Re: Gnutella 3

Quoting Max <[email protected]> from ml.gnutella.dev-forum:
:Hi Arne, that is a good idea,
:i just recommend to not only look about pro and cons for the state of art,
:but to look about the needs as well.
:So a higher evolution model should be the reference, and we should search as
:well for the off-gnutella 3 for pro and cons.

I agree with Arne that it's good to brainstorm. However we must refrain from
keeping an open mind and try to sell a preconceived idea under the disguise
of desirable properties or requirements.

:One pro would be:
:+ able to upload a seed and then go ofline
:+ dht environment
:+ torrent like speed
:+ seperation of blocks and urls-meaning, which makes it un-sueable
:+ no IP adress attached along the off-magnet downloading, you do not see
:other swarming peer IPs
:+ offload client (as example client we could work on) has community feature
:integrated over retroshare/retromessenger wx group f2f panels, for web of
:trust
:+ bootstrapping over off-urls and IP cache

For instance, if you allow me to comment on the points above:

"DHT environment" means nothing.  The need for a DHT should come from the
adequation between the features of the DHT and the algorithm needs (efficient
distributed lookup of items by key).

"seperation of blocks and urls-meaning, which makes it un-sueable" is an
instance of you trying to sell a preconceived idea.  It's also very
implementation-specific.  Is your need "anonymous file distribution"?  Or is
your need "ability to not be sued for distribution of illegal content"? Or ...?

"bootstrapping over off-urls and IP cache" again wrongly phrased "pro".

"torrent like speed" is good.  Remember however that bittorrent is actually
less powerful than today's Gnutella swarming features.

Etc...

:Contra:
:- Off-url has currently only the filename and hash in the url integrated
:(maybe extendable off urls can be developed easilsy)
:- once publsihed blocks cannot be taken down again (which is the same as in
:gntuella seeded files)
:- f2f needs to be implemented in the offload gui

I see nothing in your "contra" points which is applicable to the Gnutella 3
discussion.  You're criticizing a system which exists and is completely
alien to Gnutella 1, which is what we're trying to criticize.

By comparison, Arne's points are very good (in that they are relevant to
the discussion):

:> + Strength: Requires very little bandwidth for global connectivity (a leaf
:> can
:> run off a 56k Modem, an UP off a ISDN line).
:>
:> + Strength: Powerful fully decentral swarming.
:>
:> - Weakness: No included community building (like torrent sites with
:> statistcs
:> and verification).

Cheers,
Raphael