also forgot what |x| did for a while :-) Of course this means take the absolute value of x, which I remembered after a while.

You also talk about 'K' Hostile features. Why not call it 'H' hostile features? You are not using H anyplace else after all?!

The formula was hard for me until I read footnote [3]. At the time I thought an example might have been useful to get a feeling for the values of Vx and L. Maybe two examples to show a 'secure' network and a 'hopeless' network.

In '====== Motivation' the second 'Correctness' issue seemed the most troublesome to me because of nodes refusing mail or mail from 'honest' nodes getting dropped by powerful adversaries.

At the end in footnote [2] I wondered if there should be a 'back' mix in addition to the first and the last one because mixminion tries to solve the pseudonimity think that mixmaster doesn't solve. It is an issue in my (unfinished) protocol, so I think it should also be so in yours perhaps??


- Where should I circulate this to try to get some answers?

I would try APA-S for some general feedback about how hard the problem is and if people are waiting to learn about these kinds of things.

I would try alt.security.pgp for some more serious input after that and sci.crypt and sci.crypt.research (moderated) for the most 'scientific' input.

And I would try the remops list on shinn.net for some input from would be operators ('experts' in this document) and maybe the mixmaster development list, but I think the above groups already cover that group of readers.


Please do _not_ send emails of the form "Doubtlessly, the FooBarBaz
community has solved this."  Every time I've followed up on a suggestion
of this form, I've found a great deal of research on a superficially
similar yet completely unrelated set of problems. ;)

I usually give up before that ever happens to me '-) Still, such documents can be nice to browse through I think. Unless they are in .ps format and you don't have a ghost viewer installed on your MS Windows XP machine...


Thanks for your help,
				Nick

Hope this is of some use to you. With the amount of feedback you got so far I think anything helps!! Mixminion is not dead before birth or anything, is it? Also, if there isn't going to be a windows port I don't think there will be much users. Running nodes is fine, but having lots of users is better!

Regards,
Thomas J. Boschloo
--
Mycon: "We are Juffo Wup, you are non. All the non must become void"