Simple way to implement JM for bitcoin websites

[chris-belcher]

On walletexplorer.com look at localbitcoins. Most transactions are like this one:

https://www.walletexplorer.com/wallet/a8ef21ebe191b165

The main localbitcoins wallet closure (that people deposit into) pays out to a new address, some time later this new address which is a new wallet closure does a sendmany paying to everybody.

JoinMarket sendpayment could do a coinjoin to the new wallet closure. It would collect up many different UTXOs as inputs so sites like walletexplorer would not be able to find all the addresses belonging to localbitcoins.

Of course looking at one of these coinjoins it would be obvious which is the localbitcoins output because it's the one that goes onto do a sendmany, but even so this partial solution can benefit the website a lot. It's cash flow is no longer trivially found online. With some rewriting, tools can still get some information. But this simple addition can really help in the constant arms race.

[chris-belcher]

Analysis like this should not be possible

https://www.reddit.com/r/BitcoinMarkets/comments/3qnrwh/daily_discussion_thursday_october_29_2015/cwgycyz?context=2

Trading in financial markets is ultimately about secrecy and taking advantage of information, which is being ruined here.

There was another post (I may try to find it) of a trader who looks for large transactions going to known exchange addresses. After 3 confirmations there will be a sell order, so the blockchain analyzing trader just has to adjust their offers accordingly to front-run the seller. Note this can easily be beaten by generating a new address each time (although bitstamp only allows a new address to be generated every 24 hours)

[chris-belcher]

With using this method it's still possible to find a closure within a coinjoin by looking at the input amounts because they're not all equal. But with many inputs it can become a difficult knapsack problem(?) to solve so it can still provide a lot of protection. I'd recommend doing coinjoins with many other makers (-N 10 maybe)