You're not factoring in denominated units (and subsequent rounding at send) and "dead change" being sent to the network to remove linkages in future tx's.
It was a simplified example explaining transactions in general. Remember: the issue we're discussing is the average user deanonymising themselves through inadvertently. Dead change and denominated units do not solve the problem when the user has 50 DRK in their account, they send 20.72368 DRK to pay for some dodgy item, and then because they have some crisis they empty their wallet and deposit the entire remaining 29.27632 DRK on an exchange. Normal actions resulting in unavoidable and unwitting deanonymisation.
Assuming the wallet is already denominated, then in your scenario, the DS inputs would be two 10s, nine 1s and three .10s with the rounded up change going to the network... yes, it very much solves the issue you are trying to point out.
You're not factoring in denominated units (and subsequent rounding at send) and "dead change" being sent to the network to remove linkages in future tx's.
It was a simplified example explaining transactions in general. Remember: the issue we're discussing is the average user deanonymising themselves through inadvertently. Dead change and denominated units do not solve the problem when the user has 50 DRK in their account, they send 20.72368 DRK to pay for some dodgy item, and then because they have some crisis they empty their wallet and deposit the entire remaining 29.27632 DRK on an exchange. Normal actions resulting in unavoidable and unwitting deanonymisation.
Assuming the wallet is already denominated, then in your scenario, the DS inputs would be two 10s, nine 1s and three .10s with the rounded up change going to the network... yes, it very much solves the issue you are trying to point out.
How does that solve the issues my post mentioned?
So you can calculate the security of your darksend by yourself with a few assumption you have to take (because you cant know) like
darksend with 50 rounds, masternode network has 2000 masternodes, and i assume for me in worst case 1500 of these are bad actors.
So i got something like:
(1500/2000)^50 = 0.000005 => its a chance of 1 : 1750000 that a bad actor (with 1500 of 2000 MN) statistically can observe my mixing.
For me thats enough secure to say its anonym. But for some it may be not enough, because they cant know if there and how many bad actors are in the net. so if all 2000 out of 2000 are bad actors, you can be sure it won't be anonym anymore. (I think thats the point im reading about MNs are not trustless, because you can't know if they save the sarksend or not)
You can just own the major amount of coinjoin-transactions to trace back what happens; no need to mess with masternodes. Combined with other statical analysis approaches this is quiet powerful?