Now that RingCT is on and transactions are many times bigger, would a mixin of 1, for example, be safe enough to keep a good level of privacy? I'm mainly asking because many pool operators have decided to opt out of RingCT (and usually have a mixin of 3). Some of them are loosing money on each block found.
Lowering mixin lowers tx fees, so I'm trying to know if it's a better solution than opting out of RingCT. Thanks!
As in my comment:
The transaction size is mainly determined by the number of outputs. The number of inputs and mixin have smaller impact on the tx size. To be more precise, each output comes with a range proof using Borromean ring signature containing 64 pubkeys (representing commitments of all digits) and 2*64+1 scalars. This amounts to 193 instances of 32byte data, i.e. 6kB per output.
Smaller mixin won't help in reducing the signature size. For example, see these two transactions with mixin 6 and 2, respectively:
ae199fd5a6869be3d3b0de5334bba3d33c8579ffaae8ce8add3b05fa82a84b0c
ba734dcdb78a98c1fb369252c5dd3ce388cedd1bf1c432d5a70705de8fbc992d
The fees are exactly the same.
Also note that the protocol required minimum mixin is 2; mixin 1 is not allowed.
