On Mon, Nov 5, 2012 at 2:17 PM, Michael Hrenka wrote:

Sounds good for short-time testing, but a permanent instance might be a bit expensive when compared to virtual servers.

In the EC2 it's likely that the bandwidth used will be the primary driver of cost and not CPU power. Here is an article I've written that covers that ground: https://drwho.virtadpt.net/archive/2012/03/11/tor-in-the-elastic-computing-c... In short, as far as the EC2 is concerned bandwidth is bandwidth, regardless of what it's used for.

Yeah, but this opens up the danger that someone finds out your ID and can see all your transactions. I guess it's hard to avoid that danger in a reputation economy.

Depends on what your ID is. For example, this is a Bitcoin public address (used to send Bitcoins to whomever has the matching private key): 1Q7YCdDhzjxHf1CG2g4Akbq8EADPSzVYjj However, for a social network a Bitcoin-like Id will never work thanks to Zooko's Triangle (https://en.wikipedia.org/wiki/Zooko%27s_triangle), which means that human readable IDs will be necessary. So... practice good OPSEC if you don't want people to know who a pseudonym corresponds to. As for reputation economies, reputation is attached to an identity (pseudonym or otherwise) - as long as the entity presenting that identity is doing good business with that identity, it doesn't really matter. There are cypherpunks and hackers out there who never came in from the cold (i.e., dropped their pseudonyms) and have a significant amount of social credit attached to their handles. Sybil prevention techniques based on the connectivity characteristics of

social graphs can also limit the extent of damage that can be caused by a given sybil attacker while preserving anonymity, though these techniques cannot prevent sybil attacks entirely, and may be vulnerable to widespread small-scale sybil attacks. Examples of such prevention techniques are SybilGuard and the Advogato Trust Metrichttps://en.wikipedia.org/wiki/Advogato#Trust_metric .[4] https://en.wikipedia.org/wiki/Sybil_attack#cite_note-3

The part of the Wikipedia article you linked to describes a failure mode of the Advogato Trust Metric, where someone claiming credentials that were not earned managed to beat the system. Perhaps Advogato isn't the right solution.

My Esteem Trust System uses short-range connectivity characteristics of the "Esteem Network". I'll need to take a look at other approaches to see whether they have significant advantages.

I'll reread your paper and see what else I can work out.

That sounds like a useful idea. I guess implementing such a scheme would be quite useful in some situations. You could set up an account while several present and already trusted users verify the legitimacy of that newly created account. This nicely complements the mechanisms that I've come up with already.

There are a few problems with such a scheme, most of them having to do with the amount of work a person has to do, as opposed to the amount of work that the software has to do. The "core group" (the folks who were there when something was done) have to document what was done; people generally put off writing documentation until the last minute when they absolutely have to. So, people have to be comfortable with writing up what was done and by whom, and applying signatures to it prior to posting. The software can hopefully automate the digital signature bit, but the people have to do the writing. Farther out from the "core group of foo", everyone who's one hop away has to be prompted with a yay-or-nay to sign off on what happened. It's easy (and common) to just click "Ok" and get on with things, but that also means that people can potentially sign off on things that haven't actually happened. Because verification of a writeup has to be done, laziness can torpedo the system.

My approach is to handle most of the Trust stuff by the automated and implicit Esteem Trust System. Everything else is a bonus. I guess I will have to make the Trust System so flexible that you can easily balance secure ID verification with user friendliness depending on how paranoid your targeted users are.

That would be a good thing.

Where can I find these papers? A quick Google search didn't produce a lot of relevant results.

I'm doing some searching online and in my personal archive and I've found references in Timothy C. May's _Crypto-Anarchist Manifesto_, Wavyhill and Andre Goldman's DefCon 12 talk _Toward A Private Digital Economy_. I have a few more papers put away but I'll have to go through the last... 200 or so hits in my search manually. You may also find the paper _Distributed Marketplaces Using P2P Networks and Public Key Cryptography_ by Signorini, Gulli, and Segre interesting. I can send you a copy if you like. The paper _Implementation of P2P Reputation Management Using Distributed Identities and Decentralized Recommendation Chains_ by Satheesh and Naresh helpful. _Formalizing and Securing Relationships on Public Networks_ by Nick Szabo might be helpful (formalize.txt). _The Cyphernomicon_, section 19.4.80 was where I first heard of reputation networks. Section 16.7 also has relevant information. Reed S. Abbott's thesis _Closed Pseudonymous Groups_ will no doubt be of interest. If anyone wants copies of the documents I've found, I'll be happy to post them. Thanks for your ideas! I appreciate them. Do you want to be mentioned in

the next version of the QP documentation?

If I was able to help somehow, I wouldn't mind being mentioned. -- The Doctor [412/724/301/703] [ZS (MED)] https://drwho.virtadpt.net/ "I am everywhere." -- -- Zero State mailing list: http://groups.google.com/group/DoctrineZero ----- End forwarded message ----- -- Eugen* Leitl <a href="http://leitl.org">leitl</a> http://leitl.org ______________________________________________________________ ICBM: 48.07100, 11.36820 http://www.ativel.com http://postbiota.org 8B29F6BE: 099D 78BA 2FD3 B014 B08A 7779 75B0 2443 8B29 F6BE