[Mnet-devel] Freenet, Mnet, GNUnet, Circle
Zooko
zooko at zooko.com
Wed Jul 14 15:20:15 BST 2004
Dear p2p-hackers:
Someone wrote to the mnet-devel mailing list and wanted to know how
Mnet compared to similar open source p2p systems, because he wanted to
work on one of them. You can read the <a
href="http://lists.mnetproject.org/pipermail/mnet-devel/2004-July/
003429.html"> thread so far </a> in which Kademlia, eDonkey, and
BitTorrent are dismissed and GNUnet, Circle, MixMinion, and Tor
introduced.
In writing the following response, I thought it was of sufficiently
general interest that I should Cc: it to p2p-hackers.
> Why does Freenet get all the press if Circle and GNUnet perform
> essentially the same service???
Beats me. I guess nowadays it doesn't, anymore. Anyway, it isn't
exactly the same service. Circle doesn't offer anonymity, for example.
(Although <a href="http://thecircle.org.au/about.html"> the Circle web
page </a> confusingly tells me that it offers anonymous "news" but not
anonymous "file-sharing".)
> How mature are Circle and GNUnet
> and easy to pick up relative to Mnet???
I would say all four are similarly "mature". Freenet and Mnet are the
oldest (both began in 1999 or so, if you count Mnet's ancestor Mojo
Nation), but the newer GNUnet and Circle probably benefited from
starting with newer ideas and a fresh codebase.
Mnet v0.7 has newer, simpler ideas and a fresh codebase relative to
Mnet v0.6, which had simpler ideas and a fresh codebase relative to
Mojo Nation. :-)
I think you should try picking up one or more of these projects and
then report back about how easy or hard it was to pick up, or other
observations you have. Trying it out is probably a better way to
decide which you like than talking about it is.
There are some important architectural differences. Mnet is
(currently) strictly a "publish/download" model of shared decentralized
storage, as is Freenet. GNUnet offers the option of "file-sharing" in
which you can store your own file on your own hard drive and also make
it available to peers. I don't know about Circle. There are lots of
other differences too, of course.
> I would imagine we
> want the best one to get better and take over the world to avoid
> duplication of effort. Do any of these other P2Ps have design
> flaws like Freenet?? Which is the winner is the best design, best
> documentation category???
I rather disagree. These four projects are exploring different parts
of the design space. Any one of them is quite likely to fail (or at
least to lose a lot of time) by attempting to colonize parts of the
design space that turn out to be inhospitable. Also, these various
parts of the design space might prove to be good for some uses but bad
for other uses. Finally, inasmuch as these are all public, Free
Software and non-patented projects, they can learn from one another's
successes and failures. I say the more independent explorers the
better! Happily, there are a lot of them. I'm sure there are at least
half a dozen other similar projects which I don't even know about.
Also, open source projects that are primarily exploratory probably
don't scale up well to very many developers. For example, I perceive
that Freenet has suffered more than it has benefited from its
popularity among would-be helpers. In the entire four year history of
Mojo-Nation-then-Mnet as an open source, volunteer project, we've had
eight people who have contributed lots of work over a long period of
time, plus fourteen people who have contributed enough work to be
credited in <a href="http://mnetproject.org/repos/mnet/CREDITS"> the
CREDITS file </a>, plus uncounted hundreds of people who offered
suggestions, criticism, or encouragement.
Anyway, I'm quite confident that having an argument about which Free
Software decentralized filesystem(ish) project is best would waste a
lot of time and energy and would result in all projects progressing
more slowly rather than any project progressing faster.
Regards,
Zooko
More information about the Mnet-devel
mailing list