So I guess the problem this is trying to solve is as follows:
- TITAN requires recipient to scan every transaction
- Which requires downloading all transactions (lots of bandwidth) and running crypto on every transaction (lots of CPU)
- Transaction senders upload a pair (receiver_account, txid) to a "mail server" specified in the receiver's account
- Then the client downloads txid's for their account from the "mail server," thus learning which txid's belong to them without scanning all of them.
- Then the client downloads transactions for only those txid's (either from the mail server or from peers) and runs full TITAN protocol (but only on a few txid's as opposed to all txid's in existence).
I think I have a workable protocol design which can solve this problem without the need for "mail servers." There are some trade-offs, but I think the costs are acceptable (~20 bytes for all transactions to accounts which register for the protocol, plus additional computation/bandwidth over mail server design but the exact amount is configurable, allowing user-settable tradeoff between anonymity and computation/bandwidth). I think the benefits of not requiring "mail servers" are substantial.
So far I've received a grand total of 10 BTSX for my contributions to the BitShares ecosystem (and of course my genesis block BTSX / LTS / etc. and return from trading activities, but I paid for those ^_^). Is there any possibility that publishing my decentralized solution to this problem could earn me some reward?