Tor introduced guards, which are - oversimplified said - relays the client picks and holds for some time as the first hop, to protect the client against the so-called 'predecessor attack'. Entry guards prevent the client from connecting to the majority of the network in a short amount of time.

Now there's some planning towards increasing the guard rotation period and reducing the number of guards a client picks for that time-frame.

How are fewer guards per client better then more guards? Why is a longer rotation period (a client sticks with its guards longer) beneficial for the client?

  • You know, bad luck and such things. Like probability.
    – bastik
    Commented Oct 6, 2013 at 9:56

1 Answer 1


The short version is that when your adversary owns your first and your last hop you lose.

Assume you have an adversary that controls a couple of nodes.

You change your last hop all the time. If you also change your first hop all the time, they'll find you sooner. If you pick it once and then stick with it you either picked badly, and you lose, or you were lucky and made a great choice. Stick with it.

Since the network changes over time, Tor picks new guards once in a while.

As I understand it, recent research such as Elahi at al. and Johnson et al. suggests we should rotate them even less often than we do now.

It turns out that it is not entirely trivial without breaking the network too badly. See Ticket#8240 or Roger's blog post about the lifecycle of a relay.

  • Yes, when a client's first and last hop are adversary controlled hops all hope's gone. The answer is good, but I let it sit around for a while to encourage others to answer it, too.
    – bastik
    Commented Oct 6, 2013 at 10:48
  • I've never quite understood why it's better for a few users to be more-likely compromised while most users are less-likely compromised. Is it that, once a given user has been compromised, further compromise doesn't matter? Do these models assume particular patterns of Tor usage, such as daily versus occasional?
    – mirimir
    Commented Oct 8, 2013 at 5:57
  • @mirimir I think that's the case. Once a user is compromised, they're compromised. Commented Nov 8, 2022 at 5:31

You must log in to answer this question.

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