How true is this?

It's true. But Google couldn't care less about you posting loli on Vietnamese pottery forum.

100% true

disabling 3rd party cookies breaks captcha but they don't actually need cookies to track you. They use cookies t o determine how difficult a captcha to give you.

i don't doubt there's something of this nature going on. while mass data mining is a largely a social issue that most people have no awareness of, you can usually avoid contributing to the system by blocking third party scripts.

>you can usually avoid contributing to the system by blocking third party scripts.
Yeah try blocking captcha (Without fucking paying chinkmoot, retard) and see what happens.

most sites use recaptcha only for things like confirmation of a booking or creation of an account. nothing significant can be collected from few isolated captcha solving cases.

Jow Forums is an exception, since the volume of captchas solved contributes to Google's AI and other social experiments greatly, which i can only assume Google helps maintain financially in return.

You are offered the choice to opt out of this scheme, either by purchasing a pass or by ceasing to post comments on this website. Whining about it only makes you a hypocrite.

Collected data is always potentially significant.

don't move the goalposts. a single use of recaptcha on some service can only tell google that you're using that particular service. and yes, that is significant to the user profile they have on you. however, in most cases this is unavoidable because failure to use recaptcha results in failure to use the service. this is not the point i'm making though.

recaptcha data collected from Jow Forums is different in that it's not used to prove you're not a robot and let you off the hook, oh no. it's used to prove you're an obedient automaton, for which you get rewarded with more complex captchas. the average 4channer spends around a couple of hours a day, solving captchas one after another. this means the more captchas a user solves, the more willing they are to solve more of them, often resulting in more complex ones. this means Google can determine the threshold of complexity they can afford to set in the subsequent captchas without causing the users to quit in frustration, which maximises the efficiency of their experiments.

it's AI training as much as it is gamified obedience social engineering.

You're late to the party, they can easily check the timestamp of reCaptcha and trace back to your post.
So if you have ever used any Google services, your Jow Forums ghost account is already linked to your profile.

just accept that google is skynet and decide what to do with your life
you don't have to use google, you know.

Through HTTP referrers they can see the thread you're posting in, your location, and what time you post. It's easy to infer which post is yours, or at least narrow it down to a couple posts

>Cookies are per-site
>what are third-party cookies
You're an retard

nah captchas stopped training AI a long while back. The captcha difficulty is directly correlated to the stored age of tracking cookies. They do directly connect their profile on you to your Jow Forums posts though and that is inherently valuable (at scale).

Soon™

Attached: 1540239497645.webm (482x542, 691K)

Yea it's true
Just block gaagle cookies through umatrix or somethin

umatrix is a special case cause it allows sites to succeed at SETTING cookies so certain code appears to work to their scripts but then disallows reading them back. Not sure how effective this is as a strategy but maximum possible tinfoil seems to break captcha every other day.

>nah captchas stopped training AI a long while back
got a source to back up this claim?

>The captcha difficulty is directly correlated to the stored age of tracking cookies
how does that contradict what I said? the more captchas you solve on Jow Forums, the more (or more difficult ones) you get given to solve. the tracking method itself is irrelevant.

>They do directly connect their profile on you to your Jow Forums posts
I doubt that. You don't need to make a post to solve a captcha. I don't see why they would care about your shitposts in the first place.

>and that is inherently valuable (at scale)
I also highly doubt anyone, let alone Google, sees any kind of value in the content of Jow Forums posts, at any scale.

they stopped training for/against values in the captcha, for instance choosing a bus vs. things that could be confused for a bus to tighten the recognition. Their incorrect image selections are static enough and unrelated enough to not be valuable at all for training. They do update the set over time but that's probly to stop other bots from training against it.

Captcha difficulty is EASIER after a longer period of time, not more difficult. They recognize who you are and thus you can't be a robot so they don't need to spend extra resources to test that fact.

The captcha does send extra data, as any request would like said, also all of the posts are archived so even a timestamp and board alone would be enough for a non-google entity to identify you through.

It's not so much the value of the content but the associations of content to groups of users. Market penetration (look at intel/amd/phone threads) and Jow Forums alone is a self contained focus group.

Partly. Google associates those who have/are logged into a Google account with lower CAPTCHA rates on their services (such as Google Sorry/YouTube), but it doesn't extend this point of reference to external websites such as Jow Forums. It assigns difficulty based on IP and ISP; higher CAPTCHA challenges solved correctly = higher difficulty. Sometimes the difficulty gets so high it asks you to "please try again", even when you did everything right and it just wants you to do multiple ones to be extra sure. ReCAPTCHA solves are not assigned by IP though, so you could change to an easier IP if the challenge isn't able to be solved in one attempt, then change back and POST whatever it is your sending (e.g: Submit button).

>Captcha difficulty is EASIER after a longer period of time, not more difficult. They recognize who you are and thus you can't be a robot so they don't need to spend extra resources to test that fact.
very strange. in my own experience, ever since click captcha was introduced to the point when i bought a pass, the exact opposite has been true, as explained in my reasoning here there's no point in debating this with you further if you disagree on such a fundamental point of the conversation.

theinquirer.net/inquirer/news/2453931/bake-your-own-cookies-to-crack-googles-captchas-say-researchers

At a certain age you get increased chance of not needing to solve a captcha so people are aging their own cookies en masse to completely bypass them.

>Captcha difficulty is EASIER after a longer period of time, not more difficult
there's 2 different ways to read this.

1. Captcha is easier to solve after a longer period of time of NOT solving it, resulting in old cookies, which is consistent with my experience, my reasoning, as well as the article you linked to.

2. Captcha is easier to solve after a longer period of time spent dedicated to solving multiple captchas, resulting in new/young cookies, which is the opposite of my experience, opposite of what the article says, and is what you're seemingly suggesting by going against what I'm saying.

I guess multiple attempts would always decrease confidence values, I don't think the paper referenced by that article was what I am remembering anyway. But just having an old tracking cookie gave added confidence to captchas iirc. Don't remember enough specifics to say for sure what their results were except that just having a store full of old cookies was a low cost attack on it.

all of which backs up what i'm saying. maybe you misunderstood my point that more frequently occuring / more difficult captchas aren't used to test whether you're a robot at all, but are rather spring into action when Google knows for a fact you're a cooperative human?

>ever since click captcha was introduced to the point when i bought a pass, the exact opposite has been true

Yes and no:
A high number of capcha attempts in a short amount of time triggers tests.
But over time, you get more "instant checkmarks" as your cookies get older and their profile of you grows. Being logged in on a google acc helps a lot with getting instant checkmark, most likely because it builds the profile.

>more frequently occuring / more difficult captchas aren't used to test whether you're a robot at all, but are rather spring into action when Google knows for a fact you're a cooperative human?

Seems like it's the opposite, if you're a "cooperative human" buying into the botnet you'll get mostly instant check marks. The easiest way to get lots of tests (and slow loading images) is by not having cookies, using VPNs and possibly other privacy enhancements.

Attached: thiswasatruestarwarsmovie.jpg (491x341, 21K)

by "cooperative human" i don't mean using Chrome and being logged into Google account. I mean solving captchas continuously for an hour or more, straight. i don't doubt that this is yet another incentive to "buy into the botnet", but it's not the point i'm making. my point is, whether you're logged in or not, botnet browser or not, the more captchas you correctly solve, the more complex and/or frequent they get. which leads to my proposal that this isn't used to determine whether you're a bot, but rather used to conduct social experiments on a cooperative live specimen.

How difficult is it to train a NN that can pass google captcha? I am assuming that a few corporations have datasets that are comparable to google and have nets that can pass captchas quite reliably.

Captchas have gotten harder because computers have gotten better at solving these sorts of problem. It's not a static battle; if you succeed, they'll make captchas harder until your bot fails.
Good news is they'll probably offer you a job.

I don't have to solve any captchas at all, google just lets me through

They probably know who you are anyway if you have an Android device on your WLAN

Well then what would be the situation that makes the system think you're human but not cooperative? Which should result in almost only instant checkmarks. Clicking random images in frustration? Aborting?

When you had to solve a lot of captchas, were you using privacy enhancements? Or posting a lot?

I'm full bonnet and I get almost only instant checkmarks, except on 4g where I have to solve simple captchas maybe 30% of the time, which makes sense since my IP is changing.

It wouldn't be THAT hard, but captchas change so you'd have to do it over again a lot and using third world labor is cheaper.

Or just that its an adversarial response to multiple captcha attempts that would not match any site pattern except bullshit like 4chaan.

Older versions of image captcha could apparently be beaten by feeding the 9 images and the hint image back into google image search to get a text descriptions. The description could then be used to solve the problem.
As long as you can access some API that can add a description to all 10 images, you do not have to train a NN yourself.
Some people have noted that the difficulty increases with the number of solved captchas. This is probably because the system suspects the user to be a bot, which makes it obvious that there are probably many bots who can reliably solve image captchas.

So that's why google made their reverse image search suck.

>disabling 3rd party cookies breaks captcha
no it doesn't.

In chrome you can't disable cookies from google, FYI.

>how to deal with normalfags 101

and i never have any problems with captcha

Attached: 1531141807995.png (1113x729, 62K)