fully anonymous p2p system idea

[spanx0r]

I had an idea of a p2p network which hides the IP addresses of participants. It would involve a double proxy scheme:
A---pA---pB---B

where A is sharing, B is receiving, and pA and pB are proxys (other p2p participants) chosen by A and B respectivly.

The idea is that since A chooses pA and B chooses pB, each can trust their proxy not to betray them. It doesn't matter if B is a lawyer and chooses pB to be his co-lawyer, because pA will still protect A. Similarly if A is a lawyer sharing files (or is that entrapment? I wouldn't put it past them) pB will hide B's identity.

I'm stuck though on creating some kind of web of trust in the system. Some kind of "distributed-kudos" system is needed (probably involving public key crypto). A needs to make sure that B hasn't sent a request packet with time-to-live of 1 straight to A, because then if pA sends pack the offer, B will know that it was A. I'm thinking some kind of history of hops in the request packet; A checks out some of the hops through linked trusts back to A. If the hops check out, A can be sure that it is not the only node to receive the request, so it can safely reply through pA.

One idea: every node has an ID which is not at all ascociated with it's IP (it's an alias). So when B downloads off A illegaly, A gives B some "kudos", a referance, which others can request in order to trust B.

Second problem. I'm worried about the overhead of such a system. In a couple of years everyone should be broadband and pings will be better, will this be feasable?

[method]

This is very true... Filetopia is attempting something similar with 'bouncers', basically like bridging nodes..

I've already heavily considered the implementation of this in applications... There's a reason I will include it and a reason I don't think anyone will want to use it...

Why I'd implement it:
It adds room for doubt, with reasonable doubt.. legal cases are a lot harder to pin down

Why it won't get used:
Let's assume one of those central nodes is a dial-up user, A and B are both 1mbit users. See the problem. Let's assume the central nodes are standard cable users.. 128kbps upstream probably..

(512kbps/us)-(128kbps/us)-(128kbps/us)-(512up)

I doubt the central nodes would want 15KB/s up and down by acting as anonymizer nodes for the 2 faster users, even though all involved are on cable/dsl.

These ideas for anonymous transfers will only really be used when broadband has taken hold properly. It's bad in that it'll not REALLY be used for a few years due to viability but it's good in that you can use the anonymous-transfer/bouncers/etc excuse to deny being intentionally and knowingly involved in distribution of copyrighted materials.

[spanx0r]

true, I can't really see a system like this working until everyone is 256kbs or higher. Is it really worth it if the system is only half implemented? I'll increase anonymity slightly at a high cost to speed.

I'm thinking of a system which implements total anonymity, with ISP logs not even being of any use. Granted I havn't figured it out yet :) but if it is possible, the benefits greatly outweigh the speed costs. I dunno, maybe I'm just madly obsessed with encrypted total security...

The key to p2p is not security. but decentralized with mass users. they cant sue us all. we are their customer base. its not a good business move :)

One of my concerns is that suing/nasty lettering random people will create a legitimate excuse for leeching. "sharing is too risky" sounds fair enough, especially when your university says it'll disconnect you from the network because it doesn't want to get its hands dirty. How long before lawyers find a way to threaten ISPs unless they ban a subscriber? They don't have to go to the effort of suing; threats alone can do damage.

(Of course ISPs will never be able to have an anti-p2p policy, it too would be a bad business move.)

I reckon the demand for an anonymous p2p system will increase because sharing in safety will become more of an issue.