[Cryptography] [tor-talk] Dark Web should really be called the Twilight Web

[grarpamp]

On Fri, May 29, 2015 at 3:31 PM, Ryan Carboni <ryacko at gmail.com> wrote:
> In order to have a proper padding system, a lot more information needs to
> be leaked about current bandwidth demand.

That's only if you choose to attempt a padding-across-the-net
management scope, which is also going to be hard and slow to
manage and respond to bandwidth and other net dynamics.
(Though this was about GPA, it's probably also vulnerable to
endpoint interruption attacks that monitor your stream, unless
someone is there making up the padding slack at the far end.)
A wide scope seems hard in a low latency demand based net.
I'd suggest examining some form of next-hop, next-peer, or link
local padding scope negotiated with such peers. If you or your
peers get hit with demand, your negotiation distance is shorter.