since once finally broadcast to the network a TX should look no different than any other, I don't see why it couldn't be done without any change at protocol level.
I also recall some unspecified hypothetical risk of an observer being able to catch what dandelion attempts to hide (origin IP for a TX) but since that's definitely revealed to remote node regardless without dandelion, wouldn't that just mean that in a worst case scenario the TX would simply be as revealing as all TX are now? Wouldn't that mean dandelion would be worthwhile just in case it works as intended even if that's not a guarantee?
Last but not least, regarding the unspecified threat to it's efficacy, I can only think of two weaknesses but am curious if it's something else.
threat #1: if a nosy adversary controls enough nodes participating in dandelion, they could potentially follow the stem even though the extended path is not revealed to any one hop.
threat #2: anyon e with ability to dragnet surveil much of the internet or many large ISPs could deduce from watching packet flow & timing which encrypted hops likely belong to one stem/path
if these are the two threats, would the following countermeasure potentially have any benefits?
having the dandelion enabled light wallet que transactions for a round time (or block) increment, so that all participating nodes (and light wallets) would coordinate to unleash dandelion transactions in big bursts of network activity to make packet timing analysis significantly harder. Could do a broadcast pulse at 1:00 UTC, then 1:30, then 2:00 and so on (every half hour, on the half hour) or every time the current block height is a multiple of 5 or something. This would mean dandelion transactions would usually have somewhat of a wait but that may be worthwhile if it makes their efficacy more resilient to large scale network analysis.
I know IP address is but one of many pieces of info revealed with a transaction, but if you could drastically reduce the odds that a wallet address is ever reliably associated with an IP address, other things done to mitigate other information leakage could be far more effective. Also simply hiding your true IP (borrowing another by some means or another) doesn't stop multiple wallets from being correlated to each other by all being used from the same IP, even if that IP reveals nothing else useful, so long term ips and session ips are both very revealing when you use multiple wallets even if your origin ip masking method of choice is actually working (and it might not be)
forgive me if these are stupid questions/ideas
enhanced privacy without resorting to alts is an exciting notion
submitted by 
No comments:
Post a Comment