Hacker News Reader

Comment by woah | Hacker News Reader

Any 'SocialCoin' scheme pales in significance to the prospect of a decentralized network that can prove individuality, which the author brushes off with a 'oh someone will figure it out'. This kind of technology would have huge, far ranging impacts, much greater than any impact that Bitcoin, Ethereum, or even Dogecoin could ever hope to aspire to.

If you have proof of individuality, proof of work goes out the window, proof of stake goes out the window. The entire thing that makes Bitcoin and any decentralized network hard goes out the window. Proof of human individuality in a decentralized network would completely rewire society and computer science in ways we can't imagine.

This SocialCoin thing is rooted in the world we have today, but the idea is predicated on a technology that would make it irrelevant.

View on HN

I'd be interested in reading about any potential research in this area, it's quite a fascinating unsolved problem. Like Bitcoin's solution to the P2P untrusted consensus problem (Two Generals Problem), I would think that the solution to ensuring individuality (see the Sybil attack [2]) lies in some approach that isn't 100% failsafe but probabistically ensures the desired result is highly likely (see 51% attack).

I think a web of trust [3] is something that could work. Given a system of nodes with public keys (the most common form of identification in P2P), a possible approach to estimate the 'individuality' of some identity/key would be to measure its degree of relationships with other nodes who vouch for its authenticity, weighted according to the 'individuality' of the other nodes (something that could be solved with an iterative approach, as in PageRank/EigenTrust). The idea being that while a malicious node might vouch for many of its identities, it doesn't have any link to any other node and thus would be weighted lower.

Turns out this exact approach had been developed in 2011 [5] (just found out now, wow, see fig.1 on page 3) and also in other research with SybilGuard [4]. I'm not sure of the limitations of these formalised definitions, but it looks to me like much of P2P research (see PolderCast, a marvellous innovation) -- the possibilities are never realised until someone implements it in software.

[1] https://en.wikipedia.org/wiki/Two_Generals%27_Problem

[2] http://en.wikipedia.org/wiki/Sybil_attack

[3] http://en.wikipedia.org/wiki/Web_of_trust

[4] http://www.math.cmu.edu/~adf/research/SybilGuard.pdf

[5] https://ccl.northwestern.edu/papers/2011/kurve.pdf

Any known identity system can be gamed, including web of trust.

Proof of work is the only known viable sybil attack prevention mechanism.

Yeah, I stopped reading there too. Sybil attacks [1] are the biggest problem with decentralized distributed systems, you can't just hand-wave them away.

1. http://en.wikipedia.org/wiki/Sybil_attack

Could you elaborate on some of the "completely rewire", &c?

We already have proof of individuality. The post mentions drivers licenses or official government documentation, but you could also use phone numbers or credit cards or something like that (although it would allow some fraud.)

Now why isn't it magically "rewiring society and computer science"?

For example, the spam problem would be solved.

Facebook and Google use phone numbers for proof of individuality. However, you can still buy accounts for spamming etc. Apparently, it does not work.

It might work well enough for social networks, because one account is not worth that much. However, if one account gets guaranteed basic income the incentives are much larger.

Defeating spam requires changing standards and getting lots of different entities to cooperate. Google doesn't want to spend a bunch of money invasively verifying everyone's identity when their phone system and spam filters works fine. But it is theoretically possible.

hey lets talk. do you live in SF? Can i have your email? -siraj

Reply on news.ycombinator.com