3 ms·

Tor Browser does letterboxing for anti-fingerprinting of usable window size, which can be kinda painful, with sometimes large margins of lost display space. An

• mancerayder 3y ago
  Can you explain letterboxing to a non front-end dev?
  • neilv 3y ago
    Normally, a browser window has a rectangular area of the display that is "visible" and used for the Web page, with browser controls and scrollbars and window decorations around it. To a company that wants to identify who the person is (such as to build a profile of the person and sell that information) one way is to "fingerprint" the browser and how it's used. The dimensions of that visible display rectangle help with this, since individuals will often tend to end up with a window size that's more particular to them than of all possible users. One browser anti-fingerprinting privacy feature targets that telltale rectangle size, with "letterboxing" (different from movie letterboxing). Which means that range of different visible rectangle dimensions of what could be many users get reduced to the same smaller rectangle dimensions. Which (like movie letterboxing) results in large unused margins for all users. So, say that you have a 1920x1080 native window size, including all the decorations and controls. That's a pretty common size, and doesn't fingerprint you very well at all. But your particular preferences for things like window decorations and font sizes and such narrow that down quite a bit. And if you don't have a common display size, nor a maximized window, but you tend to have the window the same size, that can fingerprint you even more. With letterboxing, your 1920x1080 native window might give you, say, a 1600x800 display rectangle for the browser page. And a bunch of other people's native windows, of various sizes in the vicinity, and of various desktop settings, will also get the same 1600x800 display rectangle. With letterboxing, everyone is wasting precious screen space, but they're supposedly harder for adversaries to identify. I'm not thrilled with the tradeoff here. Though it's easier to stomach when so many Web sites make poor use of the display anyway, I've been desensitized. (Sorry, I'd make this explanation shorter, if I had more time.)
    • mancerayder 3y ago
      Thanks much for this!
• charcircuit 3y ago
  I don't know about Brave's implementation, but Firefox is lack luster as you can actually see the letter box 99% of the time. I feel like if real resources were poured into it then you could have sites almost look no different. In Firefox it really looks like the developers just made a smaller view port and called it good enough.
• autoexec 3y ago
  > I see a request for letterboxing in Brave, and a PR that does randomizing of some values instead: randomizing is the better way to go. Tor's approach to fingerprinting seems misguided to me. Their idea is to make every single user appear to be identical so that you can always tell when someone is using Tor Browser, but you can't track which Tor Browser user a visitor is. The problem is that new fingerprinting techniques are found all the time, and as soon as even one is discovered that Tor Browser doesn't know about or hasn't updated to address you become 100% trackable. Instead of trying to always make sure that every browser will produce an identical "Tor Browser" fingerprint and just hoping you account for everything that might go wrong, it seems like the smarter thing to do would be to make sure that every browser (and even every session) used a totally randomized fingerprint. That makes it impossible to track anyone across sites/visits and impossible to even tell the difference between someone using Tor Browser or any other browser that has a unique fingerprint (or randomizes to create one). Most people's browsers are already creating a unique fingerprint, the trick is just to make one person's browser differently unique every time. Every website will see your browser visit them once, but then never again.