Comment by gruez
3 days ago
>I increasingly encounter outright blocks rather than any sort of captcha when visiting cloudflare "protected" sites.
???
Unless you're browsing around with the googlebot user agent string, you should be getting turnstile challanges at most, not blocks. And if you're getting a turnstile challenge it's unclear how it's different than an anubis challenge. If you're outright blocked, it's probably a site decision (eg. block all VPNs or block everyone not from a given country) rather than cloudflare's.
Try using Firefox with "resist fingerprinting" enabled, on OpenBSD, if you want an idea of how insidious Cloudflare's increasing prevalence is for the openness of the web. Inexplicable 403 responses for entire domains for no apparent reason, no "captcha", just blocked outright. Heck, it happens on Linux too, and even without "resist fingerprinting" enabled.
Happens with older chrome browsers as well. Last supported windows 7 version. Started this month.
> And if you're getting a turnstile challenge it's unclear how it's different than an anubis challenge.
You are guaranteed to pass an Anubis challenge eventually [0], whereas it's possible to get stuck forever in an infinitely-looping Turnstile challenge.
> If you're outright blocked, it's probably a site decision (eg. block all VPNs or block everyone not from a given country) rather than cloudflare's.
Cloudflare blocks legitimate users itself sometimes [1].
[0]: Unless you run into a bug, but Anubis is open source, so you can always submit a patch upstream. I've done this myself, and I can confirm that it's relatively straightforward.
[1]: https://news.ycombinator.com/item?id=43329320
>You are guaranteed to pass an Anubis challenge eventually [0]
That's a double edged sword because bots will eventually get through too, and unlike humans, their time is dirt cheap.
>Cloudflare blocks legitimate users itself sometimes [1].
I never ran into this issue despite using seemingly maximally suspicious configs like tor browser. I can't say the same for some other vendors.
> That's a double edged sword because bots will eventually get through too, and unlike humans, their time is dirt cheap.
Yeah, I really have no idea why Anubis works right now: residential proxies are far more expensive than compute, yet the bots seem to have no problem obtaining millions of residential IPs, but they give up on even short-ish Anubis challenges.
> I never ran into this issue despite using seemingly maximally suspicious configs like tor browser. I can't say the same for some other vendors.
Yeah, I don't like the Cloudflare challenges, but in the past 5 years I've only had it outright block me once, and that fixed itself after 15 minutes. And I use Firefox on Linux with various privacy extensions, so my browser probably appears at least moderately suspicious.
Whereas I've been trapped in impossible ReCaptcha loops quite a few times, which is still better than vague error messages that magically go away when I switch to something not running Linux. So I'll begrudgingly accept that Turnstile is the least user-hostile product on the market right now.
2 replies →
The bots that are problems right now are not the ones that run Anubis challenges. There is no universal anti-bot, only ones that work at certain times.
> unlike humans, their time is dirt cheap.
On the contrary I only want to visit a few pages on a given site and have an entire laptop at my disposal. Meanwhile for bots efficiency is key. A serious scraper (ie the type of actor that actually causes material problems for site operators) is performing tens or hundreds of pages loads per second per core spread across thousands of sites. Making a single page load take even half a second of cpu time is a massive win for the site operator.
I’ve been blocked by Turnstile without possibility of challenge, no way to access nor know why. And it was with standard iOS and Windows machines and no VPN or anything. It happens.
Yesterday Cloudflare told me I was a bot and couldn't access this.
yeah, googlebot user agent thing was a thing maybe around 10 years ago but I don't think it's working today. Maybe if you start the http session from gcloud, it might still work.