Suggestions for better privacy protection #1
Reference in New Issue
Block a user
No description provided.
Delete Branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Assign a separate temporary container to each secondary domain and randomize the browser fingerprint to prevent cross-site browser fingerprint tracking.
Related Firefox extensions for reference：
https://addons.mozilla.org/firefox/addon/temporary-containers
https://addons.mozilla.org/firefox/addon/fingerprint-spoofing
They have an open source version called fingerprintjs, even if it's not as powerful compared to the paid version.
You make some really good points, I hear what you're saying, and your results seem to bear you out, I think. The results I'm most interested in are the ones which come from https://fingerprint.com/demo, that project in particular takes the fingerprinting techniques and combines them into an identifier which it attempts to use to track you.
So next version will we use a different per-second-level-domain container. That much is settled.
What remains is to figure out which things to randomize. I am still hesitant to randomize things which aren't necessary. I know for sure I can't use
fpsas an example, either, because they don't publish source code and the javascript shipped in the.xpiis highly minified. It's completely unusable as an example, unfortunately. So instead we need to find examples for the ones we might be able to meaningfully spoof, and then figure out if spoofing them produces a desirable result. Using thefpsextension settings page as a guideline, possibly:are useful targets for randomization.
I ran some tests with fingerprint.com/demo, browserleak.com and coveryourtracks on my self-hardened Firefox and TorBrowser, with private browsing mode turned off to use temporary containers.
Test environment. Self-hardened Firefox on manjaro OS and Whonix TorBrowser running in a virtual box
Random fingerprints with containers allow containers to really reach a level of isolation that is almost as good as multiple different browsers. Hiding everyone under the same type of random fingerprint is easier and harder to monitor than hiding everyone under the same fingerprint, and even if TorBrowser had an identical fingerprint , we still couldn't hide the fact that you were using TorBrowser, and the price we paid for that was not being able to install most of the extensions that can cause fingerprint changes. But if you use randomized fingerprints, the fingerprint changes caused by these extensions will be harmless
Leaning ACK on the use of temporary containers per secondary domain. It will require significant changes to the part of the extension where the containers are initialized and the part of the extension where the I2P history is cleared. That's probably going to take at least a week to implement and test.
Leaning NACK on the use of spoofing for the purposes of obfuscating the fingerprint at this time, but I could be convinced differently pending some research and/or discussion.
So there are these two broad things I think we can target to help reduce the impact of fingerprinting by reducing the usefulness of a fingerprint. Assuming a thorough fingerprinting defense implemented in a browser(i.e.
privacy.resistFingerprinting, hypothetically)What the doext/fps extension does is entropic, and it does it in ways which interfere with convergent approaches. In fairness, they have checkboxes to change that behavior. In any case, though I can't just copy that approach, it's safer and more effective to do entropy only when a convergent solution cannot be applied.
I'm open to entropy based approaches where necessary though, if you can enumerate some new ones to me.