The addresses is a non-issue. There is no connection between different transactions just because they have the same public address (unlinkability).
Sorry if I'm misunderstanding or asking the question wrong. I was trying to ask if one could use their private key to identify an output of theirs as a mixin in a different transation. So, if an exchange were to begin using mixins, then their ability to identify they one of their outputs was used in someone else's ring would increase proportionally to the volume they handled?
The issue of someone controlling a huge share of transactions is a real one, and amounts to a form of sybil attack on traceability. Some transactions will become traceable in this manner but transactions with high mixes or multiple hops become effectively untraceable even by someone owning 90% or more of the outputs due to the exponential function.
Oh I see, maybe. So, if a typical user were to use a mixin of ten, then their chances of having someone with 90% of the transactions identify one of them would be .9^10, or 34.8% of being de-anonymized? I think there's also the denominations to deal with as well, so let's say for this example that all transactions ever done were just '10'?
Also, just as a matter of general privacy, exchange transactions are easily identifiable since they have a payment ID, and many people don't change their payment ID very often, so you can find all their transactions that way. Given this, we can tell that the number of exchange transactions is high but not extremely high. There are still a lot of mining transactions, pool payouts, and other incidental stuff (donations, MEW memberships, private trades, people moving between their own wallets, etc.)
Is this why you initially advocated for the txid field to be removed completely? If the field was filled and encrypted on a protocol standard with random data when not in use, then would nobody know but one party (here, exchange) how much of the transactions were theirs? So, if someone were trying to mount a sybil attack on mixins, nobody would ever see it coming? Would removing it prevent the ability of an exchange or anyone to be able to know that a transaction was theirs? Maybe there's more to this? Would the best case to be to encourage the widespread random usage of the txid field then?
As a second practical matter, I don't think (most?) exchange transactions today even use mixing, so even the sybil attack doesn't apply. That could change the future, although one would hope that there would also be other uses besides sending coins to an exchange constantly, otherwise who even cares.
Me! Regardless of the eventual level of penetration into society these coins get, pretty much every aspect of them is still fascinating on a technical level
Finally it is important to remember that unlinkability (stealth) and untracabilty (mixing) work together to frustrate blockchain analysis. Even when you can partially overcome one, the other often makes the results useless. So for example, if you can defeat untracability on some tranasctions, you just get links between anonymous one-time keypairs that don't identify a person or link with other payments to or from that same person. Conversely if you can link some keypairs together, you can construct an "identify" (still not necessarily linked to a person) but you can't see flow of funds to or from that identify without also defeating untraceability. You really have to defeat both simultaneously on the same set of transactions to get anything useful, and that is much harder.
Right, so you can have the framework built up of transactions linked by one address, but still not have a clue where the terminations are because stealth addressing. Really, all you can do is prove that you were the majority transactor on the network, which you'd likely already know. So, if there were two major major network transactors (likely scenario at this point in time - two hi volume exchanges) then could there be any cross-referencing done in which some of those terminations could be revealed? Let's say cryptsy picks up xmr, and matches plx in volume and that's all that really changes in the next couple of months. Can the exchanges collude to identify users within a high probability on one exchange as users of the other exchange? I realize this can obviously be done without the blockchain, but I'm trying to learn so if anyone's under the impression that I think this would ever happen then I'm sorry but I don't mean it as it's just the easiest thing I can use to identify and understand how this works.
There was another thing that I was wondering: If ~100% of the coins that are mined go directly from pool to exchange, and then onto the rest of their life, can that cause a long-lasting effect on anything? I understand that this is not the case currently where all mined coins go to one address, but if mining were to become strictly a business where mostly nothing (<1%) is held, and 99-100% of coins always originated from <10 addresses (pools) to be used by a group of 10,000 people, as opposed to a very small network of 10,000 people who are just solo-mining and using the coins, then which network would have greater cryptographic anonymity using cryptonote?
