Is there any truth to this, or is he misrepresenting the issue?
- Not even VPNs offer RPKI only exits:
- The security slider issue was pushed off because they generally don’t favour short term fixes and had planned to completely overhaul it
- The user agent “issue” was already discussed to death in another topic here, and it is invalid.
12 Likes
He’s insufferable when he refers to “the most common way to access the darkweb”. I will be also willing to bet there are caveats with the “counter raptor” solution, that he simply doesn’t discuss, instead goes off on a conspiracy that “ThE ToR BrOwSeR dEvS conspiracy”.
The user agent issue is the weakest of the argument, that about:config option is absolutely useless as it can be exposed through other ways, and then be contradictory giving off a unique fingerprint.
Realistically if you’re worried about JS leading to some sort of bypass the only solution is Tails or Whonix, and not worrying about JavaScript. We’ve moved past that.
7 Likes
I didn’t watch the video, please tell me they didn’t recommend changing about:config in Tor Browser?
1 Like
Yeah he was complaining that users should be able to spoof their navigator user agent and then whinging about how that it is reducing “safety” for people on the “safest” setting, but totally misses the point. It’s mentioned under User Agent Spoofing Changes.
cringe "correctly fingerprinted as “Linux”. He further shows how little he knows by talking about extensions making you “stick out”, that requires JS, which you’ve disabled so no, not in that case.
Either way it’s really a nothing burger.
And it concludes with the even weaker argument “gIVe The uSeR tHe ChOiCe”, absolutely not, it’s up to the Tor Browser developers to make decisions about their browser and how all users will look, so they don’t stand out. If there was a shitload of windows and Linux users that data point would stick out a whole lot less.
He should really watch our video: What Is Browser Fingerprinting? (And How to Stop It!) so that he learns about entropy. It’s also very easy to convince a user they need JavaScript by simply making the content unavailable to them until they enable it, so yes, those users should be a priority.
7 Likes
I’m glad someone brought this up (and that I’m not the only one who thinks it’s nonsense). I’ve enjoyed some of his videos in the past, but this one really put me off. It was unpleasant to see so many comments saying things like “Tor is compromised”. I personally believe they’re mostly made by bots/intelligence agencies/etc, but it still sucks to see them.
3 Likes
But how is he incorrect about the Raptor fix? This seems like a big deal that may present a way to identify tor users
1 Like
I personally immediately write off anything like this that doesn’t mention that Tor is currently not quantum resistant and is likely the #1 target for store-and-decrypt-later attacks. This is because that’s probably the largest legitimate outstanding issue with using Tor today, and anyone serious about bringing meaningful criticisms or actionable warnings would mention it as part of their list of grievances.
2 Likes
but why?
anyone capable of doing this is also already capable of doing trivial netflow analysis and deanonymizing any circuits on the fly, which tor explicitly doesn’t protect against.
should Tor have quantum resistance? yes.
does it matter in practice? absolutely not.
quantum computers are always “10 years away” and decrypting the 3 layers of encryption for each and every circuit with a quantum computer gets you no added information over netflow analysis
To clarify, I’m not saying I think this is a significant concern, and I would still use Tor (albeit with this factor in mind). I’m saying this is obviously more of a concern than your user agent in Tor Browser and anyone acting in good faith would mention quantum resistance over the user agent “issue”.
I don’t think quantum resistance is a massive hole in Tor for all scenarios or anything (right now, at least), but it is an actionable warning to give users to consider in their threat models before using it. I do think it’s somewhat worth mentioning when discussing Tor the possibility that your Tor traffic could potentially be fully decryptable by intelligence agencies in a conservative 20 years time, although mostly for people like whistleblowers who would remain a target after that amount of time (or I guess serious criminals, though them I don’t have sympathies for). At least more worth mentioning than the user agent thing.
That said, I don’t think it’s accurate to say that anyone capable of storing and decrypting later your Tor traffic can already deanonymize your circuits as a matter of necessity (which implies the NSA can already deanonymize any Tor user right now, as they are certainly store-and-decrypt-later capable).
1 Like
What I don’t understand is why doesn’t tor protect against this. Tor has always been associated with anonymity and they can obviously see that this presents a problem. I’m sure they also know that technology now exists which can protect against this sort of thing. I’m thinking of NYM specifically. I’m no expert in any of this, so maybe I don’t know what I’m talking about and maybe something like what NYM does wouldn’t work with tor. I just don’t see why it wouldn’t.
2 Likes
If they are capable of storing the data to decrypt later as you are suggesting, then by definition they are capable of performing metadata analysis on that data they’ve just stored, so @SkewedZeppelin is correct.
If an adversary sits between my network and my guard node to collect the traffic for storage and later decryption they by definition can also analyze all the traffic between all the other Tor nodes in the circuit to deanonymize the traffic in real time? Is that what you’re saying?
Perhaps I am misunderstanding, but this is how I am reading these assertions, and with this reading I don’t think that is a logical conclusion to draw at all.
1 Like
It would increase latency dramatically to properly mitigate that.
See Anonymity Trilemma | Freedom Research Lab
Sorry yeah, I’m not really thinking clearly. I suppose a local passive adversary could pose the threat you’re suggesting.
But why are you expecting them to bother breaking the encryption on the Tor circuit instead of just the end of the chain terminating eg. HTTPS?
You mention a local adversary intercepting your connection to the guard, that’d be 4 layers of encryption they’d have to consistently and efficiently break for each and every circuit.
That is an immense cost even with a hypothetical capable quantum computer
They’re more likely to just break the end to the server you’re connecting to, since that is only one layer.
And in the context of onions that just adds more layers of encryption.
And I still posit that an adversary with both interception and capable quantum compute available to do this would just partner up with other jurisdictions and instead combine their netflows to deanon the circuit instead of bothering to break it.
I’m not saying it is impossible, just that there are easier ways to achive realistic goals that they have.
Breaking the HTTPS coming out of an exit node wouldn’t always be deanonymizing, whereas breaking the Tor circuit traffic necessarily would be.
Sure, I didn’t say that anyone was close to having a QC where this would be a reasonable attack. That’s why they store now and plan on decrypting later, when it is reasonable. I think it’s still worth consideration though if you would be concerned about your traffic being deanonymized even decades out (again, mostly people like whistleblowers, activists opposing authoritarian regimes, but even if that doesn’t fit your situation that doesn’t mean it should be discounted entirely).
Potentially, but that’d kind of be a separate line in a threat model, so to speak. That also being a potential concern doesn’t entirely negate other concerns.
1 Like
I will simply note that there is a ton of ongoing research into migrating Tor to a quantum secure network. It is not something they are slacking on, and the nice thing about them being primarily a research project is that all that work tends to be compiled in academic papers.
10 Likes
Hopefully it goes smoother & quicker than the v3 migration.
I agree, again I only mentioned it to draw a comparison to what I would say are not legitimate concerns with Tor like the user agent “issue”. If someone is warning people about user agents instead of something like store-and-decrypt-later, I don’t think they are acting in good faith.
That’s all my original point was.
1 Like