Post
Topic
Board Archival
Re: delete
by
YouWillBeProvenWrong
on 09/10/2014, 01:51:42 UTC
Long ago you and I discussed in public that linking together transactions if your IP conveys your identity thus breaks down the anonymity of rings. That was when you guys decided to add I2P because I informed you about that problem.

...

2. You're confusing linking and tracing. The purpose of ring signatures is to impede tracing. Linking can be avoided even in Bitcoin just by not reusing addresses (though is more convenient in Monero). Tracing can't be done at all via IP-level attacks, as far as I can tell, since the blockchain is effectively broadcast. No one can know from analyzing your network activity which transactions you are receiving, only which ones you send.

We had this same exact discussion publicly long ago, but at that time I hadn't yet conceived the bounty algorithm.

You know who the spender is of the transaction, you break them. Even though you don't know which public key corresponds to which person's identity (but you know that too if users want to use lite clients!), by linking together multiple rings spent from the same person, you can break down the rings by cross-correlation. It is similar to the bounty algorithm concept, but will converge absolutelywith higher certainty if you know every IP address. Or if you only know some of them, it will aid my bounty algorithm as more information for de-anonymizing the rings.

On chain just adds vulnerability

Okay, now it is pretty clear that you are "selling" your approach.

No it was a factual statement, not even FUD.

It is not fair to criticize me for not rushing to release some anonymity design that is not well characterized, if you simultaneously criticize DRK for rushing and releasing anonymity design that is not well characterized.

Also you know I've asked for implementation help.

Edit: the point is you can't claim (you haven't proven) rings add anything over off chain. And they apparently can't efficiently layer multiple encryption.

I do find rings more well understood than DRK mysterious "masternodes". But that doesn't mean I am saying CN rings are extremely well characterized. Overlapping rings and IP tracing can de-anonymize the rings with cross-correlation. None of this is proven affirmatively or disproven.