Because TBB has javascript on by default, turning it off increases your signature. It would be better if TBB defaulted to js off, with a front panel button to turn it on.
JS also dramatically improves security. TBB is stuck in a 90s mindset about privacy, as if Firefox exploits were not dime a dozen. Especially with AI making FF exploits more available, we can expect many tor sites to be actively attacking their visitors.
Tor endpoints are pretty easy to identify, there are plenty of handy databases for that, using it to begin with increases your uniqueness. If noscript was set to strictly disallow javascript by default, that decreases the degree to which it increases your signature relative to the baseline of using tor.
Then we have to account for the simple fact that many, many fingerprinting techniques rely on javascript, so taking them out of the picture reduces the unique identity that can be gleaned.
Are we absolutely, positively sure that the tradeoff is worth it? Without a strict repeatable measurement, I think I'm highly skeptical about whether or not a default of "allow" is a net boon to hiding your identity. I remember the rationale about the switch mostly being directed towards "most of the web is broken otherwise and that's bad."
Every server knows that you're using tor, we're only talking about whether they can match your traffic to you repeatably, and particularly across sessions, which then enables traffic analysis that can lead to complete deanonymisation.
If TBB changed to js off by default that signal would be less evident, and also, fingerprinting would be harder.
Disabling JavaScript actually greatly increases your fingerprint as not many users turn it off, so that instantly puts you in a much smaller bucket that you need to be unique in. Yes, not having JS means it limits your options for gathering other details, but it also requires much less effort to be unique now without JS.
Tor Browser also doesn't spoof navigator.platform at all for some reason, so sites can still see when you use Linux, even if the User-Agent is spoofing Windows.
> increases your fingerprint as not many users turn it off
We're talking about users of the Tor browser, and I'd be very surprised if this was the case (that a majority keep JS turned on)
Basically every Tor guide (heh) tells you to turn it off because it's a huge vector for all types of attacks. Most onion sites have captcha systems that work without JS too which would indicate that they expect a majority to have it disabled.
> Disabling JavaScript actually greatly increases your fingerprint as not many users turn it off, so that instantly puts you in a much smaller bucket that you need to be unique in.
I've heard a handful of people say this but are there examples of what I would imagine would have to be server-side fingerprinting and the granularity? Since most fingerprinting I'm aware of is client-side, running via JS. While I expect server-side checks to be limited to things like which resources haven't be loaded by a particular user and anything else normally available via server logs either way, which could limit the pool but I wonder how effective in terms of tracking uniqueness across sites.
In addition to server-side bits like IP address, request headers and TLS/TCP fingerprints, there are some client-side things you can do such as with media queries, either via CSS styles or elements that support them directly like <picture>. You can get things like the installed fonts, screen size/type or platform/browser-specific identifiers.
I have my problems with that argument. Yes, less identifying bits means a smaller bucket but for the trackers, it also means more uncertainty, doesnt it? So when just a few others without JS join your bucket eg. via a VPN, profiling should become harder.
> Motivated to understand the immediate physiological response to saunas, we looked at the same-day effects across ~59,000 daily records from 256 users.
I recently flashed GrapheneOS on a Pixel for a friend. I was very surprised that you can do this entire process from the browser using WebUSB - the only downside being that it required me to launch Chromium.
You can flash GrapheneOS on a Pixel from another pixel, no pc required at all. I've done it several times, this is what sold me on the utility of WebUSB. You can use GOS' own distribution of chromium, Vanadium, if you have a GOS device and you want to avoid Chrome.
Is there something specific in that process that required WebUSB vs just normal USB? Sounds like phone makers could have done this since forever if they wanted to, what makes WebUSB particularly useful for this?
Native android apps can talk to regular USB devices, if granted the necessary permissions. But it's exposed through a Java api (and Kotlin I suppose, these days), which is fine, but it means you need to write your client logic twice. If you target the web, you can do it once.
(Yes, you could try to bulid some common interface, libusb-style, but I think you'll have a bad time with minor behavioural differences, especially around permissions. libusb itself does ostensibly support Android but there are several caveats: https://github.com/libusb/libusb/wiki/Android#does-libusb-su... )
Cross-platform compatibility comes to mind. WebUSB is available on macOS, Windows, and Android; a native Android app would pose a bootstrapping problem for a probably not insubstantial fraction of all potential users.
Web USB and Web Bluetooth are amazing. I've used the former for the excellent Web MiniDisc [1], and the latter to flash custom firmware [2] on cheap Xiaomi Bluetooth LE thermometer/hygrometer devices that Home Assistant can pick up.
Truly opening new possibilities, since I wouldn't have been comfortable running some sketchy script or local binary.
Comments like this scare me. Things look amazing when people with benevolent intentions are making interesting things, but as soon as someone with malevolent intentions does something that becomes the reason we can't have nice things people will start asking if this is something we should have actually done.
I just have no faith in humanity, and do not understand why we think this is a good idea to give a browser this much access to local system resources.
There isn't much to fear here. Web Bluetooth has been around nearly ten years now and nothing monumental has sprung forth from it. It is wonderfully convenient to have at your fingertips, especially in the ChromeOS world, but it's not gonna turn everyone's devices into Flipper Zero targets.
Sorry to hear that. I thought this was a safe space for hackers to express enthusiasm about pushing their own hardware and software further (and in this case even in a comparatively safe way).
> I just have no faith in humanity, and do not understand why we think this is a good idea to give a browser this much access to local system resources.
The browser already has all that access, it's just further granting it to web apps, and on a page-by-page, device-by-device, explicitly user opt-in basis at that.
And as I've mentioned, the alternative here is to install a potentially untrusted native application that gets the same access and so much more.
If that's what the Github page tells users to do, many of them will just do it without thinking twice. Is that better?
> I thought this was a safe space for hackers to express enthusiasm about pushing their own hardware and software further (and in this case even in a comparatively safe way).
Nothing is preventing said experimentation nor discussion of it. I am merely offering my more conservative views of the situation as a contrast to the echo chamber gungho nature of the experimentation. Just because we can doesn't mean we should is often left out of the conversation. At some point, the net negative that comes from the use of something "cool" is never contemplated by those creating the something "cool" simply because they would never fathom using the "cool" for "uncool" purposes. Sadly, someone else will and weaponize it in an uncontrollable manner. If the creators can't think of how it can happen, it is vital that those not so involved in the creation speak up when there are potential issues.
I wouldn't describe it as "conservative" but as "pro-native-apps and anti-web-apps", which seems irrational in this day and age where "native apps" means platform lock-in by monopolies, less sandboxing and user-control than on the web, much more gatekeeping and control over published binaries, and these days the web app is usually a more private/secure alternative to the native app (which also bundles a marketing SDK, now, and fingerprints you invisibly via iCloud Keychain, tracks you with various identifiers, and more).
If native platforms removed USB or Bluetooth, the "control over my own hardware" crowd would flip a table. I just wish they also understood the benefits of the web compared to native. The Chrome/Project Fugu team's dream of making the web platform as powerful as native platforms is the correct one from a user freedom standpoint, or at bare minimum a "user choice" standpoint.
I'm not saying pro-native-apps outright even if that might be what it gets boiled down as. I'm saying I do not trust anything that runs in a browser. I actively block as much nonsense as possible. I do not trust devs that write code to run in browsers. There's a lot of devs getting taken out in the blast radius, but the only way to be sure is to take off and nuke it from orbit. There are devs out there hell bent on writing malicious code. I am willing to take a stand and refuse to use things when the net result is negative. I do not use social media. I do not shop at Walmart. These are the decisions I'm willing to live with even if it makes life slightly less "easy" because I've made a moral decision to not open myself up to nonsense just to later ask "what happened...".
Sure, you do what works for you. But why advocate for even more limits to how other people use their computers? One person's nonsense is another person's treasured hobby.
Yes, bad actors exist, but why concede every single nice thing to them?
again, net negative is being glossed over. whatever good and nice things there might be, if it is being used more for negative purposes, you need to consider is it worth it at all or was it rushed and needing more thought before the PoC was pushed to prod
How can you tell that the negatives were glossed over? Just by WebUSB's eventual launch? Have you considered that different people might have different value preferences than you?
The web and native app platforms have very different security models.
Nobody is vetting websites for you. There is no guarantee the same company operates a website today that did yesterday. There is no obvious distribution or regulatory authority instituting penalties for illegal actions (and often is no legal presence in a country when illegal actions take place).
That means for the web, every consent prompt has a large, sometimes even unbounded amount of harm behind it if the user picks incorrectly, and browsers have limited capacity to help them pick correctly outside of reactive block lists once substantial harm has been done and recognized.
This is why, for example, the major browsers have all moved to restricting web extensions behind their own review processes/stores, and put restrictions that make unaudited web extensions difficult to install outside of development workflows. The risk is just too great.
Chrome pushed many of these API early in the Chromebook product cycle, because their idea was that you would only build apps using web technologies. I somewhat doubt they would have pushed for WebUSB themselves if Chromebook started in its current state, where it primarily runs android apps and is about to transition to be android-based.
> The web and native app platforms have very different security models.
Yes, and as a result, the web is much more sandboxed than native app stores (which are mostly based on the illusion that vetting apps can somehow achieve better security than minimizing what resources apps can access in the first place and making access more fine grained).
This is exactly why I'd rather run e.g. shady USB aftermarket firmware flashing apps in my browser (where I know they can at most compromise the device I'm flashing) than as a native app (where USB access is the default and requires zero permissions to be approved).
> This is why, for example, the major browsers have all moved to restricting web extensions behind their own review processes/stores, and put restrictions that make unaudited web extensions difficult to install outside of development workflows. The risk is just too great.
Web extensions very often have access to your complete browsing data, including all cookies. That's orders of magnitude more risky than access to an explicitly selected USB device, in my view.
> I somewhat doubt they would have pushed for WebUSB themselves if Chromebook started in its current state, where it primarily runs android apps and is about to transition to be android-based.
Android has an USB API as well, and if Google only wanted "apps" to have USB access, nothing was stopping them from making Web USB "Chrome App Store" only.
> where "native apps" means platform lock-in by monopolies, less sandboxing and user-control than on the web, much more gatekeeping and control over published binaries, and these days the web app is usually a more private/secure alternative to the native app
Please add “mobile and/or proprietary” before “native apps”. Linux and BSD on PC are still very much free. The web as a platform is just a NIH effort.
>Sadly, someone else will and weaponize it in an uncontrollable manner.
Except it isn't "uncontrollable". You have to explicitly allow every single website to use WebUSB. Without that explicit allowance, the website can't access anything.
Plenty of things can be weaponized, even household utensils. Should we ban all forks?
The sky is not falling, and WebUSB is not going to cause it to fall.
Sorry, no, I am not supporting Windows, MacOS, iOS, Android or multiple other platforms when I could just target one single platform - the web browser.
> those creating the something "cool" simply because they would never fathom using the "cool" for "uncool" purposes
I can definitely imagine a ton of things going wrong with Web USB, and I think the spec authors did a pretty good job at bolting everything down that can be, while still shipping something actually capable at providing USB access.
And that's my point: Sure, fewer capabilities are always safer than more capabilities. But some capabilities are nice and arguably worth the risk, especially if the obvious alternative (blindly installing native applications) isn't much safer.
Sure, but some people are concerned about any website being one confirmation prompt away from being able to have full access to hardware in the user's physical environment, and being able to permanently change the behavior of that hardware.
A hacker may think such things are convenient for them, but an end user does not know the ramification of a random website (WebUSB IIRC still does not have origin restrictions) getting hardware access - nor can we categorize the risk in order to protect them.
What physical access and what permanent behavior changes in particular are you concerned about? Most common "dangerous" USB device classes are explicitly excluded in Web USB.
I've heard about rogue keyboard firmware, but that requires having a programmable/updatable firmware keyboard in the first place. And that closes the loop of my argument: People that want to update the firmware in their keyboard will do so, whether it's in the browser or by installing a potentially shady and not at all sandboxed third party application.
At least in the browser, permissions are time limited and scoped to explicitly granted devices.
> WebUSB IIRC still does not have origin restrictions
The most important USB thing I have are storage devices. Keyboards/mice/etc are much less of a concern. If something rogue happens to a drive, that's a "major problem in Australia. Please help us stop it" situation.
It's a good thing that history has shown us that things have never happened that were designed not to happen. Sure, my tinfoil hat is securely fashioned, but I've been around long enough to see things get subverted even if it's not until long after release.
You can press a simple button on a webpage and it will install malware on your iPhone. Plenty of exploits have been out there for a long time.
Should we disallow clicking on anything on a webpage too?
WebUSB is no more risky than any other tech. You have to explicitly opt-in to use WebUSB on any site requesting access to it. And I'm sorry if someone's grandfather trusts a malicious website and gets hacked, but that isn't a reason to prevent the rest of us from using tech that enables functionality on non-malicious websites that serves a useful purpose.
“ I just have no faith in humanity, and do not understand why we think this is a good idea to give a browser this much access to local system resources”
As opposed to dodgy windows-only installable software from
some weird site to flash devices instead? I’ll take my chances with webusb, thanks.
I love that, in order to reprogram my very expensive model locomotive, I not only have to buy a special device (ESU LokProgrammer) but I can also only use said programmer with windows.
Yeah I could have used some other decoder, but they all have limited functionality compared to Lol sound
There's no real reason a future iteration of the LokProgrammer couldn't run a simple webserver, or connect to Web USB.
This also all applies to most PLC programmers. Proprietary crap you have to run on Windows, just to build ladder logic
What if we implement them but hide them deep in the settings or as experimental feature inside the hidden developer menu, behind multiple warning messages and password prompts? Only the very determined developers and advanced users would be able to unlock them. Then it's safe enough?
Users will unfortunately click on absolutely anything that a trusted (deservedly or otherwise) source tells them to, and you won’t be able to reliable convince them otherwise with UX alone. This includes all “developers only”, “click 5 times” etc. UX interventions.
You have to decide whether the feature warrants the remaining risk after all mitigations, or at least exceeds other, simpler attack vectors.
I think in this case it does, but it’s not an easy decision and I can understand most opposing positions as well.
Thank you. I was confused that nobody pointed out that this "zero downtime migration" requires there to be no write to the original host for the time of the migration...
The subtle difference is between American parking minimums imposed on property owners - “you must reserve space on your private property for this many cars whether you own them or not” vs Japanese parking requirements imposed on car owners - “you must reserve space on some private property for your car if you want to own it”
I don’t believe this is a meaningful distinction when we’re not going to agree on how to judge performance of software engineers. If this were solely about income, it might be an important distinction.
The article assumes a normal distribution, making the distinction moot
But it is useful to question whether that is true in all cases. The cases that aren't normal-distributed might be exactly the cases where it pays off to be neither average or median
there is a major shortcoming in this assumption; everything we've seen related to the internet and technology in general suggests there is rarely a normal distribution. I think it's way more valuable ato frame the questions as a long tail (pareto) distribution and a "good enough" cut-off point.
reply