31 ms·
Consensus identity algorithm?
I've got a side-project I've been kicking around for a while. I think I've gotten sufficiently far to start some basic implementation EXCEPT one issue I'm still not sure about.
Are there algorithms out there that could be considered (what I'm calling) "consensus identity algorithms"? In other words, I don't want a centralized (or even semi-centralized) authority in charge of establishing the identity of an actor within the network. Such as the CA within SSL. I
So I'm comfortable with the idea that digital signatures will get me pretty close to the holy grail, but the kicker is I want those actors to be able to build consensus in such a way so as to prevent (for example) any single actor within the network from creating multiple identities. Either that or make it statistically improbable (impossible?) for such an actor from doing so. To put real-world context, imagine a social network where it's improbable that a single person could create multiple identities without it being detected by the network.
Now, from the looks of it I've almost convinced myself that this is impossible, so unless someone has the info, I'm totally comfortable entertaining ideas that reduce probability of any single actor within the network from creating multiple identities as opposed to a mathematically pure solution that strictly prevents the possibility.
- cjbprime 12y agoI don't understand how you would even make it improbable. It sounds both very possible and probable.
- datashovel 12y agoI'm not sure I understand what you're saying. What sort of solution would you propose?
- wmf 12y agoOther than proof of work I don't know what you can do here.
- SEJeff 12y agocough botnet. I used to work for Ticketmaster and proof of work is one of the layers they use to prevent abuse. It doesn't stop brokers who sell their goods on stubhub, it only slows them down, and makes them spend a lot more CPU time doing it.
- datashovel 12y agoThat sounds like an interesting direction to go. I had thought of something like that. My take was (paraphrased) if you make it somewhat expensive (ie. proof of work) to get into the network it makes it less likely people would go to the trouble. One direction I've been pushing toward is (ignore the impractical nature of the solution for a moment) what if folks within the network were somehow invested in the integrity of the system somehow. If there were incentive not just to participate in the network but to ensure all actors they interact with are real people. So as folks "confirm" one another by signing digital signatures (or something along these lines) you build up a crowd-sourced effort to ensure the identities of other actors within the network. Put all the identities / confirmation data into a blockchain and all of a sudden heuristics and statistical analysis can be used to begin dissecting the data to tell members of the network the probability that a given actor within the network is real or not. In more practical terms, the idea is along the lines of some analysis I've seen in the past about Twitter accounts, and how a huge percentage of fake accounts are only interconnected among themselves, and not interconnected among real people. It's a very distinct characteristic that they've been able to identify about those fake accounts through data mining.
- SEJeff 12y agoLike a PGP web of trust?
- datashovel 12y agoRight. I think so. Although I'm not that familiar. I will have a closer look. The differences seem to be the idea of incorporating a blockchain. Also adding a layer of analytics to help the network establish the probability that the actor is identified within the network with only a single identity. Also my original concept is less about establishing the connections with only people you trust. One idea I had was that if you're only confirmed by people you trust, it may not be as easy for the network to detect fraud (ie. multiple identities) as compared to a network that requires you to confirm your identity across some random subset of the entire network. If you're forced to confirm your identity across the entire network (or some random subset) my intuition tells me it would be less likely that someone could circumvent the rule that you're only allowed to have one identity on the network. Thanks for the info.