Tor nodes build encrypted TLS sessions between neighbors to multiplex circuits of users. Multiplexing helps to anonymize users (?).

With an increasing number of middle relays multiplexing becomes rare (because Guard nodes... and Exit nodes are some "predefined" set).

Can it happen that with an increasing number of relays (often middle relays) timing attacks become much stronger? (maybe I could use netflow, etc. ...) Extrem case: each circuit has it's own middle relay.

2 Answers 2


That is wholly incorrect

The power of the Tor network is in numbers. Since there are currently over 6,000 known nodes, the probability of targeting enough of them to even attack a few users is highly improbable. Your middle and exit relay change constantly, every 10 minutes or so, but your Guard doesn't. It's set to a (hopefully) trusted node for a couple months

That being said, if an attacker had enough nodes within the Tor network, and the resource to do so, an attack is almost feasible.

Just from a mathematical standpoint, having more nodes is better, and certainly in practice

  • You are talking about compromising/attacking nodes? I'm talking about loosing multiplexing-anonymity and timing attacks based on netflow.
    – user674907
    Commented Mar 23, 2019 at 16:04
  • @user674907 I'm talking about both
    – unixandria
    Commented Mar 23, 2019 at 23:29

No, it is not, and - actually - my privacy project has started from Tor patch set that was not communicated by devs at all with the options that were allowing to tweak this particular matter. The thing is that 3-way hope is hardcoded, so if you will be using a different length - it can be a fingerprinting/deanonymizing pattern. All of us should use a random length like 3-7 as my simulations are showing and we should run as many of an intermediate nodes as possible

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .