For clarity, while the link is to Google Patents (a public service for searching patents), the actual patent is from Juniper Networks, not Google.

I thought this item would be related to that very popular item... I saw it mentioned in discussion.

At work I just recently managed to trip it three times in one day, which I consider a record since in the past I've encountered it at most a few times a week. What's more infuriating is that my queries were far less complex than the ones that tripped it before (trying to find information about some API constants), basically one quoted term and one site: modifier. I can understand if I was querying 24/7 and hogging their servers (in which case a nice "please slow down" message would be much better), but it tripped within a few minutes of, admittedly intense, Googling.

The security device may notify the attacker device that the solution is incorrect regardless of whether the solution is actually correct.

In Google's case, it doesn't do that at all --- solve one CAPTCHA successfully and all you get is another, immediately. If you actually do deliberately give the wrong answers, it does tell you they were incorrect.

I wonder what can be done to stop it from doing that, besides the old tactic of evading IP bans by changing IP --- it's somewhat creepy to think that it's probably capable of detecting that too and banning the entire subnet. On second thought, it might be worth it... if it means I can get thousands of others blocked from Google for a nontrivial amount of time, all the more mouths to complain and maybe force some reconsideration. I am not a robot. I am not a competitor scraping your pages or doing anything else against your ToS. I am just an intelligent human with over two decades of Internet searching experience enthusiastically using your service for the exact purpose it claims to do: to find something on the Internet.

>In Google's case, it doesn't do that at all --- solve one CAPTCHA successfully and all you get is another, immediately.

I have experienced Google doing the endless captcha when using Opera's built in VPN.

The first time I went through about 3 captchas and then I "knew" they were messing with me.

So, now, I just change my location in the VPN, and get back the searching...

> I am not a robot. I am not a competitor scraping your pages or doing anything else against your ToS. I am just an intelligent human with over two decades of Internet searching experience enthusiastically using your service for the exact purpose it claims to do: to find something on the Internet.

Maybe it’s time to stop using Google and use another search engine which actually treat you like a human being, like DuckDuckGo.

That's what I do whenever it comes up. For the longest time I used Google because they had the biggest index, but if they're not keen on letting people actually make use of it, then DDG and the others suddenly look a lot more appealing.

Happened to me too, but not over my normal DSL IP, but while browsing through Tor. As you note, one CAPTCHA is replaced by another seemingly without end. I gave up after 3 or so tries and switched to DDG, which is what I use when browsing over Tor. It's Google's loss, certainly not mine.

Which shows how important it is for Internet health that we have search competitors to Google who maintain their own indexes. No one company should become the sole practical gateway to the Net.

Happened to me as well. As far as I can tell bing doesn’t seem to do that, or at least it is harder to trigger.

I think most people don’t realize this “endless” CAPTCHA is actually an average of three iterations. You’re supposed to stop when you see no squares with X in them. That is, you should click submit even when there are zero squares matching the CAPTCHA criteria.

Once you realize this it’s way less confusing, and in fact it even clearly states the instructions at the top of the CAPTCHA. But it’s super confusing and definitely a usability nightmare.

The goal, as I see it, is to increase the number of round trips to google required to solve a CAPTCHA, in order to increase the cost for those using CAPTCHA solving services.

Not in my experience. It would give me “select cars until none are visible” and it would be replacing the last car by another one for a long time. Then, when there are finally none, it would just offer another captcha, such a as “click on storefronts until there are none”. And so on and on

Their mapping AI isn't going to train itself, and unlike the proofreading CAPTCHA, this one actually benefits Google's bottom line by providing free input for a proprietary data set.

These Google CAPTCHA's are deployed all over the internet, but you mostly won't notice them until you use a VPN. When you do, you get them at what seems to be their strictest setting: multiple rounds of 'click the object that is a car/storefront/road/mountain/road sign/bus/river', where all pictures clicked are replaced via a fade-out-fade-in that lasts about five seconds before you can continue evaluating the new picture.

That happens to me when browsing normally on my home Internet connection (no VPN). I think it's because Google thinks I'm a robot (I frequently don't even touch my mouse and check the CAPTCHA box using my keyboard, or just move my mouse cursor in a straight line to the CAPTCHA), so it appears we've entered an era where you're punished for being too efficient at using a Web browser.

I do not blame Google, they do what they do for their own benefit. However, when I do get to solve a never ending captcha in a mobile app, while being logged with a user account, while buying a cinema ticket—I get really pissed. Not at Google, but at whoever thought that putting a captcha there was a good idea.

Same here. I'm starting to wonder if they're doing that on purpose to profile real users according to how fast they solve multiple captchas, say by measuring reaction times between every click on a "car", or to refine their anti bot algorithms by doing a deeper analysis of real user behavior. Whatever their intentions are, it's frustrating.

My guess is that most of the "automatic" stuff is done using cookies and other fingerprinting techniques. In safari I hit the longer captchas far more often than in chrome, and it is quite possible that it is because of the ad blocking tech inside.

Nope, I've been caught in captchas which kept on going for about 5 minutes until the algorithm finally decided it had enough. Even when there were no tiles to click anymore, I would click ok and then more would appear and it just kept on going and going and going.

That's not the issue here though. I had this happen to me with the old 'type these two street numbers' captcha as well: it would just present a new captcha after the old one was solved correctly. Makes me give up instantly whenever I got a captcha while doing Google searches.