Quote from: TPTB_need_war on August 24, 2015, 03:23:58 AM

Of course that is the point of spreading your trust around to as many nodes as possible and target isolating nodes with a Sybil attack. You can also isolate nodes by identifying the patterns that make some nodes more important to isolate than others, so a lot of nodes are isolated by isolating a fewer nodes. No P2P network is perfectly distributed.

Again I don't know what the ratios will end up being between T and 51% - T. A proper modeling has to be built.

And I doubt that is the only attack. I haven't expended more than 5 minutes yet thinking about how I would attack it. And I don't even know all the fine design points so I can try to identity more vulnerabilities.

Also it is not clear if your economic model isn't gameable. I haven't delved in that.

For example, nodes might even pay other nodes to accept them as their peers. Remember Vitalik's point that consensus is an undersupplied public good.

Quote from: Vitalik Buterin @ Ethereum

Unfortunately, altruism-prime cannot be relied on exclusively, because the value of coins arising from protocol integrity is a public good and will thus be undersupplied (eg. if there are 1000 stakeholders, and each of their activity has a 1% chance of being pivotal in contributing to a successful attack that will knock coin value down to zero, then each stakeholder will accept a bribe equal to only 1% of their holdings).

In summary, I don't want to comment more other than to say it is a complex model and I would want extensive peer review.

That doesn't really answer my question in any manner at all.

I want to know how you isolate a node A, that is honest and free to connect to any node that it wants, from either connecting to that node, or sending it information.

Lets say that A is an important node, lets also say that it has a slightly above average trust weight. It has many connections incoming to it, and many outgoing from it that it makes of its own. Assume that is also has a list of preferred outgoing connections to nodes it knows are of similar importance to it. If you can isolate it then it improves your chances of success slightly.

How do you isolate it so that it can not receive or relay information to anyone without being physically near it?

If you fill up its inbound connections so that regular clients cannot connect to it, it can still make outbound connections to any other SN nodes like itself. If/when that fails (because you are filling them up too) it makes connections to regular non-SN nodes instead. All the other important nodes in the network are doing the same thing, and by doing so, they have a route to each other (albeit it a bit slower). The regular clients become the service providers for the SNs, by providing a route for data between them.

The network topology is such that for every 100-200 connected users of the network, on average you need at least 1 SN available. With 10000 users that is ~50-100 SNs, with 100000 that is ~500-1000 SNs, 1M you are 5k-10k SNs. With a large number of SNs comes a good distribution of trust, so there wont be just a handful of SNs you need to isolate. You'll need to isolate a lot of them (1000s as the network grows), and as all of these SNs will have inbound connection pools of 100-200 in size, you'll need to fill up 10,000s of connection slots in total.