6

I have stumbled upon the idea of disabling cache - completely.

On one hand, I am thinking that due to having to load all elements afresh when visiting a page, I appear more like a fresh 'New Identity' (thus decreasing linkability and trackability across Tor IPs within one Identity), EVERY single time I load a page; but then I'm wondering how much of a fingerprinting issue it also raises, decreasing anonymity instead.

Is it a case of an anonymity trade-off? (An analogy would be: using an email masking service where you're trading having 1000 websites all knowing your email address and having it stored on all of their vulnerable, breachable servers, for ONE website having the outline to your entire digital life and storing those same 1000 sites in one solitary but juicy basket, on THEIR potentially, equally breachable server.)

Would it be a similar trade-off here?

Perhaps it depends what other anti-tracking add-ons and settings you have enabled. For me, I have:

  • NoScript set to global blocking
  • Third party cookies blocked via FF settings
  • Self-Destructing Cookies with default set to expire 'after you close the tab' (so I assume with such cookie sanitizing I will more properly look like a truly fresh user on every given page)
  • RefControl with default set to 'no referrer' (again, making sure that I truly come across to every page as if I've literally entered the URL into the address bar and visited the page for the first time, ever)
  • RequestPolicy to massively sandbox every domain I visit to keep other domains from knowing I'm visiting that domain (via blocking third-part cross-site images, scripts, css, scripts, objects, etc.)
  • layout.css.visited_links_enabled set to false in about:config (to plug this in case you have enabled history)
  • FlashBlock and BetterPrivacy
  • HTTP Nowhere to block EVERY HTTP connection in the entire Tor Browser (short of direct handshakes to establish a domain's HTTPS with the Firefox default trusted CA store)

And the only exception to all this stringent blocking, is custom (often temporary) whitelisting/unblocking for sites I want to use more functionality on, which largely thanks to RequestPolicy are still very much sandboxed even with things like JavaScript and session cookies enabled.

So although it's very unusual behaviour, I'm guessing that it still gives stronger anonymity for each individual site I visit, via my stringent blocking of cross-site object tracking. I simply come off as a phantom user, someone maybe visiting the page from a saved bookmark, just with a Tor IP / User Agent / HTTP_ACCEPT Headers, no previous identifiable cookie, and a few other 'blank' returns that just make me look like a total unlinkable (but unique) ghost, to that one site.

To me, that's anonymous, IF that unique behaviour can't be linked, by that website's sysadmin, to other sites, which unless there was some collaboration or communication between the two (and it would have to be so for every single site on the web, two-way), I don't think is likely.

Or is the issue, that parties like the NSA, who are hoovering EVERYTHING up in their servers, will be the ones that can detect this extremely unique fingerprint across multiple sites, and so this being useless against the NSA and I should just fall back to 'crowd anonimity' once more?

Is it a case of, would I rather look more anonymous to the NSA, or more anonymous to each website that I visit on the Internet?

If so, is there a set of behaviours that can then combine the best of both worlds, with more knowledge of how TBB's default crowd behaviour of cache expiry, cookie settings, session length (how quickly things are 'reset' by default in TBB already), and other such fingerprinting considerations, work?

Improve this question

2 Answers 2

Reset to default
1

Is it a case of, would I rather look more anonymous to the NSA, or more anonymous to each website that I visit on the Internet?

Largely all of your concerns are addressed by Tor Browser. From a useability perspective blocking the referrer would break some sites. I think it's interesting that we rarely hear about consequences of individual websites attacking a person's privacy. Rather we hear about the protests precisely because of the consequences if 'Big Brother' can get their hands on this data.

To me, that's anonymous, IF that unique behavior can't be linked, by that website's sysadmin, to other sites [..]

True, this is one way of looking at the impact of caching. From the perspective of a website. Consider, however, that these same sites could look at how many of their guest users use Tor Browser and your customizations. If you're one of a few on Tor they can look at your use of Tor Browser, and your use of exits, as examples of fingerprinting.

Or is the issue, that parties like the NSA, who are hoovering EVERYTHING up in their servers [..]

This is where your choice of disabling caching will be to your detriment. These surveillance organizations aren't just looking at exiting traffic. They also look at your Tor client and it's interaction with your entry guards. If caching is disabled entirely it presents (potentially many) more opportunities to perform a correlation-attack against client-to-guard, and exit-to-site. You can expect these organizations are really good at positioning themselves. It also presents the opportunity to perform website traffic fingerprinting based on a known signature of accessing some site. If you use a particular site and your adversary knows this, you've just given them (potentially many) copies of a reproduce-able fingerprint.

If so, is there a set of behaviors that can then combine the best of both worlds [..]

This is exactly the choice Tor Browser makes. Caching isn't disabled entirely because, subjectively, the greatest modern threat against your online privacy won't come from collaborating websites. For that case you fall back to crowd anonymity -- and it doesn't always work. (You may be the only one viewing a site using Tor, or be identified by your Browser configuration). The 'new identity' feature is designed around this threat (and leaving persistent traces). The protocol, and caching as it stands, has the objective of preventing surveillance in general.

Tor Browser's design document should put your mind at ease. Except maybe for bugs.

-- leeroy

Improve this answer
10
  • "Consider ... that these same sites could look at ... your customizations." - Yeah, but it's still sandboxed. I stick out like a sore thumb, but they only see this very unique profile's activity on THEIR site, NOT others. So I may be a unique phantom, but I'm still giving them less information in other ways, as demonstrated. Really what I'm doing is giving Google less info of what sites I am visiting, which is the real benefit. So I guess it's just a toss-up between sites/services profiling me, or a wiretapping adversary like NSA doing it.
    – user1006
    Commented Jan 7, 2015 at 21:25
  • I have read the link and I am not convinced by their arguments, at all, I can see other non-privacy rationales coloring their arguments (like usability and thus success of the Tor project) - if high privacy is your goal and you have javascript disabled with requestpolicy etc., and usability as a low second priority, I think the set-up I have outlined above is wise, for protection against everyone but NSA types. So I think it's a choice, if either approach is to the detriment of the other. But not leaving so much IP linkability on 3rd party sites means less info being collected by NSA too.
    – user1006
    Commented Jan 7, 2015 at 21:29
  • Except disabling cache entirely isn't a question of useability. No sites will break from disabling the cache. Of course it's a choice. The point is if you think you're not traceable to malicious group of sites by turning off caching entirely you've never tried supporting Tibet from inside China.
    – user5341
    Commented Jan 8, 2015 at 0:47
  • Not to mention Tor Browser already makes cross-site unlinkability a primary goal. Tor Browser already takes your concerns into consideration for everyone. All disabling your cache entirely does is slow you down. Those same sites can track you by making educated guesses -- which they can do anyway. You've got other adversaries than big bad NSA. Your VPN operator, school network IT, that free wifi you use, the compromised router, they can all wiretap you.
    – user5341
    Commented Jan 8, 2015 at 1:32
  • I'm sure the Tor Browser by default does a lot of excellent things to help bolster cross-site unlinkability - but let me tell you, after discovering the eye-opener that is RequestPolicy, default TBB doesn't even approach that add-on's ability for that. However, I am super clear now on what I guessed in the OP itself - it IS a trade-off, between either individual sites knowing more about your browsing activity (i.e. activity other than their own site), vs. NSA types linking more easily your footprint around the web as they hoover it up everywhere.
    – user1006
    Commented Jan 8, 2015 at 8:52
0

It will not change the anonymity in case of persistent cookies like Evercookie. If a simpliest or too straightforward tracking is used - then an answer is Yes, but it's hard to beleive, that if someone is doing the tracking being so stupid about the tracking methods at the same time. Never underestimate your enemies

Improve this answer

You must log in to answer this question.