i wonder if the image on slide 19 can be actually created.

i know what they present is a cartoon to demonstrate fundamental differences

however, i would surmise that the actual curve for a decoy approach flattens out at some point.

i.e., the amount of "privacy", if that is a quantifiable term, is probably no different between 1024 decoys and 1024*n decoys

i wonder where i picked up the word surmise

The graph is a plot of tx size vs anonymity set size

not privacy

i sees a privacy arrow on the x axis

it's just saying privacy generally increases as anonymity set size increases

right. my point was, it might be interesting to actually try and analyze that. if "measuring privacy" is possible.

because, if so, we could possibly find a number for the maximum-benefit-gained-from-ringsize-increases number, and say with mathemagical confidence that there's no good reason to increase ringsize beyond n

I doubt anything less than 'all the outputs' can be considered good enough.