On Thu, Jul 13, 2006 at 04:14:32PM +0200, Lean Fuglsang wrote: > I still think that opennet can be more secure than darknet.
It can't. Sybil attacks are the obvious thing: One node, with a lot of bandwidth, pretends to be several thousand nodes, harvests the network, always resets the data source to another one of its fake nodes, and connects to every node on the network. There's nothing we can do to prevent this. And once it's connected to every node on the network, it can do a lot. Second problem: Premix routing is very unlikely to work safely on opennet, partly because of the above. Even if a way to do it could be found, it would cost a very large amount of work, separate from the significant work involved in implementing it for darknet. I am also doubtful that securing the location swapping algorithm as currently planned could work. Third problem: Load balancing. Load balancing is much harder on opennet due to having to give newbie nodes some slack, when they may just be some malicious node pretending to be a new node in order to exploit you. > As long as it > is legal (or you do not care if the world knows you run a node) to run a > opennet node. And how long do you think that will be? > The reason is that content allways comes from nodes. So if content exist > on the net, it means that somehow you are connected to nodes which are not > part of a sybil attack. You can even put data into the network from > different locations, and see if the data shows up in the network. > When you know that other people is on the network, it should be possible > to route a way which real node is looking for what data. Eh? > > The trick is that not everybody has to probe the network to avoid sybil > attack. If just some people do it, the attacker will have increasing doubt > of knowing which node is probing and which one is not. So the attacker can > not attack at all. People who do not use probing can know this, and do not any > probing at all and still be secure. I don't see what you are proposing. An attacker who can connect to the entire network is unlikely to attempt to DoS it; he will instead attempt to trace the origins of data, and with some considerable success. Much better to give people a false sense of security and then knife them in the back (not literally) when they're not expecting it. > > I have thought a little about this, but have not come to a conclusion. But > my gut says it can be proved. So just think loud and hard, and maybe we > can show if it is true or not ;) So prove it. > > The main point is that my friends is one of my biggest ressources, and > there is no way in hell I'm going to expose them by connecting directly to > them over the Internet (ip). Face reality. The police, let alone the spooks, can already read your email headers without a (judicial) warrant. This information is retained for a year. They know who your friends are; at least, they know who you exchange emails with, they know who you phone on VoIP, your cellphone or your landline (and where you were when you made the call!). They don't know how much you trust whom, but they DO know your basic acquaintances. But the bottom line for me is that opennet is not viable in an even remotely hostile regime. This means it is not very interesting to me because most of the places where freenet would be most useful are hostile regimes. Further, it is inevitable that if freenet gets big it will be banned in the west, even if it isn't blocked by some stupid misguided broad legislation such as IPRED2 first! Now, I accept that connecting to your friends over the public IP internet is not great; we need stego, we need support for slow transports such as exchange of boxes. But darknet is a big step forwards. Stego and especially whacky stego (wifi links, exchange of disk boxes, PDAs with short range wifi, etc etc) is pointless or even impossible without darknet. And ANY regime with a national firewall and with a legal basis for blocking freenet (BOTH of which are coming to a western european country near you VERY soon) can block opennet. > > --Lean -- Matthew J Toseland - toad at amphibian.dyndns.org Freenet Project Official Codemonkey - http://freenetproject.org/ ICTHUS - Nothing is impossible. Our Boss says so. -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 189 bytes Desc: Digital signature URL: <https://emu.freenetproject.org/pipermail/devl/attachments/20060713/bef21dff/attachment.pgp>
