Re: [GNUnet-developers] Miscellaneous Ideas

[Igor Wronsky]

On Sun, 28 Mar 2004, Ian Clarke wrote:

> > On a gut feeling, if we gave +/- or nothing for efficiency,
> > I'd say it would go something like
> > +bittorrent
> > +edonkey variants
> > (the rest somewhere in the middle)
> > -gnunet
> > -freenet
> > For example, if someone is wishing to share some of their
> > own products with only low bandwidth available, BitTorrent
> > is _definitely_ the choice.
> That depends on your definition of "efficiency", but for most reasonable
> definitions I can think of, I think you have it backwards.

Ok, to do you the due courtesy, and in the light of

"I gave some concrete benefits of GNUnet and Freenet over
 BitTorrent, to which you have failed to respond."

lets analyze your message again.

> For example, with Freenet, when you publish content, you only need to
> upload the file you want to publish once (together with some check
> blocks for FEC).  With BitTorrent, you must distribute it from a central
> server, and it might be downloaded many times over from that server
> until there are enough people out there sharing the file.  In most
> cases, BitTorrent will require orders of magnitude more of the
> publisher's upstream bandwidth than Freenet would.

So which part of this is a concrete benefit of e.g. GNUnet
over BT? None, because there is no such thing in GNUnet as
non-local insert. You have to stick around and contribute
bandwidth, and you'll never know when is enough, when you
can leave. The Freenet part I already acknowledged by saying
that you can't generally rely on pushed content staying
in the net.

> Personally, I don't find BitTorrent to be a particularly interesting
> technology.  It is even more centralized than Napster, offers no
> anonymity, and publishers must set up and run BitTorrent trackers for
> the entire duration of their content's availability.

These also I did acknowledge by simply making the statement
that _supposing_ the user does not need anonymity
or censorship resistance (for many, this equals no need
to avoid centralization), then how do the things stand?

Even at the risk of again merely restating my subjective
opinion, I'd like to go through that 'most reasonable
definitions that you can think of' part again, so that
after this, you could think of one more - the one that
apparently very many people think of these days.

Suppose Jack wants to share a big file. Say, a 600 megabyte
file. He has 5kbps upstream that he can sacrifice 24h
for this purpose. Suppose there are k different Charlies
each day, some drop off while new ones arrive, and
they all jolly well wish to download the file, and
eventually get off.

Now would Jack go for Freenet or GNUnet? I don't think
he should. He should be using BT, because that can pull it
off. People using BT successfully to download and provide
large-size content with small resources proves the point
empirically. If you are not aware that this has been
happening, or where, you've been sleeping at the wheel.

So I don't think the concrete benefits you mentioned
are much else than subjectivity in values. Supposing
a user with different values, he might want to know
what system he should use to get most bone for his
pennies. And thats what should be in the FAQ as well,
if we see it as our business to start doing some
comparative listings.

If you need to call it p2p, maybe BT is not that. If you
call it file-sharing, it is definitely that, and works
surprisingly well for it.


Igor