Cloudflare is free of CAPTCHAs; Turnstile is free for everyone
blog.cloudflare.com
blog.cloudflare.com
> those users can immediately understand the issue and make a conscientious choice about whether they want to allow their browser to pass a challenge.
So they will block privacy-focused users. I guess VPNs will be blocked too. It seems they have just removed the captcha part of their system and are spinning it as a good thing to block certain users.
[0] https://dash.cloudflare.com/sign-up?to=/:account/turnstile
Hopefully blocking Tor users will mostly be a thing of the past... :)
edit: Was able to get a mix of a few successes and a few failures upon multiple browser restarts. Maybe I was lucky to avoid a discriminated exit node? I'm not sure if IP address is a factor here.
My Firefox installations are not even Arkenfoxed. I just use couple of privacy oriented add-ons and a few about:config tweaks.
Cloudflare verifications are destroying the web.
I wonder what percentage of collateral damage they're expecting. I know I'm not alone.
They're saying 15% of people don't even attempt to complete the captchas. I wonder how many false positives are they getting with Turnstile. Hopefully less than 15%…
This line of thinking assumes that without captchas, things that were captcha-walled would remain but be unprotected. I think the actual counter-factual is that those things would either require a nominal payment or some other harder-to-spoof-at-scale action first.
I run a service that provides arbitrary compute (Jupyter notebooks) on demand. Without a captcha, there was a period where it would have been overwhelmed by crypto miners to the point that it wouldn’t be available to anyone else.
Generously offering free services that incur costs for the hoster are much harder to solve, no doubt about that. You essentially can't have anonymity because, if you did, any normal user will look indistinguishable from a cryptobro. You'll need to either have them cover the costs or have pseudonymity at best.
But in instances where anonymity is fine, I share the experience from the person you're responding to: just to read resources, to browse read-only material like a normal person, cloudflare is a barrier to the open web.
Anything that has a dynamic backend of any kind (or, let's be real, even stuff that's static-built) will get relentlessly hounded by hackers the very same second it's online on the Internet. Be it spammers trying to sell you dick enlargement pills or questionable supplements, pedos looking for a place they can use to host their shit, botnet operators looking for good connectivity to abuse in DDoS attacks or whatever, targeted attacks against your site by extortion gangs (very common in business), or (particularly if you're active in the gamer/streamer scene) pseudo-"trolls" that just want to cause you harm for the lulz.
The problem is, as I've written multiple times here, that our governments are doing nothing against the bad actors, their ISPs and the countries that allow them to operate. That needs to be fixed, and then we won't have to rely on Cloudflare and friends any more.
It's like saying "we shouldn't have to lock our doors, but the governments are doing nothing to stop the robberies". And the internet is much more wild than the real world, IMO. The biggest obstacles to the safe internet I see are:
* Globality. Aspiring Russian cybercriminal can hack from their own basement with no OPSEC and VPN without any fear of repercussions as long as they don't bother anyone in Russia (and neighboring countries). Before COVID we used to have literal "wanted posters" (as a joke) in our office with names, addresses and photos of known cybercriminals that we could nothing to arrest, because they resided in another (usually Russian-speaking) country. Even in Europe it's not trivial, because Europol has relatively high requirements to start an international investigation and extradition, and "regular" cybercrime doesn't qualify - so one can send malware across the border as long as they want without any (real) fear.
* Velocity: there is no such thing like "bad actor ISP". I mean ok, there are so called bulletproof hostings, but in general the bad actors buy things like anyone else. The typical phishing campaign starts with criminal buying a domain (using fake data), cheap (or usually free/demo) hosting account, getting a letsencrypt certificate, uploading some fake login form HTML, now just send a few hundred thousand emails/SMS and you're done. Next day tweak some things and do the same, just with a different domain. The typical phishing campaign is live only for a few hours. There's no way a lawful country can make a proper legal decision in a few hours. And this is assuming no international cooperation - usually the server and domain are hosted by another country than the attacked country, so we're talking about contacting another country to make a legal decision in a few hours timeframe.
* Censorship fears: Ok, so a proper timely expropriation is not possible. What we have in practice is a list of more and less formal blacklists, like Google safe browsing, (extremely annoying and opaque) spam blacklists, and many lists of unsafe domains (my country provides one). In most cases the lists are DNS based, and they work great. In my country, our list is also applied by some ISPs automatically (so home users are protected as long as they don't change theit default DNS). It would work even better if every ISP applied it, and blocked malicious plaintext DNS requests on the wire, but I can already feel HN readers becoming tense as they read this (I don't like it too). We all hate censoring the internet, and want to preserve the right of normal people to be scammed by visiting a phishing website.
* Money: Having a dedicated abuse team cost money and brings no revenue. Just look at how google does it, and they are drowning in money. Imagine how responsive are smaller providers. In many cases you could as well try to contact /dev/null.
* Privacy: This rant is getting a bit long. I'll just mention that one of my colleagues (frustrated when a known criminal managed to turn off his computer and was let free later, since his disks were encrypted and there were no strong enough evidence to jail him) said "law abiding people have no reason to encrypt their disks". Of course I strongly disagree, but I share the frustration. With god-mode on the internet (the ability to read all the communication, get into any server and take down any domain) I could do so much more to help people in my country. I guess that's also what drives NSA and similar agencies to get more and more power. Unfortunately, we live in a democracy, so we have to make compromises, and I think the compromises we make (i.e. that my powers as someone fighting cybercrime are reasonably restricted) are good ones.
Personally, I think a big issue is that there are no legal repercussions for ISPs/hostings that repeatedly host malware/phishing. Even if they respond to abuse and take down something (maybe even block the account - the horror!), the same happens few days later and there's nothing that can be done. I think financial fines for gross negligence would really help to align the incentives here.
I'm just looking at the dozens of billions of dollars lost to scams in the US alone each year [1]. And that's just scams, not the other forms of cybercrime. And Europe isn't much better off.
(I won't copy your points for a quote since they're too long)
> Globality
Agreed. But Western governments, united, could mandate their ISPs and phone traffic to cut off all traffic from these countries. Most international carriers are based in the US and Europe. Guess how long India would take to dismantle their scam callcenters if cut off? A week tops. Russia wouldn't cave, but I see no reason for this country to be connected to the Internet at all, at least not as long as they are invading Ukraine. And China? They've been running rampant with espionage campaigns for years. It's time to accept this declaration of war and retaliate.
> Velocity
Oh hell yes there are bad actors. Phone providers providing connectivity to scammers and spammers, residential ISPs not acting against abuse reports and thus allowing compromised residential devices (e.g. cheap IoT crap) to continue to attack infrastructure... if I had anything to say, I'd mandate that three credible abuse reports should yield in the disconnection of any Internet participant, and that ISPs were to assist their customers in cleaning their devices. As for domain providers: mandate verification of domain names, and boot off providers that repeatedly violate this requirement. The only thing that reverses profit incentives is serious sanctions.
> Money
See above. Fine providers that don't respond to abuse requests similar to GPDR, up to 10% of yearly worldwide revenue. If they don't comply and show no credible efforts to become a good citizen of the net, cut them off.
[1] https://www.statista.com/statistics/1050001/money-lost-to-ph...
There's the obvious things like public forums or contact forms where spammers submit messages, but "anything that has a dynamic backend of any kind" is just not true. Most things don't lend themselves for making money.
I do agree that we could do better about tackling the abuse that does happen, both by law enforcement and by sysadmins simply banning IP ranges whose abuse center doesn't make any attempt to solve the problem.
I suspect that more people have their blog behind Medium, so that’s even lower hanging fruit.
However, on my phone I cannot get past it. It's just an endless loop. The phone is an older model for the Indian market and the Firefox lags behind. So what, it's perfectly usable on 90% of the sites I want to visit. (Which excludes everything full of ads) I hate today's wasteful computing à la Android.
Edit: The article says there would be a report problem form. I have never seen that.
Seconded.
At least for me, on Tor.
A lot of sites are behind Cloudflare.
The other killer problem of course are cookie-consent banners, and the considerable proportion of sites which have taken to emitting full-page modal dialogs the instant you view the site, or after you scroll down, or when you move the pointer upwards to pick another tab.
The amount of stuff you have to wade through to get to or use a site has reached the point where casual browsing isn't viable, not with Tor, not for me.
Are you seeing more blocks than that?
At this point I'd honestly take CAPTCHA over this bullshit.
I don't know how to vocalize this exactly, but this wording really gets under my skin. It shifts the entire responsibility of being blocked over to the user; it's not Cloudflare blocking privacy configs from browsers, it's not Cloudflare circumventing the notion that user agents have agency to present how they want, it's not Cloudflare blocking people from changing their user-agents: no, it's the user who just decided that they don't want their browser to pass the challenge.
It also trivializes the impact, there's something about this phrasing that's like: "oh, it's no big deal, some users have just decided they don't want to allow their browser to pass a challenge."
I am not the person who decides whether or not my browser passes a check that Cloudflare invented. Phrasing it this way has some real, "everyone has a choice whether to give me their wallet, but people also have the freedom to decide whether or not they're going to allow me let them go without shooting them" vibes.
Bullcrap.
And this is a huge deviation away from the idea of purely clientside checks like Privacy Pass (not that those methods are perfect, but they were at least headed somewhat in the right direction). Cloudflare is basically admitting that they are going to focus on verifying hardware and restricting software configurations, and they want to phrase it like a good thing. It's soft DRM because they can't go all-in on hardware attestation yet, Cloudflare is openly saying, "we're going to restrict hardware categories and software from accessing the Open Internet."
Oh sorry, I should rephrase that. They're going to give users the power to decide whether or not they want their devices they own to be allowed by Cloudflare to access the Open web.
Just a rant / modern web sucks / CloudFlare effective monopoly sucks. There are other providers - please give them a go to diversify the web a bit.
I've let airlines and other companies know that Cloudflare isn't letting me buy their products but unsurprisingly they don't care (or tell me to clear my cookies)
Ovh and Hetzner provide their own solutions. Opinions are mixed there, but they both have big pipes at their disposal.
For one reason, I recently spent months with Cloudflare blocking my Firefox ESR from many sites, including 2 key paid ones for work.
When both companies showed zero interest in trying to figure out why Cloudflare was doing that, even if they lost a customer, I had the thought that this is probably what it's going to be like with a lot more things.
We all know some of the key huge companies that do the math about how many users they're willing to lose/screw, to gain some other advantage or to reduce some cost.
But Cloudflare might be one way that small companies will have no choice but to lose/screw users in an uncaring way, like they were those uncaring huge companies. That might hurt the companies eventually, but it will tend to hurt many users first.
Fingers crossed that at least Cloudflare is rising to the challenge to not become a resented Kafkaesque institution.
We need to stop treating this like it's a civil conversation and put more energy into decompiling and breaking Captchas and other attestation methods, and normalize the idea that breaking these systems is good and should be accessible to normal people.
Because increasingly, breaking the systems is going to be the only way to run really privacy-focused software. It's like adblocking, you don't make progress by negotiating with advertisers, you block ads. We have no negotiating power with Cloudflare and no way to force companies to care about user privacy or freedom unless we have ways to circumvent these systems.
----
I'm tired of trying of entering into a relationship with Cloudflare where they break everything and we send them a bunch of bug reports begging to be let back in. Nah, we need to start decompiling their crap and circumventing it.
So how would this work with financial websites like Stripe? Should they let bots open ready to use their service?
Headless Chromium with some tweaks is already good enough that it's practically impossible to distinguish from regular Chromium, and Google knows it. That's why they were pushing for WEI (even though they were claiming it wouldn't impact browser extensions or debugging protocols). Google knows that environment browser/fingerprinting isn't an effective solution to stop automated requests, because normal user browsers out of the box allow sending automated requests.
Basically the only reason why botters don't already run full browsers is because of computer power, and botnets get around that problem. When attacks aren't limited by IP or compute power, attackers can just run regular browsers and bypass all of these checks. Even that isn't always necessary, a Raspberry Pi 5 is going to be perfectly capable of running a full Chrome instance to send automated requests through.
At least IP rate limits require actual work to circumvent, you need to acquire a botnet or a lot of IPs. Browser capability testing is comparatively easier to get around, you just run a full browser.
----
People have a tendency to lean into whatever outcome they want as if it's the default outcome that everyone else needs to disprove. But the reality is, Turnstile is a new system that is being newly developed as a response to changing conditions around traditional CAPTCHA methods, and it's not any privacy advocate's fault that robots now know how to do OCR.
And "I don't need to figure out how to preserve your privacy" should give you some clarity into why privacy advocates are not very sympathetic when you say "well, we have to start doing fingerprinting now because the robots broke everything, what is the alternative?"
Nah, we don't have to find you an alternative if you have no interest in actually working with us to meet everyone's needs equally. It stinks that robots broke your audio CAPTCHAs, but that's not Firefox's fault and it's not a privacy advocate's job to fix. Particularly not in the situation where you're unwilling to advocate for our needs alongside your own.
So you are complaining that the existing solutions aren't good?
I mean, I'm perfectly fine sticking with existing CAPTCHA audio systems or proof of work. They're problematic but workable; over time we can push for systems like privacy pass that improve them for everyone. But if you want to propose that we go all-in on fingerprinting and capability testing, then what's your plan to mitigate the new privacy risks from that?
Cloudflare: Hey guys, we have a solution for telling humans and bots apart!
Commenter: Your solution sucks!
Me: What would you recommend?
Commenter: Why don't you recommend something?
etc etc.
Right, and what I'm saying is that when a solution is proposed that stinks, it doesn't just magically get the default seal of approval, it doesn't suddenly become everyone else's problem to find an alternative.
Clouflare is proposing a solution with a negative impact on privacy rights and user autonomy. And you're kind of jumping right to, "well, unless you can perfectly solve the problem then we'll go with what they propose." That's not really how solutions or negotiations or anything works.
If your solution to "how do we distinguish bots" is that we'll stop caring about privacy, then I don't see why it's any less of a solution to say, "the way we'll preserve privacy is to stop caring about bots" -- other than that you've arbitrarily decided that the solution to this problem has to prioritize bot filtering over privacy.
The situation we're in is that CAPTCHAs are easier to defeat now than they used to be. Cloudflare is proposing that we get rid of privacy rights and user autonomy to solve that problem. The rest of us are saying, "sorry, it's a nice idea but privacy rights are non-negotiable."
It is not our responsibility to solve Cloudflare's problem, it is Cloudflare's responsibility to come up with a solution that isn't terrible. They're the ones who are saying that existing CAPTCHAs aren't good enough anymore, privacy advocates are fine sticking with status quo.
It's not Cloudflare's responsibility to do anything, really. They could just as well not have released this. As a service owner, I want to be able to distinguish legitimate users from bots. If I can't do that, there's a real possibility that I will stop offering a subset of my service. If someone can find a privacy-preserving solution to distinguishing legitimate users and bots, amazing, but, until then, users will have to choose between the existing solutions and not accessing my service at all.
I don't see how this could be any other way, and complaining about Cloudflare's implementation of a CAPTCHA doesn't do anything constructive, as far as I can see. The problem still remains, and no amount of "your solution isn't good, do better" is helping.
I don't think privacy advocates are trying to be constructive, we're trying to tell Cloudflare that their solution is bad. We're not offering them advice about how to write a novel and this isn't a support group for their developers; we're telling people who are refusing to prioritize privacy that we're not going to prioritize their needs either if they're not willing to care about ours.
> It's not Cloudflare's responsibility to do anything, really. They could just as well not have released this.
Great, I'm on board, let's do it :D
> If someone can find a privacy-preserving solution to distinguishing legitimate users and bots, amazing, but, until then, users will have to choose between the existing solutions and not accessing my service at all.
I don't see what this has to do with Cloudflare independently offering an attestation service. Again, you're jumping right to the assumption that it's our responsibility to solve your problem. It's not.
If you can come up with a privacy-preserving solution to distinguish legitimate users and bots, fantastic. But until then, you'll have to choose between not releasing your service or dealing with bots. We don't want to change the nature of the Internet to accommodate you. If that means you can't launch your service, I do have sympathy but... what are the alternatives? Privacy advocates aren't just going to be OK with having their privacy violated just because it makes it easier for business owners. We built a system around user agency and autonomy, and if you want to make changes to that system, if that existing system as it is today doesn't work for you then it's your job to figure out how to make the changes you need without breaking everything.
> and complaining about Cloudflare's implementation of a CAPTCHA doesn't do anything constructive, as far as I can see
It discourages Cloudflare from launching the service.
> The problem still remains, and no amount of "your solution isn't good, do better" is helping.
Again, I would flip this back on you. Complaining that existing CAPTCHAs aren't effective enough doesn't change anything about the privacy problems and restrictive nature of attestation, and no amount of "but how will we block bots otherwise" is going to help move that conversation forward. It's not any more constructive than telling business owners that they'll have to tolerate bots.
I feel a bit like: I'm sorry, but I don't know what you want me to say. I'm sorry that existing CAPTCHA methods today aren't good enough for you, but it doesn't sound like you have a suggestion about how to improve them without putting people's privacy at risk, and that's kind of a nonstarter. Let us know if you come up with an idea, but I don't know what to tell you in the meantime; you're the one who's saying that audio/image CAPTCHAs that exist today aren't suitable for businesses.
This could just be used on its own as a way to annoy someone. Or it could be used to try to hide some legitimate email that you will receive, like if they are compromising your account somewhere that will notify you of the new login and hope you don't notice in the sea of spam.
How many IP's do would think would be needed to create 10x your current user base in a single day? 100? 1000?
The "Just a moment" security check does not complete - sits there, with the busy icon, "Verifying".
Using Tor Browser, which will be why.
This is normal behaviour for Cloudflare sites.
Making a new circuit sooner or later gets past the check, usually after between two and five retries.
Curiously, I can never view Shopify sites. They always and fully block Tor.
I'll add the source when I find it.
Doesn't this wording imply that they actually do collect this data, but just don't use it for the bot decision?
The goal of CAPTCHAs is for users to prove they are human. It's okay for a CAPTCHA to have multiple solutions that are human like to accept. Is giving up on audio and visual captchas an admission that bot owners have won and they are allowed to freely roam people's sites?
I don't understand how this can work though. Lets say someone tries to password stuff on your site. These CAPTCHAs don't do anything to stop the attacker. The attacker can just automate clicking the box if they get the challenge.
I wonder which competing/opposing interests could provide counter-pressure to this trend, because let’s face it: caring for user interests is a legacy goal for big tech browser vendors.
In few lines they have positioned their product in a much more competitive way than Google reCAPTCHA, highlighting Google privacy pain point.
You do realise hCaptcha[1] is a thing, right ?
Most privacy conscious websites switched from reCAPTCHA to hCaptcha years ago.
It's quite typical in our industry to churn something out quickly, which isn't always suitable, but sticks around for a very long time, before something better comes along. I hope captcha can be relegated to a historic footnote some day.
A pay-as-you-go option for advanced features will be available in early 2024 for self-serve customers.
Users can continue to access these advanced features below our 1 million siteverify request limit.
So it seems it will be not so good for free users.
I emailed you and asked you to take my money, but your sales team disqualified us because we wasn't big enough (reasonable, sales people are expensive).
But I think many smaller firms would like to pay a fair price for it, and have more configurability (logo removal, etc).
If there was a relatively turn-key solution that people could deploy it might help wean people off of cloudflare.
You aren't paying with computation, you're paying with your data.
https://community.cloudflare.com/t/near-infinite-loop-of-hum...
Edit: actually not, I'm full of shit, ignore me
Probably some fingerprinting trick.
The same is done by shady news websites which load a single sentence and ask you nonsensically to “Click to expand Article”.
It's a tricky problem to solve, and I guess it's technically better than nothing, but we really should be focusing on surfacing these permissions in a more user-understandable way. There's friction there -- we want to be able to go to Youtube and click a single button and have a video play. But stuff like video autoplay does not make sense to hide behind a user click that can be anywhere on the page or a keypress that can be basically anything at all.
It's too easy of a bar to cross imo; it breaks certain applications and makes them less reliable, but also makes it far too easy to just have a popup banner with an X, and poof, now you have all the permissions you need.
(You can tell easily because Cloudfare always puts their logo somewhere... and those captchas had none)
Generally, you shouldn't block fingerprinting from inside the browser, it is only a fool's errand that provides a false sense of security while being open to exploits, new techniques, bypasses and broken sites.
If you want to be safe from tracking, run your browser sessions in separate ephemeral containers or even VMs and trash them regularly for example. Qubes OS is at one of the extreme ends of this, but Chrome profiles are sufficient for the vast majority of use cases.
but i'd argue its a bad thing overall; if we cannot use curl/lynx/ect to hit a website to view somebody's squabbling on their blog, to not see the url to an image/video without loading it when you want to, something is amiss.
IMO, JS and other newer components are trying to make a general purpose OS out of the browser, and its a bit too much; which yes leads to needing to handle fingerprinting outside the browser; but really all this to stop the adtech from tracking our asses?
I do totally agree to regularly trashing of profiles; though the accessibility for more folks needs to be there, IE: the local-system automation of retaining plugin block profiles and a few others. just takes time and energy - but in an era where we uselessly burn energy and call it a currency and folks cling to it, well, it's just another wave to drown in