Is "gatereloaded" a Bad Exit?

[John Case]

On Sun, 13 Feb 2011, John Case wrote:

> I was not aware that "the plan" had evolved into a 1:1 mapping between ports 
> you usually use in cleartext and ports you usually use enciphered.
>
> So, if my exit has TCP 19 open ... what will you accept as a suitably secure 
> chargen ?  What about 37 ?  I don't know of a standalone time protocol that 
> is encrypted ...
>
> Since there is not a clear 1:1 mapping of cleartext and not ports (we _can_ 
> agree on that, right ?) then I suppose some system of "weights" will have to 
> be applied.  How would this weigh out:
>
> 21, 23, 25, 53, 1337, 2105, 5555, 32245
>
> There are four TCP ports that you usually use cleartext, and four that you 
> usually use encrypted ... is that exit policy acceptable ?  Or do you assign 
> different weights based on how well you know the protocols ?  I've sure never 
> heard of "MMTSG-mutualed over MMT (encrypted transmission)" so ... I'd assue 
> a low weight.
>
> But then what about the malicious operator that takes a few seconds to post 
> to or-talk and "explains" their exit policy that didn't weight out properly 
> ... then what ?  Is it a panel of three ?  A panel of five ?  Are there 
> appeals ?


Where's the answer to this ?  I chose edge-case scenarios above, for sure, 
but this is the real meat of the implementation of your plans, and I'd 
like to know if you've given any thought to this whatsoever.

What _is_ the "proper" corresponding open port for 25 ?  What _do_ you 
find an acceptable match for 53 ?  What system of weights will you give 
ports that don't have an obvious correlary ?

Oh, by the way - I used TCP port 80 this morning for something other than 
cleartext HTTP.
***********************************************************************
To unsubscribe, send an e-mail to majordomo at torproject.org with
unsubscribe or-talk    in the body. http://archives.seul.org/or/talk/