Re: Re: Implementing a Gnutella-structured network only with basic HTTP possible?Date: Wed, 24 Feb 2010 09:08:15 +0000 (UTC)Date: Fri, 26 Feb 2010 12:55:15 +0000 (UTC)

The tag search service is not good because it treats all files/contents as isolated ressources.
There's a lack of structuration, i.e. a content model similar to HTTP hierarchical filesystem-like system, and lack 
of a DNS like system to name services.

All we have is the possibility to browse isolated hosts, and no way to clearly bind several ressources as an unified 
service (possibly built collaboratively with many contributors), and no way to support active services (which would 
be supported by cooperating proxies, discovered each other using the P2P Kademlia, working together in a cluster to 
build a response, or transform objects, create compound documents, caching their results.

The way I perceive the P2P evolution is that it should become a true network layer, supporting a transport layer to 
decentralized clusters of collaborating hosts.

By integrating web-like services (including web 2.0 concepts) the P2P network would be more creative and more useful 
(and would contain many more legal contents). And by allowing people to work collaboratively, they can create 
various groups of people sharing some progressive level of trust, to build something that is currently only offered 
for sale by web hosting providers (or with very limited hosting on classical web sites, that strictly limit what 
type of content they accept to host.

Giben that contents on the Gnet have quite a short lifetime, it's also normal for it to be used first as a 
collaborative platfom for interactive services. Social networking on such platform would also be much safer than on 
a giant server that builds up a dangerous giant database and completely reirients what you will see for the profit 
of the most powerful advertizers.

As I said, I am very unsatisfied by the fact that job sites want now us to register on Facebook, where everyone can 
use the personal details and even add to it though many cross references. Some other companies will anyway eliminate 
candidates that reveal too much about themselves in their public profile on Facebook: even the customers can then 
look at the private like of those working for their providers. This completely contrasts with the image that 
corporations want to exhibit. But the bad thing is that people that share a lot about themselves on Facebook are 
very likely to reveal things that should have remained secret (remember that search engines are very powerful, they 
can collect insignificant elements, and then pack them into usable data that can compromize what you thought was 
secret, so if you are in a critical position in a company, such discovery by a search engine could place the 
corporation at risks, exposing for example projects, their customers, their possible partners, and loosing 
opportunities).

The same is true for the private life. Do you want that your new girlfriend discovers everything you did before 
meeting her, would you like your former girlfriend to take revenge by linking your public profile with photos of you 
or proove your "contradictions" with your new life?

The fact that P2P networks offer shortliving contents is an advantage. Everything does not have to be archived, and 
the bazaar approach is good, even if it clearly requires some organization, but an actively changing organization 
reflecting what users really want to do in some short period of time (they'll decide later if they want to continue 
the cooperation). The P2P network is then for such dynamic contents. It is a facilitator to increase the 
possibilities of contact that can be made gradually without revealing everything about you immediately at the first 
time. This means that we start completely anonymous (well, not for your ISP or for a law enforcement service, that 
can log your access or request and parse these log), and then progressively will reveal some info that will not stay 
in the network for ever, and that will be difficult to aggregate later if some pieces have been archived.

And finally, the best collaborative projects and groups will be able to progressively build things that will finally 
find a place on the traditionnal web. The P2P network could be a great place for making experimentations, trying 
concepts or refining them.

And that's why I woould see it evolve as a secondary Internet, reproducing all its core functionalites in a 
decentralized and failure-tolerant way. Searching for files by name has, in my opinion, little value (it's much 
easier to do that within existing websites, or with the free hosting space that most ISPs give to each one for their 
personnal webpages, and easier to look for these contents using existing one of the many web search engines). If it 
is used just to download contents illegally shared, it will not be better than on personal web pages.

> Message du 27/02/10 00:28
> De : "Arne Babenhauserheide" 
> A : [email protected]
> Copie à : 
> Objet : Re: [the_gdf] Re: Implementing a Gnutella-structured network only with basic HTTP possible?Date: Wed, 24 
Feb 2010 09:08:15 +0000 (UTC)Date: Fri, 26 Feb 2010 12:55:15 +0000 (UTC)
> 
> 
> Am Freitag, 26. Februar 2010 21:23:16 schrieb verdy_p:
> > The existing tag search service of Gnutella is widely deprecated. The best
> > service now is the Kademlia-based "Mojito" DHT, routed via the Gnutella
> > infrastructure, and the host discovery system (that allows the network to
> > remain fully connected, independantly of what they can transport).
> 
> My last status is that Gnutella based searching by tag (partly fuzzy) still 
> clearly outperforms kademlia, because it has no publishing cost. 
> 
> Did that change? 
> 
> Best wishes, 
> Arne
> 
> PS: That kademlia wins at searching by hash is clear - that's what it's 
> designed for. 
> 
> 
> [Non-text portions of this message have been removed]
> 
> 
> 
> ------------------------------------
> 
> Yahoo! Groups Links
> 
> 
> 
> 
>