Firefox expands fingerprint protections | Hacker News Reader

Firefox expands fingerprint protections

blog.mozilla.org

323 pointsptrhvns170 comments

I exclusively use private browsing, but I know that doesn't do much in preventing tracking, so it's nice to see this finally starting to roll out.

The fact that I have to go to great lengths to browse anonymously - and companies desperately try to circumvent my genuine decision to opt out of their tracking - tells me everything I need to know about those companies. Words like sleezy, shady, and predatory come to mind.

I would love to see this taken one step further and have states/countries prevent companies from tracking me altogether if I reject their cookies, but I fear it's more likely those companies will lobby to prevent Firefox from protecting us.

You could try to use profiles instead of private browsing. It keeps things separated.

Also profiles can be configured and used with CLI, no need for UI (old or new).

    ./firefox -CreateProfile "profile-name /home/user/.mozilla/firefox/profile-path/"
    ./firefox -profile "/home/user/.mozilla/firefox/profile-path/"

And, you can run it directly, no need to launch default firefox profile:

Given that /usr/bin/firefox is just a shell script, you can

    - create a copy of it, say, /usr/bin/firefox-hn
    - adjust the relevant line, adding the -profile argument

If you use an icon to run firefox (say, /usr/share/applications/firefox.desktop), you'll need to do copy/adjust line for the icon.

Instead of needing to know scripting for a core feature, it would be nice if I could tell the program to ask me every time I open a new window which profile that window used. Right click would have an option like their containers "opening new profile window" .

Better if they would allow some configuration like toggling js by domain. uBlock is great, but I would like first party support.

NoScript extension...

Right-click on the Firefox icon and choose "Open Profile Manager".

Or add "-p" to the startup command to do the same thing without right-clicking:

    firefox -p

> Also profiles can be configured and used with CLI, no need for UI (old or new).

AFAIK, they can only be created at the command line, not configured. If you want to do things like change default settings or install extensions from the Firefox Add-On store, you can't really do that at the command line.

You can do that by mucking around in the user.js file and manually adding .xpi files to the extensions/ subfolder, but that's probably stretching the definition of "done at the command-line" since most people aren't creating Puppet modules to manage Firefox profiles.

Perhaps someone knows an easier way to do this, though.

Tor? Although I wish there was a way to make a reddit account.

You might want to check out the Mullvad browser. They work with Tor and are based on Firefox. It won't connect you to the Tor network but still

Library

Out of curiousity, how would you steelman the argument that fingerprinting is no different than a store owner, standing behind the counter, taking note of the faces of who enters his store, and maintaining a log?

The store owner visibly responds to the customers differently. Fingerprinting is invisible. It's more like the store owner recording everyone on hidden camera.

So no, you cannot steelman a broken analogy.

To make that analogy closer to the Internet reality, I would say that Internet tracking is more like a cabal of shop-keepers, librarians, neighbors, utility pole workers, and so on who are keeping track of all the faces, all their habits, what they look at, what they say, who they interact with, and share this information amongst themselves, recording it in perpetuity. They also share details with the police and anyone who cares to purchase them.

When you talk about a "shopkeeper" it gives it a small community charm. The Internet is anything but that.

Exactly. The "shopkeeper" is cross-correlating my sleeping habits, my browsing data, 27 data-leaks, my credit score history, the proximity of other devices and WiFis, the pictures my in-laws posted of a get-together, sentiment analysis of voice messages...

All while showing me 2 advertisements before I enter the store, trying to trick me into clicking a mysterious "track me more" button while I try to get toothpaste, and never lowering the price of pasta for me because my wife mentioned on a post that she loves eating Italian.

And he's the town's least creepy shopkeeper.

The difference is scale and intent. A mom and pop store owner “remembering” my face versus big tech tracking is like comparing a nosy neighbor to the CIA.

One of them might peer out their window, the other will infiltrate every aspect of your life. One of them is bored, the other has no qualms about doing significant harm to you if it serves their interests.

It's automated data processing at scale rather than a local mom and pop country general store. The profit seeking, decision making, management culture driving decisions is a fundamentally different relationship. Also I don't think store owners do that?

Rather than presupposing an analogy to something importantly different, I would propose that the steelman would be along the lines of noting that ads and hyperpersonalization are effective at meeting and predicting your needs, and steering you towards an interpretation of your own needs that finds their fulfillment in deepening a consumer relationship. And if you get steered into lock-in with one company's ecosystem, you get the convenience of a stack of vertically integrated services.

I'm fairly confident I could sue that store owner for stalking if they were logging every time I entered that store and left, along with all my activities.

I'm absolutely positive I could if they were getting other store owners to help them track me.

What I don't understand is why this is unacceptable if they do it to a single person but perfectly normal if they do it to all their customers. IMO that should make things worse, not better.

Let's put it this way. You'd get a restraining order against someone if they followed you around all day, logging when you woke up, ate, who you talked with (even if they don't hear the conversation), where you went, and when you went to bed. That's clearly stalking, right? So why us it suddenly acceptable when it's being done by some guy named Mark who is stalking a billion people instead of just one?

We clearly differentiate this from being a regular customer at a store. If I'm a regular at Joe's Corner Market and get a sandwich every Wednesday for lunch then he remembers me because we're talking face to face and making conversation. It's personal. There's clear consent in what I'm sharing and there's a clear expectation that Joe isn't going to use that information to manipulate me or follow me around town. Our interaction is limited to the store and maybe bumping into each other on the street. It's clearly not stalking, we're just friendly. The same way your partner might know about when you wake up, go to sleep, eat for breakfast, and all that same stuff. Your partner isn't stalking you.

[Edit]: I want to encourage the above comments. Doesn't matter if recursive4 believes the other side or not, I want these conversations to be front and center. I like to see the other responses than mine as well and I think these help us refine our arguments and by being prominent they help others be convinced and join us. So while I know we don't usually talk about how to upvote/downvote, I'll just say "vote strategically rather than agreeability" :)

Doesn't your (proverbial) Costco membership card track every time you enter and leave the store? Doesn't seem like anyone is suing them...

Also, if they were logging you specifically, you may have grounds to stand on. But if they're logging every customer that comes in/out (like websites do), I think there is a lot less grounds for a restraining order or anything

Edit: Found out I'm using 'proverbial' wrong but I think you get the idea either way.

> Doesn't your (proverbial) Costco membership card track every time you enter and leave the store

No (you have to use it at the register for Costco to know you were there),

and they don't track your every movement in store either,

and they don't track your every movement outside the store either,

and there isn't a standard way to say "I don't want this" which they nonetheless choose to ignore.

> No (you have to use it at the register for Costco to know you were there),

Not the case where I am: Costco scans everyone's card upon entry into the store.

> and they don't track your every movement in store either,

It might: Costco certainly uses security cameras, and it's possible that Costco may be using some sort of AI / facial recognition software alongside the cameras. Perhaps someone who has worked with Costco in security / loss prevention could chime in.

> and they don't track your every movement outside the store either,

Probably true, although who knows whether Costco purchases data about its members from brokers?

> and there isn't a standard way to say "I don't want this" which they nonetheless choose to ignore.

I think you're talking about the equivalent of cookies in the real world, in which case I'd generally agree.

> Costco scans everyone's card upon entry into the store.

Not the case where I am: You just ask for a "code 99", because they are legally required to sell alcohol to nonmembers where I am. I'm sure there are other loopholes too ("I want to browse to see if I want to join"), but nonetheless, no tracking for me.

> it's possible that Costco may be using some sort of AI / facial recognition software alongside the cameras

It's possible that they have a magic genie that grants wishes to shoppers who rub the right pallet, but if we have no evidence of either, we need not entertain such hypotheticals.

> Probably true, although who knows whether Costco purchases data about its members from brokers?

Those brokers are literally the people to whom we're comparing Costco. If Costco must outsource the act in question to the brokers even in the analogy, then Costco is not analogous to the brokers.

(I get the idea)

Costco is much more like the Joe's Corner Market. I'm only dealing with Costco. It is a bit more invasive than Joe's Corner Market, but I'll admit that I have much more trust for Costco given their history. Like Joe, Costco isn't following me around the store, unless I explicitly ask for their help finding something. They similarly won't follow me outside unless I'm explicitly asking for something like a delivery or some other service. It is always explicit and I'm always aware that I'm being "watched".

But the key difference is that Costco isn't sharing that data with Walmart, Facebook, and others. There is some tracking and I definitely don't like that, but there's a huge difference in going to Costco.com vs Facebook.com or even Google.com. Heading to Costco.com uBlock hits me with 9 blocks. Heading to google.com I get 17 and then it is constantly rising. In the time to write these few sentences it has already hit 30. Meanwhile, costco is still at 9. I mean I'm literally on google.com sitting and staring at a search page doing nothing. There's a much more aggressive and invasive attitude here.

Mind you, nuance and intent matter very much. Without them we wouldn't be able to differentiate a partner you live with and a stalker. Similarly expectations of trust. I'm glad you're asking the question of getting the steelman and pressing, but we must make it clear that if we're going to brush away detail and be dismissive of the nuances then we are contriving an environment where we would be unable to differentiate these things. But again, the consequences of that contrived setting is that we would not be able to differentiate someone's husband/wife from a stalker. And that result is beyond laughable. So maybe the better question is to ask where these lines are drawn. I'm not sure there's an easy answer, but I'm certain it is important.

Costco membership is more like voluntarily accepting cookies.

Even more than that, it’s purchasing a membership that requires authentication when you use the store.

So I go to Costco and buy wine, and next week I get a letter from DMV reminding me of the blood alcohol requirements, I get people visit home selling wine glasses, when I go to work someone hands me the best of wines coffee table book, and so on.

Ok but you say - It is not this overt online - well if we live digital life, lot of things are not overt, but we know we need to clean cookies, some of us create containers, some of us use TOR, so the sensibilities in digital are different than real life, and I am showing pretty much equivalent examples / metaphors if the same level of intrusiveness was there in real life.

Don't forget that time Target accidentally informed a teenage girl's patents she was pregnant by sending her coupons despite not buying anything explicitly maternity related. And how Target's solution was "oh, we'll just add ads for other stuff too so it feels less creepy" -____-

https://archive.is/kK1V8

Lots of moral values/legal rules are based on magnitudes and scale.

You can talk at a normal voice inside your own home at night, and even if the neighbor can hear you through the thin walls, they have no legal recourse. If you start blasting music, the police will (in principle) come and stop you.

Some things are okay in moderation and simply bad in excess.

Value of the dataset to attackers/buyers.

I am dreaming for righteous 'small' employees too, those who carry out the dishonourable practice of implementing privacy intrusion following instructions, for money. Corporates are built by thousands of ignorant grey workers.

The "Temporary Containers" extension is great here, allowing pretty easy compromise between different buckets of sites. I'll have some personal ones that I log into, others go specifically into a snoop container, and the rest get temporary ones that evaporate when closed. https://addons.mozilla.org/en-CA/firefox/addon/temporary-con...

stoically, the maintainer and creator of this extension unfortunately passed in early 2023. There's a new fork available[0], linked from github[1].

I briefly discussed this extension and how to proceed after the passing of a maintainer with Mozilla staff in their Extensions and People teams at FOSDEM this year, but there was no real procedures in place at the time of our chat.

[0]: https://addons.mozilla.org/en-GB/firefox/addon/temporary-con...

[1]: https://github.com/stoically/temporary-containers/issues/634

Oh, damn, good to know thanks! Sad news indeed.

Yes, that's truly sad that he passed. It is fortunate that he left that open-source project, where others can pick up the torch. As the forks continue, it becomes a small tribute to him.

> I would love to see this taken one step further and have states/countries prevent companies from tracking me altogether if I reject their cookies

You're in luck since EU's GDPR is about informing users of PII harvesting and consent in general (among other things). So the banner is not only about cookies. And I think nowadays there are similar regulations elsewhere.

The last thing I want to see is more banners that don't actually do anything for your privacy. Let's be real, websites/companies will do whatever they want with your data, the banner is just for show.

I was just pointing out to parent that it already happens...

This is a good use of Firefox resources. Unfortunately Firefox is at a natural disadvantage for fingerprinting by virtue of being used by such a small number of users.

Interesting. So when you try resist fingerprinting. If you dont go all the way you're at risk of making your differentiations smaller?

As an oversimplified example:

If a website has 100 visitors, and 99 of them use Chrome, and 1 user uses Firefox, it doesn't matter how good their fingerprinting resistance is, they're always the one using Firefox.

https://xkcd.com/1105/

But if another Firefox user comes they are indistinguishable from each other, while every Chrome user is uniquely identifiable, are they not?

> if another Firefox user comes they are indistinguishable from each other,

Even if every Firefox browser gave off the exact same fingerprint, that wouldn't make the network traffic indistinguishable between Firefox users. There is a lot of entropy that is provided by your network stack of your device, the networks you connect to in order to get to the end website, the behavior of your requests, etc.

Now, most websites aren't doing this kind of analysis. But it isn't unheard of or impossible. There are major websites that are known to do TLS fingerprinting.

Firefox is low on browser count but it's still around 4%[0]. That's enough that there will be lots of collisions. Even a small percent of a very large number is a very large number

[0] https://radar.cloudflare.com/reports/browser-market-share-20...

Of course.

However, if you're trying to search for somebody, and you're able to eliminate 96% of the data, you're in a much better position to accomplish your goal.

Whether or not you should care about this depends on what kind of tracking threats you're trying to avoid.

I mean yes and no. Raw numbers still matter. It's all about context. If you have a billion visitors and rule out 96% of them, sure, searching 4 million it's easier but it's still such a large number that that alone isn't enough. That's all I'm trying to say.

Completely agreed

I often think about this in connection with my user agent. I am sure it helps identify me. If I spoofed a Chrome/Windows UA that would probably be better from a privacy perspective. But if we all do that then web designers will never know that we exist. I want people to know there are Firefox and Linux users out there.

Spoofed UAs are easily detected. And if you are spoofing your UA you are among a very small subset of users.

Easy to detect but companies are lazy. I remember when Netflix first worked for Linux on chrome but not Firefox. I changed my agent and was good to go. After some months I emailed them asking to lift the agent block. They assured me they weren't blocking by agent. I sent them screenshots. They doubled down. So I said ¯\_(ツ)_/¯ and just kept using the agent until they unblocked it

Absolutely, but the parent was speaking about privacy. Access is a different story, because you can test different user agent strings, and immediately determine whether you get access. By contrast, you can't change a user agent string and readily determine whether or not you've broken someone's ability to track you.

My example of access is just a clearer example of laziness. Maybe they were tracking but it seems unlikely, right? If they were, why not block? Laziness is a much better explanation.

I can get feedback with access, I can't get feedback with tracking. That's why I mentioned access.

They probably weren't tracking you, that was probably a case of directing a user toward a supported browser for customer support purposes. I would imagine that was a requirement in somebody's Jira ticket, solved with a few lines of code.

By contrast, tracking people on the web is a multibillion dollar industry, and there are out of the box commercial libraries that do very sophisticated tracking. None of these solutions rely on user agent string alone.

The vast majority of websites by count are not doing anything sophisticated. But some are.

The announcement came with the claim about DRM. So I believe there was some "legal" issue about it. I'm also sure they didn't actually care that much.

  > By contrast, tracking people on the web is a multibillion dollar industry

Of which Netflix is a part of.

  > The vast majority of websites by count are not doing anything sophisticated. But some are.

And this is my point. Somewhere like fingerprint.com is trying to use all the tools available. But most places aren't. Facebook and Google? Sure, I buy that. But mentioning that many places are lazy is not the same thing. It is a game where we can't win completely and we still need to let people know that small gains are still meaningful. A major problem we face with privacy is that people feel so powerless that it is useless to fight back. But that's not true. Just because your bulletproof vest doesn't stop a missile doesn't make it useful. A bulletproof vest that only stops small caliber is still better than no vest, since most shots are small caliber. Pareto is still alive and well here.

Yes, there are plugins (e.g. widevine) that do DRM and they have/had varying browser/os compatibility.

But ultimately, Netflix is just trying to check a box in their contractual obligations, and/or prevent high-schoolers with chrome dev tools from sending movies to all their friends. They're not really interested in spending large sums of money to figure out your browsing history. It's just not relevant to their revenue stream.

>> tracking people on the web is a multibillion dollar industry

> Of which Netflix is a part of.

I was referring to businesses that do web activity tracking as their primary business. Facebook and Google's primary business is advertising, which isn't the same thing, and they control enough products that they don't actually have to do very much fingerprinting in order to target ads effectively. Most of their data, people voluntarily hand over. I was getting more at the big ecosystem of commercial tools that others can implement that do these sorts of things. e.g fingerprint.com and many others.

>This is a good use of Firefox resources. Unfortunately Firefox is at a natural disadvantage for fingerprinting by virtue of being used by such a small number of users.

I'd rather be trackable but secure -- the big draw for me is NoScript. Paired with uBlock, I'm safe from malvertising[1]

[1] https://en.wikipedia.org/wiki/Malvertising#Examples_of_malic...

I was wondering why uBlock is not enough since you can block Javascript globally and re-enable per site. AI's answer:

Only things uBlock doesn’t replicate:

NoScript’s anti-XSS and anti-clickjacking heuristics (uBlock just blocks the sources, not sanitize payloads).

NoScript’s control over other active content types (e.g., WebGL, media codecs, etc).

It's also a lot better UI -- it doesn't look like it's nearly as easy to see a drop down list of JS and selectively enable as NS allows.

You're more trackable by using NoScipt and there's no good reason to use it if you know how to properly use uBlock: https://github.com/arkenfox/user.js/wiki/4.1-Extensions#-don... uBlock is a content blocker so it can do everything NoScript can if you learn its advanced UI usage. Using additional extensions makes you more trackable.

>You're more trackable by using NoScipt and there's no good reason to use it if you know how to properly use uBlock

What data do you have to support this assertion? uBlock doesn't seem to have the ability to selectively enable only JS nessecary for functionality, and if it does, the UI makes it much more difficult to enable.

I just ran a test -- merely uBlock use renders me unique, whereas one in 5742.77 had the same fingerprint as me when using NoScript. (I suspect that's the number of people also using Firefox with NoScript who own this particular monitor size)

A big chunk of the fingerprinting techniques require JS -- it's pretty hard to ascertain what specific extensions are installed with it. I tested disabling it and it didn't seem to do much difference in terms of bits of entropy on EFF's tool.

I encourage you to try for yourself and then think hard on your advice.

I cannot judge the validity of your test and I have done any tests myself. I encourage you to read the link from post to the top firefox hardening resource (arkenfox) that labels NoScript redundant. This is further backed up by another top privacy resource: https://www.privacyguides.org/en/browser-extensions/ This also links to the uBlock docs which outline different modes. Medium/hard modes make NoScript redundant.

>I cannot judge the validity of your test and I have done any tests myself.

I'm going to assume you meant to say "I have not".

If you can't judge the validity, maybe you shouldn't give out advice that might be read by vulnerable populations, given the sources you list do not address my points.

There was a commenter some time back showing that browser statistics were easy to skew. Safari and Firefox are less likely to show up in analytics, so website owners think they're less important than they really are. Conflating client-side with server-side analytics showed quite a gap.

Most of the people who are just looking at browser statistics for the purpose of managing a website are using simple tools that just simply collect data from user agent strings. Determining browser from this isn't 100% straightforward, but it's enough to give website operators a rough idea of what browser to target. This data was more important in the days when everything wasn't Chrome/Android/iOS, and it actually mattered what version of IE your users were running.

If you're doing fingerprinting for tracking purposes, you're gonna be tracking a lot more in-depth data.

But in the end, there are pretty much three types of Internet user today: 1. The person who uses the default browser installed on their device. 2. The user who always downloads Chrome when they first get a new computer. and 3. Nerds who do something else.

I don't disagree, but it makes for bad web development practices. Google Analytics is still the de facto king everywhere I've worked, but it's going to more often Safari and Firefox than other browsers due to tracking protections and users being more likely to run ad blockers. Then there's all the edge cases like Brave.

I don't remember the discrepancy that the study found, but it's significant.

So… we keep optimising for Chrome, as if that's the bulk of our audience. That makes things shittier for everybody else, and we think it's okay because they're such a small part of the group. This reminds me of a former client burning almost 9 million euros every year because they excluded IE6–8 from their reporting, yet they would account for 15% of the traffic.

One thing I found that broke tracking algorithms was the ‘every tab is a new random profile’ extension. I can’t remember the name as I haven’t used it in a while and it broke a lot of logins.

They could not build a profile on you and it would break their system of tracking user login per device.

You probably mean Temporary Containers…?

https://addons.mozilla.org/en-US/firefox/addon/temporary-con...

Thanks to both of you. That seems valuable.

I've recently switched from Containerise + Temporary Containers to Auto Containers. Brand new addon, but the dev is responsive and IMO it works much better for creating new containers on the fly as you browse.

https://addons.mozilla.org/en-GB/firefox/addon/auto-containe...

https://github.com/Shajirr/FF-Auto-Containers

That's really handy! I like that it handles domains automatically

Edit: Seems to break ad blocking and there's some issues with login. Such as adding a container for YouTube requires also doing *.google.com since that's how the login is handled. Interesting and I'll keep playing around with it

Usually, with most browsers, you can't separate YouTube from Google in order to watch their videos. People can get around this using NewPipe and various forks, that are mostly for Android. There is also FreeTube for the desktop, that doesn't allow ads, but does connect to Google.

Yeah my bigger concern is that it is disrupting ublock. I want that running in all containers

It does not break uBlock Origin / ad blocking…

Well when I reloaded my YouTube tabs in it they all had intro ads and intermediate ads. Maybe it was that one time? Because as soon as I disabled the extension I was ad free again. Maybe your experience was different, but that doesn't mean mine didn't happen.

It uses the existing FF container functionality, so uBlock should not be affected. I also use uBlock and haven't noticed anything bizarre. Maybe you just used YouTube right before a necessary update to uBlock's filters got loaded.

Regarding login, it's a struggle with any container addon. But I like how this one makes it easier. It streamlines it a bit because you can right click the page and go to Auto Containers -> Exclude tab from domain exclusions, then login as normal and add each intermediate domain to the container's URL patterns. It would be nice if the dev could add a way to record all intermediate domains and add them at the end; sometimes I miss one or two due to an auto-redirect.

When I added it, and went to Youtube, it instantly started generating tabs in some kind of loop ... had to shut FF down to stop it, and even then had to kill the loop manually.

That’s most probably a conflict with another extension…

https://github.com/stoically/temporary-containers/issues/347

https://github.com/stoically/temporary-containers/wiki/Autom...

In my case the single largest contributor to my fingerprint is ... canvas size. I run full screen with a custom Firefox setup that basically makes my canvas size unique :/ The "protection" Firefox uses for this is to always open a new window at a default size, which does nothing in my case since my toolbar config still makes the canvas size unique.

It would be really useful to have something that dithers the reported canvas size by 5 or 10 pixels in different containers to add noise there.

Now I understand why I'm getting paywall limits even in private browsing :) I use Tree Style Tab, so my canvas is also of unusual size and ratio. I guess I can try making it more narrow or wider to combat that :)

to defeat canvas size fingerprinting in firefox:

about:config -> set privacy.resistFingerprinting to true

about:config -> create new boolean key privacy.resistFingerprinting.letterboxing set to true

this will set your canvas to a common size which fits in the viewport and display a grey "letterbox" border in the surrounding space.

Doesn't seem to work... reported canvas size is still some odd value (2200x1283x24). I think it uses a fixed size for the letterbox, which is useless. Right general idea though.

I wish them the best. When I last tested it on fingerprint.com, the hash remained stable even with resistFingerprinting and letterboxing from a VPN, only changing between profiles. When I daily-drove resistFingerprinting (not reduceFingerprinting that permits exceptions like dark mode) in 2021, my hash changed every restart.

Perhaps fingerprint.com has stepped up their detection game and have new heuristics to identify you, thwarting the resistFingerprinting measures.

My experience lately has been that fingerprint.com is able to identify my main profile "in bursts", i.e. it will identify me consistently for some days, then it will forget and tell me it's never seen me. Maybe the service they provide on the landing page has a TTL policy? Either way, I've observed this behaviour on both my main profile and my "Firefox Focus"-like profile (a mix of no history + automatic temporary containers). On Mullvad Browser, however, it always seems to group me with random access across the globe.

Unfortunately, Cloudflare and other protections will keep working even less than they used to. I have started to not use Cloudflare protected websites because they don’t work with Firefox. But that is a fight I am going to lose.

Symptoms? Is it limited to when a site has Cloudflare's more aggressive protection turned on? I haven't noticed any problems I've attributed to Cloudflare, and I use Firefox exclusively.

This matches my experience as well. As a FF user, I very occasionally encounter problems, but these don't seem to be correlated to their using CF protections. Much more often I find sites broken that rely on cloud domains with bad reputations, which my DNS filters block.

I was actually wondering if the stuff that Mozilla's talking about here will be used by bad bot people to try to circumvent CF's abuse protections. As I recall from when I was working with them, CF's service relies in part on being able to identify botnet attacks by doing its own fingerprinting.

The only symptom is that captcha never completes.

I have more restrictive protections on. If you use just loose settings, it completes, but advanced fingerprint protection, for example, breaks captcha completion.

This is very known issue.

https://news.ycombinator.com/item?id=35742606

I'm sorry whatever problem you've run into, but it's definitely not true that no cloudflare protected websites work with any Firefox. You've run into something more specific, I guess.

I run exclusively Firefox over known mullvad VPN endpoints and I never have any issues with cloudflare or its captcha.

I guess you don’t have fingerprint resist on:

https://news.ycombinator.com/item?id=35742606

Cloudflare and their crappy widget has done a disservice to the Internet and humanity. They really do need to come up with something else.

This seem sto be the actual list of things it's protecting?

https://support.mozilla.org/en-US/kb/firefox-protection-agai...

They are... surprising to me. And as a developer, some of them seem kind of horrible. Altering canvas data, really?

It's a real "can't have nice things" situation.

It’s a bit annoying that Firefox by default breaks all sites that use canvas imageData API. There is no permission for that, so no user-friendly way to ask for consent either.

Sites such as?

Offline friendly image editors for instance.

I'm already using CanvasBlocker, Decentraleyes, and the NoScript Security Suite; but getting more protections will be nice. Even if it may take a while for them to land in Waterfox.

How is your browsing experience with that stuff? I used to go nuts with anti-tracking measures, but enough of my browsing experience kept breaking that it just didn't feel worth it.

I use LibreWolf at work, and I exempt most internal sites from aggressive anti-tracking stuff, but otherwise it works fine.

It's fine. Sometimes I get annoyed by websites which require JavaScript to show static text (apparently HTML is too difficult?) or which block me with a 'please unblock challenges.cloudflare.com to proceed' (that second one seriously pisses me off when I see it on, for example, the website of the Belgian railways), but by and large I'm fine with just saying 'if it breaks I don't need it'. But I handle my e-mail with isync, mu, and mu4e; and as far as I understand e-mail tends to be a sticking point for those who care for their digital rights. I also don't have Xitter or Facebook or any of that nonsense.

If there's one thing I don't like its the fact that NoScript doesn't integrate with Multi-Account Containers. It would be neat if instead of having to temporarily allow GitHub JavaScript and re-disable it when I'm done; I could just allow GH JS in a GitHub or Microsoft container and it only being enabled in that container.

Libraries documentation that requires javascript to load is the lowest of the bunch in my opinion.

My experience with uMatrix: most sites work right away. Others require fiddling with the matrix of media, script, xhr, frames and the third parties serving them. After a while it's easy to remember which ones must be temporary enabled and which ones don't. Sites with videos are a little more difficult. Sites with payments require care. I whitelist the minimum set of scripts that make the sites I use often work. There are usually many scripts that can be left out. If everything fails and it's a one shot site, I start Chrome.

You are actually easier to track using these addons.

By installing Canvasblocker, Decentraleyes and NoScript you are providing more entropy to trackers and thus making it easier to track you. Imagine how many people worldwide block specifically Canvas, have weird looking network requests to certain js libs and have JS disabled for some (/all) scripts combined with your general setup (window size, font size, and many other factors that do not even require JS).

The Tor project explicitly suggests to not install an adblocker for example because of this.

Without an ad blocker and JavaScript blocker the average website would be 100GB in size and take several years to load. If I really cared about tracking protection I would just not use the regular internet and stick to Gemini. CanvasBlocker is just because the Tor browser itself has one implemented (source: <https://2019.www.torproject.org/projects/torbrowser/design/#...>) so I figured I might as well.

There has to be a happy middle between "no protection" and "complete uniqueness"

The web without ad blocking is revolting. Browsers building in these features makes them more popular.

Aside: Fuck the Washington Post. They have a line in their privacy policy that acknowledges the existence of "Do Not Track" flags in browsers. Their acknowledgement: since there is no industry standard for responding to it, they ignore it.

wow lmao

> Do Not Track. Some web browsers may transmit a “do-not-track” signal. Because there currently is no industry standard concerning how to treat such signals, the Services currently do not take action in response to do not track signals. We respond to legally recognized browser-based opt out signals such as the Global Privacy Control signal for California residents.

https://www.washingtonpost.com/privacy-policy/

Read: we ignore it because it's not legally enforced

The actual industry standard for Do Not Track is to ignore it. It is deprecated. The browser we are talking about in this post, Firefox, removed support for Do Not Track in February of this year.

https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/...

To my credit, WaPo should have said that instead of saying there isn't a standard. The standard is gone.

I was a bit glib with my answer, but I don’t think what they said is wrong. Just because something is a written standard doesn’t mean it is an observed standard. WaPo was clearly referring to the lack of an observed standard. I don’t think they were denying the existence of API specifications, nor do I think that’s a reasonable interpretation. They’re saying that nobody treated it the same way, which is true, and it’s also the reason why it was deprecated in the first place.

I more or less use those addons (uMatrix instead of NoScript) plus uBlock Origin. uMatrix doesn't load a large number of JS files. An example from an ecommerce site I'm browsing right now: the site is functional (at least in browsing mode) without the scripts from

    bigcommerce.com
    classyschema.org
    doofinder.com
    elfsightcdn.com
    google.com
    grit.software
    gstatic.com
    hexgator.com
    klarna.com
    skeepers.io
    criteo.com
    googletagmanager.com

It needs only the JS from the first party domain. So they can track me from there but all the other guys don't know about me, unless they buy data from the first party. At least they have to do more work.

I also don't get advertising in any form, maybe because I don't have ecommerce apps on my phone and I block a lot of things with Blockada, but that's another story.

I don’t really mind first party telemetry. I’m already interacting with the sites, so they can build a nice profile if they want too. But my pet peeves are loads of non functional JS and not having an html render for web content for a non app website.

> (window size, font size, and many other factors that do not even require JS

Yeah, they require CSS, which you can also block using noscript and other tools, if you want.

Also, while you might be more "trackable" to those who fingerprint, if you are blocking those cross origin or same origin scripts from loading you are already stopping some of that. You can even blacklist some known hosts completely in your browser's policy settings and prevent those requests from ever reaching their destination.

Adding noise to images sounds like a really bad idea. It will mess with any Javascript code which performs processing on images. Try writing a photo editor in Javascript and watch your browser corrupt your images.

Like the articel says those features can be disabled on a per site basis.

You are able to toggle these specific named categories:

* Cookies

* Tracking Content

* Cryptominers

* Known Fingerprinters

* Suspected Fingerprinters

But there is no separate toggle for the feature that adds noise to the image, or indication of which toggle would affect that.

It's 'Suspected Fingerprinters' that controls the Fingerpritning Protection feature described in the blog post. But yes, naming and descriptions is hard and never seems to work.

But to disable it on a per-site basis, I would just disable ETP for the entire site. If it's a service or site you use frequently you probably trust them or otherwise have a login to them that makes trying to avoid fingerprinting illogical.

On the topic of Firefox fingerprinting, how does one edit the NetworkID in about:networking#networkid without creating new profiles or user accounts?

It would be nice to see Firefox implement a few features browsers like brave have, like being able to automatically clear cookies for a site when leaving it, and to make containers available when in private browsing, ah well.

This is pretty handy and I've been using it for years[0].

I like the idea of Brave but we have a bigger fight that requires us to have no chromium. Chromium winning is Google winning, allowing them to control the Internet. I don't want that power in any single entity's hands. So I do ask that more people switch to Firefox or Safari as those are the best options to fight back and have decent market shares (even if small). If we lose the internet we'll lose our privacy too

[0] https://addons.mozilla.org/en-US/firefox/addon/cookie-autode...

The real power is in who owns the product. Since chromium is open source. A fork can be made at anytime. For the web there is a lot of common code that is useful to share between browsers.

Your take is quite naïve. The problem is that Google is controlling all the protocols. We're already in a situation where chrome, and consequently chromium, is "the most up to date browser" because Google has a heavy influence in dictating what those standards are. This is, of course, why people fork chromium in the first place, because it gives them a leg up not needing to build everything from scratch and allows them to pull in security updates and new protocols as Google releases them. But that last part is the problem.

So in a way you're right. But the owner is Google as long as you are forking chromium. Because they control to protocols. Maybe they don't own the roads, but does that matter if they get to dictate how all the roads get used and how all the maps are made? You don't need to own the roads to control them

>if they get to dictate how all the roads get used and how all the maps are made?

That power comes from the market share of Chrome, the product. It doesn't come from the market share of blink, the browser engine. If you want to challenge the power of Google you need an alternate product, and not an alternate browser engine. Maintaining an opinionated fork is cheaper than maintaining an alternate browser engine. This allows you to invest more resources into the product itself instead of the browser engine engine.

  > you need an alternate product, and not an alternate browser engine.

Sorry?

The engine is what reads the protocols. It's how the internet is navigated. It is... the car to the road.

  > Maintaining an opinionated fork is cheaper than maintaining an alternate browser engine

I believe I EXPLICITLY mentioned this.

  > This allows you to invest more resources into the product

Incorrect.

It allows you to invest more time into a different aspect of the product. A browser isn't just a UI interface. It isn't just the style of the car. It isn't just the paint job. Chromium browsers are like taking the engine and drive train, chassis, power train, and suspension out of one car and putting it into a new body. Sure, that's a different car and saves you a lot of time and effort but the cars are going to have very similar performance and capabilities.

Let's put it this way. If Jeep was the dominating car company and decided all cars have good suspension and then Jeep, due to their "expertise in cars" got an unequal say in how roads were designed, then they might want to do things like make roads more rough and bumpy because their cars can handle it and a Porche couldn't. Does that make roads a better place? For everyone? No, that just kills Jeep's competition, further embeds their market dominance, and let's them dictate a critical part of our infrastructure.

You're so focused on the trees you've missed that they're part of a forest.

The question that I have not see answered in the many, many forum threads on "browser fingerprinting", is specifically why a user seeks to avoid it

Is it (a) to avoid internet marketing, (b) some other reason or (c) both. What is the "threat model"

If the answer is (c) then is there a belief that a fingerprint collected for marketing purposes may be used for other purposes

I do not use a browser to make HTTP requests, I only send two headers, Host and Connection, unless I need to send more, e.g., User Agent, Cookie, Accept, etc. The vast majority of websites I access work with only two headers. The list of ones that require more is short and the local forward proxy adds them automatically for those sites

For me, the "threat model" is (a) internet marketing

I do not see any ads because (1) the computers I use cannot access ad or tracking servers^FN1 and (2) I use a text-only browser to read HTML. There is no Javascript interpreter, no way to auto-load resources, no way to display images, no way to store cookies, etc.

I have no issue with this information that I'm a text-only web user being revealed to any internet marketer. (More likely I am mistaken for a "bot" as a result of crude heuristics)

On the other hand, if I were using a popular browser to make HTTP requests, one that sends a "common" fingerprint to internet marketers, then this would signal a more viable target for ads and tracking. Popular browsers have default settings that enable Javascript, cookies, images, auto-loading resources, etc.

tl;dr The reasons a computer user has for avoiding fingerprinting may be different. For example, one user might want to "blend in" and "hide", i.e., avoid being "identified", whereas another user might want to "be left alone", i.e., avoid being the target of internet marketers

FN1. Markerters always seem to require access to DNS

I use FF and I paid for NYTimes. I was logged in, yet NYTimes constantly flagged my browser with a persistent captcha I couldn't bypass for months (across 2 different machines). It thought I was a bot because of the privacy features. So I cancelled my subscription using my phone.

Ha - I thought you were gonna say you switched browsers.

I just found a way to bypass the paywall on a web browser when I want to read an article. Which I figured was a easier solution than emailing customer service over a technical matter (never fun).

Is there a reason to force all these bot checks on logged in accounts that are paying you money other than insanity? Surely you could just have a max monthly bandwidth limit per account and just stop worrying about this?

The New York Times is like a microcosm of the publishing industry. They seem to spend the majority of their effort on protecting their intellectual property. I'd rather they use those resources to improve their reporting, particularly about technical topics, but alas.

We just down know from the outside how much revenue they would lose by redirecting that effort though.

That's fair. I was a little bit sloppy with my previous comment; I was mentally conflating their lawsuits about intellectual property with their dark patterns that prevent people from unsubscribing. I'm not sure if it's still this way, but five years ago they were a nightmare to disentangle yourself from.

https://news.ycombinator.com/item?id=23235341

They actually published an op-ed criticizing Amazon for using dark patterns to prevent people from leaving Amazon Prime while they were using those exact same patterns themselves.

https://www.nirandfar.com/cancel-new-york-times/

They probably don’t want you paying once and using that subscription to scrape the website. Which is reasonable.

Again, they have your login cookie and are already tracking what you've seen. Just start captcha'ing after several dozen articles per day.

I don't think there is any value of [x] for the monthly bandwidth usage you could pick that malicious users cannot afford, but legitimate users could not hit.

That's what early warnings are for. It's an easy problem to solve... except by the NYT.

How would a warning fix the problem?

"Hi, I see you've read [x-y] amount of news of new this month, we're going to cut you off at [x]"

What's the correct value of x?

If [x] is greater than or equal to the total amount of news published, then scrapers need one account.

If [x] is less than the total amount of news published, then you have now made it so legitimate subscribers cannot read all of the news.

Also, you have made things easier for scrapers, because they can determine how many accounts they need by dividing the total amount by [x].

Just use Bypass Paywalls Clean. Paying for a subscription is up to you.

I just open dev tools and look at the file in the network tab. You can read it the response sub-tab usually.

when I used to subscribe to the nyt, I had to block a few of their endpoints to kill the awful popups and etc. This, the further ads for paying subscribers, and a host of other issues led me to drop them as well though.

>Having a unique fingerprint means fingerprinters can continuously identify you invisibly

This is not right. If you have a unique fingerprint every time someone tries to fingerprint you, then they have to do extra work to try and figure out which are the same. If you make it always be the same you've made the fingerprinter's job much easier.

Agreed. And this technique becomes more effective as the number of people using it increases. It's easy to match up randomized fingerprints if only one person is doing it, but quite hard when thousands or millions are doing it.

dont use randomized fingerprints, spoof actual fingerprints, randomly.

A good fingerprint algorithm incorporates features and functionality that can't be spoofed because it is necessary for the browser to work correctly.

You can't just make your browser's APIs give erroneous outputs and still expect the browser's APIs to work.

In this context "a unique fingerprint" means that your fingerprint does not match any other user's. When you visit Site A and B you give a fingerprint X that is the same on A and B but no one else on the internet has ever sent.

In contrast a randomized fingerprint mean when you visit A you have a fingerprint X' and on B you have a fingerprint Y' and no one else on the internet has X' or Y' but A and B can't correlate you.

The protections we've put in place first try to do API normalization to make it so more people have a fingerprint X, and it isn't unique. And then they do API randomization so you use X' and Y'.

If a fingerprint goes to extra effort of detecting a randomized fingerprint, and ignore (or remove) the randomization, they will get the X fingerprint which - hopefully - matches many more users.

I dev my private fork of browser fingerprinting bypass and I can tell, this is like 1% of what commercial tracking companies use for fingerprinting.

Unless they tackle all the hidden things, all artifacts, canvas rendering and many more.

These companies will be actually happy after this change, because even users with ublock and other plugins, will think they're not tracked. Yeah, nope.

And it's not that hard to see how they fingerprint your browser, reverse any JS tracking script yourself and see.

I'm still unhappy with the user-agent header. I tried removing information but it breaks a number of sites. Would like to leave Linux in there (if feasible so it gets counted) but remove/spoof everything else.

Breaking websites is about the only thing you're going to accomplish by messing with the UA string. It's a small amount of entropy and anyone who really wants to track you, doesn't need it.

I tested firefox recently. It had some AI summary button or something that was new. I instantly wanted to eliminate this from the UI but I don't know how to do that. I guess it is possible? But it probably requires some time and research; the thing I don't need or want this, it just takes away space.

Then I remembered why I no longer use firefox. I believe we, as users, need to take back the open web. The days of some random developers ruining the UI should really be over, be it firefox, or Google chrome killing ublock origin. We need to fight back.

I use Firefox because it is better than Chrome, which is the only alternative I see.

Do you use something else?

LibreWolf, Iron Fox, and Brave are all worth a look, I think.

Not the commenter you're replying to, but I've been using LibreWolf for the last few months.

It's a bit more privacy focused, so may need some tweaking to your liking (by default it won't persist history, zoom levels, cookies, etc.)

Almost all "alternative" browsers are Chromium based or Gecko/Firefox based. If there are any that are truly scratch-built other than the text-based browsers such as lynx or w3m I'd be interested to hear about them. I'd guess they are extremely limited in features.

The graphical alternatives that I am aware of are extremely limited, such as NetSurf.

You are looking for "ladybird"

I agree with your comment, but to resolve the question it's "browser.ml.chat.enabled". A common topic on HN,

https://hn.algolia.com/?query=browser%20ml%20chat&type=all

> It had some AI summary button or something that was new. I instantly wanted to eliminate this from the UI but I don't know how to do that. I guess it is possible?

Started a fresh profile, but couldn't find an AI button. The AI stuff in the context menu? You can remove the chat bot functionality right there. As for the buttons, if there is an undesirable button, it should be removable via context menu or toolbar customization.

You have to click that button and option to hide is right there.

I feel your pain with the AI stuff, but I think I had one sidebar open one time and I was able to disable it with one click.

Fingerprinting is nearly impossible to resist these days anyways, no matter which technics Firefox uses to reduce it, and sometimes it actually makes the browser appear more unique.

Last time I tried everything I could to prevent Firefox from calling home, it was still requesting Mozilla servers. Though I haven’t given up, my plan is disabling it at source code level and build my own release.

I think this is a nihilistic view. The browser ultimately sends only what the webpage requests. If we gut the ability for websites to request large swathes of information such as every supported TLS Cipher suite and also better protections such as GDPR to make it illegal for browsers to track this information unless a user signs up and also not gating information behind said sign-ups

I couldn't quite catch what you meant, but

> The browser ultimately sends only what the webpage requests.

You should do research before making such claims.

> and also not gating information behind said sign-ups

"People should do work for free" isn't very workable.

I don't think there's anything in GDPR or similar laws about disallowing paying for a subscription with money. It's merely about killing the practice of paying with your privacy for something otherwise labeled as "free".

The quote I gave was the context, not GDPR.

"not gating information behind said sign-ups" was in the context of regulations like GDPR. You twisted that into "People should do work for free" which is not at all the meaning of what you replied to.

GDPR is nothing to do with stopping locking info behind a signup. Not at all.

People lock info behind a signup if they want to to monetise. Something stopping that is forcing advertising as the monetisation strategy, or else saying people should work for free.

Specifically, my concept is not having to fork over personal information to get access to "free content" I could have worded that better, its either free or you pay and if you pay with your information each sale of your information should be recorded and legible

>The browser ultimately sends only what the webpage requests.

You've got 6 layers under your browser before that data is sent -- some of those are useful for fingerprinting. Also, browser behavior and feature sets are not and likely will never be 100% uniform.

> GDPR to make it illegal for browsers to track this information

Unfortunately the internet is global and people outside of the reach of those jurisdictions can just exist outside of the reach of those laws. Consider the existing landscape of malicious internet traffic and scams which are already illegal in almost every country -- they are still a widespread problem.

Browser behavior and feature sets are mostly uniform apart from some specific semantics. Companies like Google should absolutely be anti-trusted out of owning the Chrome Browser and it should be required that because you are "selling a browser" you must inform each user of the transaction of data

When I say selling, I mean that in the truest form. Google chrome isn't free; you just pay with your data rather than your wallet. You as a consumer should retain the full ownership and access/control of your data and that should be legible in a way you can inspect exactly what's going on. I think it would really wake up a lot of privacy folk to what actually goes on with people's data and things like facebooks Onavo VPN is more visible

I'm working on building a bot detection platform to defend websites right now and yeah, without immense effort on the part of the browsers, you are correct. Any of the apis they injected noise into here will be updated in the competant fingerprinting programs as well. The only real solution is to not enable JavaScript at all.

Reply on news.ycombinator.com