Thanks for writing this. I think it is an important point that the safety of "mixing" schemes (including ringsigs) depends on the nature of the "decoy" inputs. I'd like to see a more rigorous analysis of what exactly the requirements are for safety of schemes like the one Monero currently uses, under adversarial conditions.

If someone wanted to gain more ability to break the privacy of future Monero transactions, the way to do it would be to generate lots of dummy transactions between now and then so that the future Monero transactions would accidentally choose your transactions as decoys, thinking that they were getting privacy that way, but actually since those are your transactions, it doesn't provide any privacy from you. Right?