On Fri, Aug 5, 2011 at 1:07 PM, Andy Parkins wrote: > On 2011 August 05 Friday, Gavin Andresen wrote: > > Transaction forwarding could be randomised slightly, by randomising the > outgoing relay order; and adding a random delay between each forward. Even > the massively connected monitor can't represent _all_ the connections on > every > real node, so it would have no way of knowing whether it got any > transaction > from the originator or because it got a fast path through the first N nodes > to > receive it. > Right, while it doesn't warrant completely changing the transport protocol to UDP or implementing onion routing, I'm all for simple timing and order randomization changes if they can make attacks like this less effective. JS