Lee *!* Clagett

> I'm not sure what you mean by this. The scheme described here only leaks: There is more information about the real spend than the current approach. Or rather, there...

> Metadata leakage about the real spend, in the context of a floating output, is mostly independent of how ring members are selected (the use of hashes here is irrelevant...

> If each of those is uniformly distributed, then no information is leaked about which bin member is the real spend. In other words, it is equally probable that the...

> If the ring-selection algorithm can filter out definitely leaked real-outputs, this changes the argument because other participants are less affected by tx-chaining (it lowers potential anonymity set size but...

> If most txs using floating outputs cannot be hidden (e.g. because floating outputs are mostly used for tx-chaining, and the delay between tx signing and floating output being added...

> Plus, you can just as easily blackball floating outputs from this proposal that are unequivocally real-spends. Then why not just use 1-member rings in these cases. Its simpler for...

> Note: Even pure ring signatures over a gamma distribution might not be heuristically neutral if the true spend distribution does not match the gamma distribution. An analyst who knows...

> Yes it would still be recommended. > >* "We show this binned sampling ensures privacy even in spite of a compromised sampling distribution." source > * The storage required...

> This response confuses me. Decoy floating members can and should be selected from the most recent 10 blocks. Why wouldn't they be? It's true that the presence of a...

In 99.9% of cases, a spent output results in change coming back to the same viewkey. So it is already possible to infer a balance. If a wallet always sends...