this post was submitted on 22 Nov 2026
13 points (100.0% liked)

Privacy

52202 readers
639 users here now

A place to discuss privacy and freedom in the digital world.

Privacy has become a very important issue in modern society, with companies and governments constantly abusing their power, more and more people are waking up to the importance of digital privacy.

In this community everyone is welcome to post links and discuss topics related to privacy.

Some Rules

Related communities

much thanks to @gary_host_laptop for the logo design :)

founded 7 years ago
MODERATORS
13
submitted 3 hours ago* (last edited 1 hour ago) by warcat@lemmy.ml to c/privacy@lemmy.ml
 

Okay I did some research on tor but I still don't know if I understand this correctly.

From what I understand the ISP (or whatever is your last proxy, like your VPN etc.) can easily detect if you are using tor. So for example if I connect to tor at 10:00 and enter a website X at 10:01 and send like 100MB and leave tor and website at 10:03, the isp can tell i was on tor for 3 minutes and send roughly 100MB. So, if the police or administrators check the website X they can tell someone at 10:01 to 10:03 send the same amount as me and can pretty much assume it could be me. Add to that a bunch more information isp would have over long time, how does tor mitigate this?

edit: To make this more clear I mean, a second example: The ISP the logs show that I send (roughly, due to encryption) ~90MB of encrypted data over tor at 10:01 and downloaded ~35MB at 10:02. And on the backend of the website at the time only one (my account) uploaded 85MB at 10:01 and downloaded 30MB at 10:02. Wouldn't that eliminate 99% of other tor users?

you are viewing a single comment's thread
view the rest of the comments
[–] LewdLemon@sh.itjust.works 6 points 2 hours ago* (last edited 2 hours ago) (2 children)

What you are talking about, OP, is timing analysis or timing attacks. The Tor Project has information on that specific threat vector on their website:

It is possible for an observer who can view both you and either the destination website or your Tor exit node to correlate timings of your traffic as it enters the Tor network and also as it exits. Tor does not defend against such a threat model.

In a more limited sense, note that if a censor or law enforcement agency has the ability to obtain specific observation of parts of the network, it is possible for them to verify a suspicion that you talk regularly to your friend by observing traffic at both ends and correlating the timing of only that traffic. Again, this is only useful to verify that parties already suspected of communicating with one another are doing so. In most countries, the suspicion required to obtain a warrant already carries more weight than timing correlation would provide.

In other words: yes, assuming a threat model where the threat actor can observe both the traffic leaving your machine and the traffic going to the website in question, you might be linked to the target website through timing analysis (though the content of the exchange would still be protected by encryption).

[–] FineCoatMummy@sh.itjust.works 3 points 1 hour ago (1 children)

I have wondered if this style attack could be made much harder, by fuzzing both times and sizes. Ofc that could severely degrade the near real time experience. But let's say a delay was added. Bob's 12K message to Alice is rounded up to 128K. It mixes with many other 128K rounded messages from Frank to Jane, and Tina to Winston. Then it comes out later, sent to its destination. It's harder to correlate the timing. It is no longer clear whether the message Alice got was from Bob, Tina, or Frank, b/c all of them were in the same timing window.

The delay might be unacceptable for many uses. But it might be OK for others. I'm fine loading a web page in a tab and doing sth else for a while as I wait. And size rounding doesn't help for larger data. So it's far from perfect. But maybe it would help in some cases?

[–] warcat@lemmy.ml 1 points 1 hour ago

Maybe the delay could be reduced by lowering the fuzzing in high traffic circumstances, like if 10 users connect to the same website the fuzzing could be higher, but when 100, be lower (like, only 2 people mix) as there's much more data going on. I can see how that could improve privacy.