Icon base by Lorc under CC BY 3.0 with modifications to add a gradient
I remember thinking how strange it is that websites can know all of your installed fonts when I was playing around with https://coveryourtracks.eff.org/ and https://www.amiunique.org/
I'm on linux and I have some extra fonts installed. Just the combination of them alone is so unique to me that you don't need anything else.
The second big one for me is how shocking I find it that timezone spoofing isn’t standard, now that so many people use VPNs. Why would someone connecting from Sweden have their clock set to GMT? Etc
One known problem is that on Firefox for Linux, every font you install via the package manager becomes a System Font, and thus is immediately "visible" as soon as Use Document Fonts is enabled, irrespective of the setting for CSS font visibility. I've even asked about here if it is possible to run multiple fontservers on a single session, as that would help palliate the fingerprinting by running Firefox profiles connected to different font lists.
As a relatively useful alternative, you can have Firefox profiles on different users, each having their own fontset available at .local/share/fonts, but for that to work you also have to remove all those extra fonts you installed via the package manager.
Oh interesting I wasn't aware of how all of these things work. So even on the strictest CSS font visibility setting system fonts are not hidden (provided Document Fonts is enabled)? Local fonts I assume are hidden?
From comments in the bug report I got the feeling they should have, but aren't; proably because of intrinsic issues with what it means to be a "system" component.
I have no means to test in Firefox Linux if local fonts are hidden because I don't know what makes a font "local" either. Any font placed where the fontserver looks for them (be it system paths or .local/share/fonts) seems to count as "system", which is why I specifically suggest using .local/share/fonts with different users: it's the only path where you can offer variability. More importantly, CSS font visibility settings seem to mostly be intended for Javascript-based fingerprinting, and they do not work at all against CSS-based fingerprinting.
I am pending to raise an issue ticket to make it so that "Use Document Fonts" becomes a Site Preference instead of a Global Preference; that should go a very long way to enhance privacy in these cases. Once I do, I'll link it here.
Hopefully this helps to get around some of the bullshit reddit banning
why not stop using reddit once and for all?
brave have it, but it doesnt look it will, reddit is too smart for that. you need things to spoof you fingerprint, IP address, device and components. im on a forum where they use anti-detect browser(not an actual browser, but to open instances of a browser using different ip/devices,,,etc) to manage all of this, plus you need an reddit account thats not tainted. i think reddit is too used to mozilla already, things adspower, dolphin anty, is one of many that does this.
I got identified on Brave immediately
Opera worked for me
Very good to see these changes, but could somebody explain this one to me? I don't understand how that helps with fingerprinting protection...
The available screen resolution is the screen height minus 48 pixels.
If I had to guess is because you can be tracked by your screen resolution too, It's just a way to minimize that. If you start tor browser, for example, it does not open full screen but only occupy a smaller window, for the same reason
You can be, duckduckgo was accurately profiling my location from resolution. Librewolf has a setting called letterboxing that fixes this, but you have to turn it on in about:config. Once I did, duckduckgo's profile wasn't accurate anymore.
You absolutely can be tracked by screen resolution, especially when using a monitor with unusual resolution (few years ago I was still using 1440x900 so I understand this well). I just don't understand how this specific change would help prevent that.
At one point, years ago, they were talking about removing the screen resolution entirely, and just make it a copy of the window size values instead.
Guessing it broke too much stuff, since it seemed like a nice idea but never eventuated.
Is that why the Surface 7 has such an oddball resolution of 2304x1536?
I don't know how that helps either, since it is a fixed diff. Just pre-check adding 48 to any sus screen resolution you get reported. Enabling letterboxing by default and reporting the screen resolution as the nearest larger "common" size would be a far more practical response.