Re: [GNUnet-developers] Found a related project

[Krista Bennett]

jan marco alkema hath spoken thusly on Mon, Jun 09, 2003 at 07:06:32PM +0200:
> Thank you for the feedback on anonymous file sharing concepts --) Next week
> I have time to study it.

You may want to study it before we continue this discussion. If you 
understand the paper, it may make what we're doing and our understanding 
of how anonymity works a bit more obvious.

> I believe more in 1 file sharing system. It could be programmed as "service"
> on Linux and Windows machines like ftp. Depending on attributes of the files
> (cds/Dvds) in database different type of file sharing will be used. 

The point that was being made here, though, is that the main purpose of
the AFS service is not to be an all-purpose file-sharing server. Its
purpose is *anonymous* file-sharing, and its construction (and the speed
trade-offs)  are inherently reflective of that purpose. AFS will continue
to have anonymity as its primary goal. If you want lightning-fast and
don't care about anonymity, GNUnet is probably not for you.

It is, by the way, configurable to some extent anyway. One can regulate
the amount of noise, whether or not the files are fully encrypted instead
of being served on demand, etc through the configuration file. Also, other
services can be implemented on top of the GNUnet core which don't have the
same overhead as the anonymity protocol. The anonymous file sharing
service, however, has the overhead it does for a reason! If you don't
understand the anonymity and encrypted file sharing bits of this puzzle,
you really don't understand anything about the file-sharing software!


> (>Being
> thrown into a prison for longer than armed robbers for writing software, on
> setting foot in a foreign country?).

Ah, yes, the good ole US of A. Mmmm. Or giving details about SARS in
China. Take your pick.

> My biggest problem with gnunet is that I should insert all my files to share
> it with others on the Internet. I have all my shared data two times on my
> computer! 

Not true. You can choose to only index the data existing on your computer 
without reinserting the whole file anywhere. The data indices are 
extremely small in comparison to the whole file. I think the encryption 
paper covers the index overhead (something like 2 percent of the file 
size, although the math is in the Efficient Sharing of Encrypted Data 
paper, page 8; the 1.02*n figure is for insertion WITHOUT indexing. I had 
reproducing math in plaintext, so I'll leave it to you to read through and 
understand the details).

The point is... you don't *need* the data on there twice. Inserting the
whole file has two advantages; speed (the blocks don't have to be
encrypted on demand, as they're already stored in encrypted form in the
database) and the ability to insert a file into the network and remove it
from your hard drive (or whatever the source medium was) and still be able
to serve the content (something the paranoid about deniability would
consider wise). But you don't have to do it.

You've been on the list for long enough now that I would *strongly*
recommend you sit down and read the papers thoroughly. I have no problem
with explaining how the system works to list members, don't get me wrong;  
the more people who ask questions, the better. It helps us find flaws,
introduces us to new ideas and points of view, and provides interesting
discussion. However, after more than a year on the list (I think), if our
main goal with the anonymous file sharing service is still unclear, it
would behoove you to try to understand our motivations before suggesting
we take a different direction :) The papers shouldn't be beyond you after
this long and this much exposure to what we're doing. Good luck with the
reading :)

- Krista

-- 
***********************************************************************
Krista Bennett                               address@hidden
Graduate Student
Interdepartmental Program in Linguistics
Purdue University
     
     "You're more important than a bowl of spaghetti!" - My mom