IDK but recapatch hasn’t worked for me on desktop. Since like… forever.
What happens is… I click the squares that contains the stairs. Or motorcycles, traffic lights, buses, w/e. Try to guess whether a sq counts if there are like 2 pixels of railing. Is railing part of the stairs? IDK. Doesn’t matter tho. Google rejects anything I try and displays another captcha. Same result with that one. Fucking endlessly. It will NEVER let me past. I’ve tried some dozens in a row.
I think that is because G can’t ID me to a particular human. So it does not matter how I answer them. “You shall not pass.” G had a patent on “endless rejection” techniques. Instead of an outright block, it presents you with infinite unsolvable captchas.
Someone once analyzed web based recapatcha. Their conclusion was, G isn’t trying to figure out whether you are human. It’s trying to figure out WHICH human you are. It’s identity resolution, not bot detection. If it can’t get a high enough confidence estimate, you get denied. Sometimes with infinite captchas.
Yay. Let’s make this asshole company the gatekeeper to the fucking internet.
Have you considered the possibility you are a robot? /s
Hmmm. It would come as a surprise. I’m willing to consider it tho. :D
Yeah I’ve had better luck with non-G captchas. Many of those work. There’s some sliding puzzle piece one that works fine.
Long run, I wonder if captchas are a dead end. AIs can learn to solve them as well as ppl can. So what remains is to mimic the signatures of a human. Little jitters in mouse movements. Or variations in timing. But AIs can easily learn those too. So we end up with many real humans excluded, while many bots get in.
Someone once analyzed web based recapatcha. Their conclusion was, G isn’t trying to figure out whether you are human. It’s trying to figure out WHICH human you are. It’s identity resolution, not bot detection. If it can’t get a high enough confidence estimate, you get denied. Sometimes with infinite captchas.
I’d be interested in reading/watching if you can recall the source.
Tryin to trawl through my saved bookmarks… it might(???) have been this link. But when I try it now, it’s an invalid URL. It was years ago. Maybe the site or domain wasn’t kept up. So I’m not completely sure that was even the right URL.
That story claims too, the more you are in google’s ecosystem, the easier you can pass the recaptcha. For example, Chrome users get past easer than Firefox.
I do have working a link about the infinite captcha block technique! Blocking via an unsolvable CAPTCHA. Warning, google domain. Goes to patents.google.
IDK but recapatch hasn’t worked for me on desktop. Since like… forever.
What happens is… I click the squares that contains the stairs. Or motorcycles, traffic lights, buses, w/e. Try to guess whether a sq counts if there are like 2 pixels of railing. Is railing part of the stairs? IDK. Doesn’t matter tho. Google rejects anything I try and displays another captcha. Same result with that one. Fucking endlessly. It will NEVER let me past. I’ve tried some dozens in a row.
I think that is because G can’t ID me to a particular human. So it does not matter how I answer them. “You shall not pass.” G had a patent on “endless rejection” techniques. Instead of an outright block, it presents you with infinite unsolvable captchas.
Someone once analyzed web based recapatcha. Their conclusion was, G isn’t trying to figure out whether you are human. It’s trying to figure out WHICH human you are. It’s identity resolution, not bot detection. If it can’t get a high enough confidence estimate, you get denied. Sometimes with infinite captchas.
Yay. Let’s make this asshole company the gatekeeper to the fucking internet.
Have you considered the possibility you are a robot? /s
There are other captcha providers. If only companies would start using them instead of Google.
Hmmm. It would come as a surprise. I’m willing to consider it tho. :D
Yeah I’ve had better luck with non-G captchas. Many of those work. There’s some sliding puzzle piece one that works fine.
Long run, I wonder if captchas are a dead end. AIs can learn to solve them as well as ppl can. So what remains is to mimic the signatures of a human. Little jitters in mouse movements. Or variations in timing. But AIs can easily learn those too. So we end up with many real humans excluded, while many bots get in.
I’d be interested in reading/watching if you can recall the source.
Tryin to trawl through my saved bookmarks… it might(???) have been this link. But when I try it now, it’s an invalid URL. It was years ago. Maybe the site or domain wasn’t kept up. So I’m not completely sure that was even the right URL.
https://hfet.org/google-recaptcha-privacy-nightmare/
I searched just now and found a few sites mirroring the claim. But they did not have the technical breakdown. Here’s something from The Register.
That story claims too, the more you are in google’s ecosystem, the easier you can pass the recaptcha. For example, Chrome users get past easer than Firefox.
I do have working a link about the infinite captcha block technique! Blocking via an unsolvable CAPTCHA. Warning, google domain. Goes to patents.google.
https://web.archive.org/web/20201114190040/https://hfet.org/google-recaptcha-privacy-nightmare/
Oh thanks.
That’s def not the one I was thinking of then.