24 hours from now there will be an MRL meeting in this channel:

Rucknium[m]: so your CCS is, as far as you can tell, a way to make Monero resistant to statistical analysis?

Inge: Yes, a particular type of statistical analysis involving the ages of ring members.

The issue is this: The age of all ring members is known. The ring members each come from a particular identifiable block, and each block was mined at a particular point in time.

So you cannot just have the decoy selection algorithm select decoy ring members willy-nilly. If, for instance, you always select decoys from more than a year ago, and users are spending outputs that are less than a year old, then the "true spend" input can be identified. That would make Monero transactions somewhat traceable, in a sense.

The best thing to do is to employ a probability density function that exactly mimics the "real spend" age distribution, since then decoys look like real spends in pretty much every way.

To the extend that there is a substantial difference between the real spend age distribution and the decoy distribution, user privacy is somewhat harmed.

So the purpose of the research is to construct a probability density function for the decoy selection algorithm that is as close as possible to the real spend distribution so that real spends are as difficult to identify as possible.

i am still flummoxed regarding how we can know the real spend distribution

If you have enough data points, like any other distribution: subtract the synthetic distribution from the observed one.

but it can change over time... hence Rucknium[m] 's adaptive thing they mentioned .... hrmmmm

<monerobull> bridgerton is the bot handling the matrix-discord bridge :D

moneromooo: and how can one get the synthetic distribution?

wallet2.cpp, grep for gamma.

Also described in Moser et al.

There might be a URL or title in the source.

Though the exact distribution is in the code, as shown by jberman[m].