Sybil Attack

This sybil attack thing bothered me, so I did some more research on it. For methods like coinjoin where you are mixing your funds with others, then clearly who you get matched up with is quite important. If the attacker is having a large percentage of all nodes, then the would be able to either prevent the transactions or glean info about the spendings

With Teleport, this is not very applicable as it is a sending from source to destination in an intact telepod. Any tampering with it would require cracking layered encryption through all the onion layers and such tampering would be quickly detected on the destination side.

When the M of N fragmenting is used, there is a random chance to forward or not forward, so if an attacker is trying to prevent the tx by not forwarding, this just increases the failure rate and presumably the M parameter will need to be adjusted to compensate. I am defaulting to M = 40% of N and so there is some headroom anyway, but this is something to keep note of as if the attacker can control 20% of all the nodes, then it would be evidenced by a higher than expected failure rate to receive enough packets.

There is no need for the attacker to make fake identities if they just want to be hosting a lot of privacyServer nodes. these nodes would then be able to determine who is sending what packet to whom. However, not knowing if this packet originated from the sender, nor if the next hop is the final destination, I do not believe there is much to be gained by this "sybil" attack. The routing can be blocked, but then the nodes would route around what looks like a malfunctioning privacyServer.

Maybe the fact that there has not been a followup post means that I actually wasnt an idiot about this

James

Source: https://bitcointalk.org/index.php?topic=762346.msg8852892#msg8852892