a new p2p structure

[alphabeater]

yeah, i know i'm new here, but this seems like a place full of people who both use and understand p2p programs, and i've been meaning to write this down for a while, instead of just leaving it forever as a loose collection of thoughts.

anyway, there are two programs that i am very fond of, and that i've been exploring a lot recently.

the first is msn messenger, which i can use to chat with friends i met back in the days of napster and managed to get contact info for before the shutdown. the second is winmx (the opennap section of it, anyway), which i have found to be a neat community of music lovers where most other p2p programs nowadays are just 'leechware' - ie. no communication between users, just a free-for-all where you can download most files you can think of.

i've thought about this a lot, and central to the problem seems to be the idea of having a verifiable and locatable identity. on msn, you have a hotmail address, problem solved, and on napster you had a username. while you have identities on networks like winmx and kazaa, they are not locatable across the entire network. the number of kazaa users who ask 'how can i find my friends on kazaa?' is amazing, and the answer (which you probably know) is that you can't.

winmx is, in my opinion, superior to kazaa, in that you can have multi-user chat communities and, as long as you are all on the same server or server network, you can find your friends. however, opennap servers are hard to set up, usually full and large targets for anyone wishing to shut them down.

gnutella and kazaa solve this by introducing the idea of a 'servent' - a combined server and client. multiple servents connect up to form a decentralised network. however, on a decentralised network, verifiable identity cannot exist almost by definition - it would require central servers of some type. while kazaa and gnutella allow you to get the files you want, it is only from a swarm of ip addresses or easily-spoofable nicknames - never a proper community.

i spotted that there could be a crossover between the network models of kazaa, gnutella, opennap and msn. to explain myself, msn and opennap both have verifiable identity lists. for msn, this is the hotmail address, but for opennap it is the domain (or subdomain and domain) of the server. i realised that the reason opennap is an easy target is the low number of servers, but that this could be addressed by making everyone into something similar to an opennap server, with their own domain - an opennap servent, if you like.

sites like no-ip.com, dtdns.com and loads of others give away (sub)domains which can be updated using a dynamic dns update client. this would mean that even non-static ips (dialup) could use the network easily.

the ideal layout for the servent, i think, would be something like msn messenger - that is, a contact list listing domains. this would be much like the ones you can download for winmx, except that you would compile it yourself, and it would be a list of your friends (or 'contacts') in the network. you would be able to send an im to any of these contacts and have a conversation in instant messenger style, and then choose the ones you like best (or the ones with the best connections, more likely), to use as 'peers', ie. connect to and join the p2p network. the network from there on could work much like gnutella, with search messages being forwarded between servents with a ttl.

this system would be a little more work for users, but would allow them to move from a decentralised, automatically-organised network (kazaa, gnutella) to a decentralised but user-organised network. this would have the effect of introducing the much-hyped concept of a 'web of trust' into a p2p network, as well as testing the 'six degrees of seperation' theory in a new way. it would also be the proper integration of im and p2p, two technologies which i feel are very close but have never been introduced properly.

ok, i'm done rambling (for now), sorry if that was all a bit of a mess but i'm having trouble articulating all the stuff that's going on in my head. if you've read this far, please tell me what you think and what your ideas are on the subject, it might help me out here...

[greedy_lars]

Originally posted by alphabeater
... idea of having a verifiable and locatable identity.

are you from the riaa?

[alphabeater]

Originally posted by greedy_lars


are you from the riaa?

...the reason opennap is an easy target (for the riaa, duh) is the low number of servers, but that this could be addressed by making everyone into something similar to an opennap server, with their own domain...

the idea, to put it another way, is that when you have a very large number of these unique, verifiable identities, it ceases to be a problem, unless the riaa is prepared to sue everyone who uses the network. also, remember that this thing isn't a music sharing tool, or even just a p2p tool - it's instant messaging too, and on top of that it's decentralised. overall, i believe this system would be just as resistant to attacks by the riaa and anyone else who wanted to see the end of it as gnutella is today, possibly more.

[TankGirl]

Welcome to Napsterites, alphabeater!

A good post with a good vision of how p2p should evolve.

Originally posted by alphabeater
i've thought about this a lot, and central to the problem seems to be the idea of having a verifiable and locatable identity. on msn, you have a hotmail address, problem solved, and on napster you had a username. while you have identities on networks like winmx and kazaa, they are not locatable across the entire network. the number of kazaa users who ask 'how can i find my friends on kazaa?' is amazing, and the answer (which you probably know) is that you can't.

Precisely. This has been the core problem of public p2p networks. Decentralized networks have lacked reliable peer identities altogether and OpenNap is not much better as it is so easy to spoof identities and you keep revealing your nick/password combinations openly to all servers you try to log in. Without permanent verifiable identities there cannot be real peer-to-peer trust not to talk about wider trust networks.

Originally posted by alphabeater
gnutella and kazaa solve this by introducing the idea of a 'servent' - a combined server and client. multiple servents connect up to form a decentralised network. however, on a decentralised network, verifiable identity cannot exist almost by definition - it would require central servers of some type. while kazaa and gnutella allow you to get the files you want, it is only from a swarm of ip addresses or easily-spoofable nicknames - never a proper community.

It is possible to have verifiable peer identities also on 100 % decentralized networks. This requires that any two peers communicating with each other for the first time establish a protected communication channel between them and exchange dedicated encryption keys that they can use in their future communications to verify each other's identity. Using encrypted 1-to-1 communications all the way they can maintain an elementary trust relationship on which trusted groups and communities can be built.

Originally posted by alphabeater
the ideal layout for the servent, i think, would be something like msn messenger - that is, a contact list listing domains. this would be much like the ones you can download for winmx, except that you would compile it yourself, and it would be a list of your friends (or 'contacts') in the network. you would be able to send an im to any of these contacts and have a conversation in instant messenger style, and then choose the ones you like best (or the ones with the best connections, more likely), to use as 'peers', ie. connect to and join the p2p network. the network from there on could work much like gnutella, with search messages being forwarded between servents with a ttl.

this system would be a little more work for users, but would allow them to move from a decentralised, automatically-organised network (kazaa, gnutella) to a decentralised but user-organised network. this would have the effect of introducing the much-hyped concept of a 'web of trust' into a p2p network, as well as testing the 'six degrees of seperation' theory in a new way. it would also be the proper integration of im and p2p, two technologies which i feel are very close but have never been introduced properly.

The idea of building permanent social networks over the constantly fluctuating 'technical' networks is a very potent one and as you say it hasn't yet really been introduced into decentralized p2p - although the idea is very much in the air. AudioGalaxy made important contributions to the idea of social self-organization but in the centralized environment. I trust that similar ideas will inevitably find their way also to the decentralized p2p, making way for even more exciting applications and networks.

- tg

[goldie]

Anyone remember the proggie that combined filesharing with the AOL IM and buddy lists?

This idea seems very similar to that..............



Sidenote: I never did get any good buddies off that system and ditched it shortly after I installed it.

Relying on anything MSN-like gives me the ebbie jeebies!!

Welcome to NU alphabeater!


<------ I'm the 'tarded one

[alphabeater]

thanks for the responses, there are some interesting ideas here...

Originally posted by TankGirl

It is possible to have verifiable peer identities also on 100 % decentralized networks. This requires that any two peers communicating with each other for the first time establish a protected communication channel between them and exchange dedicated encryption keys that they can use in their future communications to verify each other's identity. Using encrypted 1-to-1 communications all the way they can maintain an elementary trust relationship on which trusted groups and communities can be built.

this is an excellent point - i hadn't thought of encryption, although it could possibly become a little cumbersome with large numbers of people involved. for 1-to-1 communication, however, it would provide an extra layer of insurance - being especially useful if the identities were somehow removed temporarily, in that other communication methods such as email could be used to re-establish contact between the two peers.

Originally posted by TankGirl

AudioGalaxy made important contributions to the idea of social self-organization but in the centralized environment. I trust that similar ideas will inevitably find their way also to the decentralized p2p, making way for even more exciting applications and networks.

i loved audiogalaxy, and found it to be superior even to napster at its peak. audiogalaxy-style communities could start on a network of this type if some of the users were willing to run 'indexes' of the network, or individual features like forums, chatrooms and so on using the already-established identities. this would be a great way for a community to grow, and would provide true flexibility, as everyone would have the ability to do whatever they wanted to help the network without having to download and configure a seperate and far more complicated server application.

Originally posted by goldenrod

Anyone remember the proggie that combined filesharing with the AOL IM and buddy lists?

i think you're thinking of aimster (now pretty much dead and known as madster). this seemed like a good idea, but turned out to be more of a gimmick, and just decreased the program's reliability by forcing it to rely on two login servers instead of just one. i never managed to get a single successful download from aimster, and resented the clunky integration with aol im - aimster ran at startup whenever you turned on your computer, just in case you decided to use aol that day!

it's probably better in the long run to create a new im network to 'piggyback' a p2p network on, as relying on existing ones would also mean keeping all of their faults.


ps. thanks for all the welcomes, i'm liking it more here every minute!

[JackSpratts]

as a bridge between now, where we really don't have any true napster like hotlistng, and tomorrow when they'll be as common as lawsuits, a look at the mozilla style incorporation of aol-im in netscape 7 might be helpful alphabeater. i'm still not sure why no one ever created a successful "im patch" for a p2p, especially in one of the cleaned versions so popular of late. my browser opera allows me to use any of several im schemes inside the window, from aol to msn and icq. i imagine a decentralized p2p could do the same. of course by choosing one you eliminate another but your p2p name (not the same as your im screen name probably) could list what app you're using at any given time and an uploader could then act accordingly and in this way have at least some kind of hotlisting permanence.

i had lots of favorite "partners" on napster, users i miss to this day. although i keep in touch with some whose email addys i managed to get, if napster or morpheus had used aol as a comm tool (you can name your own), i'd still be sharing with dozens of them right now, long after the demise of the initiating clients.

in any event, tg knows of what she speaks. she’s given the problem so much thought if she says it’s coming i’d put money on it.

- js.

[alphabeater]

Originally posted by JackSpratts

i had lots of favorite "partners" on napster, users i miss to this day. although i keep in touch with some whose email addys i managed to get, if napster or morpheus had used aol as a comm tool (you can name your own), i'd still be sharing with dozens of them right now, long after the demise of the initiating clients.

i know the feeling.

one thing - instead of p2p clients featuring im integration, how about if a new im service could detect existing p2p and im clients running on your connection, and then give you the option to store users' information and add them to a hotlist within the new im program? if enough people could be persuaded to install the new im program and add each other as contacts, then it would become a useful tool for migration to a new, decentralised im/p2p network from existing centralised and semi-centralised ones.

[JackSpratts]

Originally posted by alphabeater

if enough people could be persuaded to install the new im program and add each other as contacts, then it would become a useful tool for migration to a new, decentralised im/p2p network.

technical conflicts notwithstanding it's true that the meta im client is a perfect prescription for the affliction of fragmentation.

- js.

[Mazer]

Wow, there's just a garden of good ideas growing in this thread. Welcome alphabeater.

I especially like the idea of a domain name for each user, not only does it make it easier to verify a user's identity but it eliminates the need for public entry points into the network. You don't need a list of peer cache servers if you alredy have a buddy list. I think with that system in place Zeropaid or someone else would host their own deidcated peer cache node for new users, which would be a huge boost for a new p2p client just being introduced.

People are going to IM each other whether their p2p provides that service or not. Any client that does incorporate IM will be that much easier to use. I liked the idea of Aimster even if I didn't use it, it's goal was to take advantage of a well established community model rather than trying to create a new, untested model. IM is intuitive, that's how p2p should be.

[petriburg]

An interesting post indeed, alphabeater, and welcome to nu.
I believe, and sincerely hope, that this thread has been/will be followed by Teclis, since he is working on 'galaxy v2' and is looking for suggestions like these.
J.S., is there any way you can bring this topic to his attention?
Good luck to the project anyway, from yet another grieving AG ex-user.

[JackSpratts]

Originally posted by petriburg

J.S., is there any way you can bring this topic to his attention?

done.

- js.

[assorted]

back in the days of opennap and audiognome's attempt at decentralization I was thinking that what was needed was what I was already using as a dynamic IP... mainly that everyone needed their own yi.org type addy for easy identification.

But that requires a central server that keeps everyones addy with username/password. Even something that barely centralized feels to me like it would be open for legal attack (because then usernames could be theoretically "banned" from the system for trading copyrighted materials).

[alphabeater]

Originally posted by assorted
Even something that barely centralized feels to me like it would be open for legal attack (because then usernames could be theoretically "banned" from the system for trading copyrighted materials).

all they'd be doing is pointing a domain to a dynamic ip.. if they could ban that then it's the end of the net as we know it, really.

"we've banned your username for linking it to an ip which runs an im service on one of its ports which could theoretically also be used to transfer files, some of which might be music files, and we may own the copyright to them...".. hmm. this is about as likely as isps shutting down people with specific ip addresses, as they are hosting large numbers of files using a gnutella servent (it could happen, sure, but it probably won't).

there are a million other non-infringing uses for such a system (more than gnutella currently has). plus, there are enough free ddns services that getting a new username would be easy (you could find your friends again using the secure email 'insurance' mentioned earlier).

remember that opennap servers have been using this system for ages, and although threats have been sent to people's isps i don't think many ddns services would shut down the domains of their (potentially paying) customers.

if the ddns was run by the same people as the im/p2p (ie. offering subdomains of their own site), then i could see a legal vulnerability, but as things stand i don't think it's much of an issue.

[napho]

This program is a cross between an IM and a P2P and more.

http://www.napsterites.net/undergrou...threadid=11933


http://www.napsterites.net/undergrou...&postid=138985

[multi]

yeah like some ideas here too guys
my idea for a wile is got to do with some thing i know very little of.SSH......(secure shell) A secure command line interface is just the beginning of the many ways SSH can be used. Given the proper amount of bandwidth, X11 sessions can be directed over an SSH channel. Or, by using TCP/IP forwarding, previously insecure port connections between systems can be mapped to specific SSH channels. You will be asked to supply the root password for the server. Then, the (insert program or process of yr choice) will appear and you can resume your downloads on the network?
Port Forwarding

With SSH you can secure otherwise insecure TCP/IP protocols via port forwarding. When using this technique, the SSH server becomes an encrypted conduit to the SSH client.

Port forwarding works by mapping a local port on the client to a remote port on the server. SSH allows you to map any port from the server to any port on the client; the port numbers do not need to match for it to work.

To create a TCP/IP port forwarding channel which listens for connections on the localhost, use the following command:

ssh -L local-port:remote-hostname:remote-port username@hostname

Note Setting up port forwarding to listen on ports below 1024 requires root access.

So if you want to check your email on a server called mail.domain.com using POP through an encrypted connection, you can use the following command:

ssh -L 1100:mail.domain.com:110 mail.domain.com

Once the port forwarding channel is in place between the two machines, you can direct your POP mail client to use port 1100 on localhost to check for new mail. Any requests sent to port 1100 on your system will be directed securely to the mail.domain.com server.

If mail.domain.com is not running an SSH server daemon, but you can log in via SSH to a machine on the same network, you can still use SSH to secure the part of the POP connection. However, a slightly different command is needed:

ssh -L 1100:mail.domain.com:110 other.domain.com

In this example, you are forwarding your POP request from port 1100 on your machine through the SSH connection on port 22 to other.domain.com. Then, other.domain.com connects to port 110 on mail.domain.com to allow you to check for new mail. Using this technique, only the connection between your system and other.domain.com is secure.

Port forwarding can also be used to get information securely through network firewalls. If the firewall is configured to allow SSH traffic via its standard port (22) but block access through other ports, a connection between two hosts using the blocked ports is still possible by redirecting their communication over an established SSH connection.

Note


Using port forwarding to forward connections in this manner allows any user on the client system to connect to the service to which you are forwarding connections. If the client system becomes compromised, the attacker will also have access to forwarded services.

System administrators concerned about port forwarding can disable this functionality on the server by specifying a No parameter for the AllowTcpForwarding line in /etc/ssh/sshd_config and restarting the sshd service.
i got no idea if and how this could be relevent to a new p2p program but some ppl may understand where or why i think iT could be used or would it be useles
just in the ideas you were saying made me think of it...

[butterfly_kisses]

Originally posted by alphabeater
yeah, i know i'm new here, but this seems like a place full of people who both use and understand p2p programs, and i've been meaning to write this down for a while, instead of just leaving it forever as a loose collection of thoughts.

anyway, there are two programs that i am very fond of, and that i've been exploring a lot recently.

the first is msn messenger, which i can use to chat with friends i met back in the days of napster and managed to get contact info for before the shutdown. the second is winmx (the opennap section of it, anyway), which i have found to be a neat community of music lovers where most other p2p programs nowadays are just 'leechware' - ie. no communication between users, just a free-for-all where you can download most files you can think of.

i've thought about this a lot, and central to the problem seems to be the idea of having a verifiable and locatable identity. on msn, you have a hotmail address, problem solved, and on napster you had a username. while you have identities on networks like winmx and kazaa, they are not locatable across the entire network. the number of kazaa users who ask 'how can i find my friends on kazaa?' is amazing, and the answer (which you probably know) is that you can't.

winmx is, in my opinion, superior to kazaa, in that you can have multi-user chat communities and, as long as you are all on the same server or server network, you can find your friends. however, opennap servers are hard to set up, usually full and large targets for anyone wishing to shut them down.

gnutella and kazaa solve this by introducing the idea of a 'servent' - a combined server and client. multiple servents connect up to form a decentralised network. however, on a decentralised network, verifiable identity cannot exist almost by definition - it would require central servers of some type. while kazaa and gnutella allow you to get the files you want, it is only from a swarm of ip addresses or easily-spoofable nicknames - never a proper community.

i spotted that there could be a crossover between the network models of kazaa, gnutella, opennap and msn. to explain myself, msn and opennap both have verifiable identity lists. for msn, this is the hotmail address, but for opennap it is the domain (or subdomain and domain) of the server. i realised that the reason opennap is an easy target is the low number of servers, but that this could be addressed by making everyone into something similar to an opennap server, with their own domain - an opennap servent, if you like.

sites like no-ip.com, dtdns.com and loads of others give away (sub)domains which can be updated using a dynamic dns update client. this would mean that even non-static ips (dialup) could use the network easily.

the ideal layout for the servent, i think, would be something like msn messenger - that is, a contact list listing domains. this would be much like the ones you can download for winmx, except that you would compile it yourself, and it would be a list of your friends (or 'contacts') in the network. you would be able to send an im to any of these contacts and have a conversation in instant messenger style, and then choose the ones you like best (or the ones with the best connections, more likely), to use as 'peers', ie. connect to and join the p2p network. the network from there on could work much like gnutella, with search messages being forwarded between servents with a ttl.

this system would be a little more work for users, but would allow them to move from a decentralised, automatically-organised network (kazaa, gnutella) to a decentralised but user-organised network. this would have the effect of introducing the much-hyped concept of a 'web of trust' into a p2p network, as well as testing the 'six degrees of seperation' theory in a new way. it would also be the proper integration of im and p2p, two technologies which i feel are very close but have never been introduced properly.

ok, i'm done rambling (for now), sorry if that was all a bit of a mess but i'm having trouble articulating all the stuff that's going on in my head. if you've read this far, please tell me what you think and what your ideas are on the subject, it might help me out here...

This was the most fucking excellent thing I have ever seen posted here in this forum!

Welcome to Napsterites.NET

please enjoy the stay.

[SA_Dave]

I was just thinking yesterday how vulnerable some of the current networks are. Gnutella's GWebCaches for example, could easily be shut down by the RIAA. Even FT clients like Kazaa, have supernodes preprogrammed into them on first install. Once you're connected though it's a different story, but if anyone had just installed the client they might experience connection problems. It would require major software upgrades & many potential users might get fed-up with the hassle.

I believe that a dynamic dns protocol would solve most of these problems. Together with encrypted communcations. In fact, I believe the entire internet will evolve to this model. Think of the servers/metanodes as a central pool or "ring" structure. These machines can see each other & know what content (specifically indexed & linked content from clients) is "located" (by dynamic linking) at which "server". Of course, the backbone would have to be incredibly secure & have multiple checks in place eg. authentication & intrusion/vulnerability detection, in order to prevent the total collapse of the network in unforseen circumstances (although it's more likely that specific machines would be targeted, much like with today's internet.) Every client connects to this central structure, or at least to one of its cascading sub-branches. The encrypted communications between clients are routed through the central backbone. You could upload copyrighted content, but couldn't tell to which ip/domain it was sent, as only the unhackable server knows. The same for downloads. Plus, if anyone hacked into the server it couldn't be legally used as evidence. However, it wouldn't be possible to implicate anyone, as this method could be used to publish websites (in Freenet p2p style), chat, e-mail, ftp, stream & anything else you can do on the internet today. Secure & flexible.

If only.

The only problem is if everything is untraceable, what about the potential problems caused by hackers? How do you effectively police a system designed to make personal security & privacy its priority?

These might be desirable features, but it'll take a lot of effort to imlement. I think that making this technology applicable to the entire internet, rather than just p2p, is the best way to get this implemented with gusto.

[twinspan]

dynamic dns redirection for nodes/servents, 1-to-1 encrypted ID verification, meta-messengers, port-forwarding...

It looks like three decades after the Pentagon started out on the path to create a completely attack-proof communications system, the RIAA and MPAA are spurring the P2P community towards making it a reality.



I love you people.

[TankGirl]

Originally posted by twinspan
dynamic dns redirection for nodes/servents, 1-to-1 encrypted ID verification, meta-messengers, port-forwarding...

It looks like three decades after the Pentagon started out on the path to create a completely attack-proof communications system, the RIAA and MPAA are spurring the P2P community towards making it a reality.


I love you people.

Ain't that the truth!

The only remaining issue is whether it takes military grade security or shall mere commercial grade security do the job...

- tg