so i just wanted to point out that there is the potential for some small amount of information to be leaked by people using very specific mixin values. for example, for 2 transactions which both used a mixin value of 23 might be surmised that its more likely than a random sample that these two transactions were initiated by the same party. standardized mixin denominations offering a range of options might be preferable. 3, 5, 7, 10, 15, 20, 25, 30, 35, ect.... (just as an example) plenty of flexability there. its unlikey that someone who wants more than 20 mixin couldnt be happy with 25 and would feel very strongly that they wanted 22 very specifically. so in this way there would be little cost to the user in inconvenience, but potentially reasonable gain to the privacy of the network for that very small inconvenience cost.
It's a good idea but people are probably going to do that anyway I would think. Who is actually going to use 23 as opposed to 20 or 25.
Probably only a dummy.
Well if that's the case its not really going to be significant on a system with a significant volume of transactions, so not worth worrying about. The idea is to avoid people comprising the system as a whole. If they want to compromise their own privacy they can take wallet screen shots and post them on Facebook for all we care.
I do think if higher ring signatures are made much more efficient (say like the the 1000 or 10000 examples given) due to some algorithmic improvement, then we should probably just pick some number and use that for all transactions. That doesn't exist now though. We are sort of doing this with the minimums, so what I'm saying is that maybe in some hypothetical instance the minimum becomes 1000 and everyone just uses that.
But lets keep this one as clean as we can. 
