Captcha

I once heard the concept of captcha described as:
>"Forcing the user through an ordeal to bypass a technical design flaw programmers are too inept to figure out how to correct."

How does Sup Forums feel about this description?

Are captchas a plague we should be working to be rid of or a necessary evil?

i'd prefer to generate proof of work beforehand and have it autosubmit but that'll never happen

shut up Goy and keep on training that Google Deep Learning™ driving algorithm.

What you're asking the programmer to do is to define down the letter what separates a machine from a person, and somehow make it something that all people across all cultures have in common and that can be instinctively known by the computer before a user inputs anything.

In short it's a flaw, yes, but asking a programmer to fix it is like asking a philosopher to define the very nature of what makes us human.

Based on how spammy Web 2.0 was I'm pretty sure they're a necessary evil as of right now. Sure having to find bridges sucks, but it's still better than 3 threads on Sup Forums reaching 500 posts with literal viagra advertisements.

I'm surprised no one's taken it the other way yet.

Build a script that spends every waking moment utilizing 100% of a system's power to feed Google INCORRECT Captchas. Just to completely fuck with the way google builds that shit.

Agreed. The solution is some kind of account/reputation system. It can still be anonymous.

Personally hate having to select all the canoes when it's a fucking kayak, all the parachutes when it's a fucking kite.

>canoes when it's a fucking kayak, all the parachutes when it's a fucking kite

This is a great argument against . If Google is learning anything based on captcha, a good chunk of that shit is going to be incorrect.

This is the most cancerous thing I've ever experienced

Trips confirm the captcha-crusade begins today.

>Web 2.0
Retard.

How about just an IP-based reputation system handled by your ISP?

If you openly spammed the administrator of a certain domain could report you to your ISP who would then give you a strike. After 3 strikes you would have to do catpchas when you wanted to post shit for a while.

You don't have control over ISPs.

>some kind of account/reputation system
Easily abused. Army of bots will work harder towards improving its reputation than army of indifferent humans.

>technical design flaw programmers are too inept to figure out how to correct
Sounds like a load of bullshit to me. Being unable to tell machine from human operating a machine is not a technical flaw.

>Being unable to tell machine from human operating a machine is not a technical flaw.

Isn't it though? Isn't it inherently up to the computer to know whether this message is being typed out on a keyboard rather than copied in from some random text generator?

What's kinda' cool about this topic is how far human emulation has come for computers compared to human recognition.

>Isn't it though? Isn't it inherently up to the computer to know whether this message is being typed out on a keyboard rather than copied in from some random text generator?
1. No, the message comes from the network, and there are no tags attached to it that tell about what source was used to write the message.
2. If there were those kinds of tags, there's nothing to stop the attacker from sending fake tags. There could be, with a totalitarian control over every computer that connects in the Internet, with cryptography modules on every devices and signatures everywhere, but this is such a sad picture that I don't even want to think about it.
3. If the totalitarian control thing does happen, nothing stops attacker from getting the robot to type on the physical keyboard.

Captcha users are second class citizens
Pass users are first class citizens
deal with it.
pay da cash to be da boss
that's how the world works kiddo

>paying to shitpost

I'm just saying the truth. I know it is hard to see the real world and accept it for you redditors but face it. Sooner the better. World is not going to change to protect your feelings.

>not using your super waifu AI to solve captcha for you

Does your pass work with brave?

>BTFO
>I-I'm just s-saying the t-truth

When did I say I use pass? Why would I give my money to gookmoot?
Who are you quoting sweetie?

Because 20 bucks is 4 cups of coffee and posting without capture is everything you imagine it to be. It also keeps the site running since pretty much everyone here uses ad blockers and don't click things that normies click

Holy shit what kind of coffee are you drinking?

go back
starbucks
I hope he dies in a car crash.

Lol. Pretty fucking accurate

No, it doesn't.

A medium flat white from a bakery on the way to work $4.50.
We don't have starbucks in new zealand.
My point is 20usd is not a lot of money for 12 months of no capture.
It's worth it to me and Sup Forums has gone to great efforts over the years to not be a huge douche about advertising and money grabs, I respect that

>"Forcing the user through an ordeal to bypass a technical design flaw programmers are too inept to figure out how to correct."
I would love to hear what great solution this smart ass would have to this problem, and why no-one has implemented this solution yet to make big money.

It would be a logistical nightmare to get every ISP in every country to use this reputation system, the system would be gamed in a couple months, either due to the software leaking and people finding exploits, or due to some ISP just being flat out corrupt and giving people a fake good rating if they pay them. It is also a false assumption that people are either good or bad. Many people would be normal users most of the time, but would sometimes do spammy things, and it would take some time until the system stops their spamming, or the system would have to be too sensitive, and it would lock users out too often. Still, the biggest flaw with this system is that it's totalitarian, and when something goes wrong, it's much worse than with captcha (think good reputation users getting hacked and then their machines being used for spamming, and then them having to go through some lengthy procedure to clear their name).

>dots captcha
>never any mailboxes

1. Google creates purposely frustrating turing gauntlet you must undergo every time you want to shitpost.
2. Some poor dev creates perfect captcha solving AI out of indignation.
3. Google purchases AI and releases it into their own botnet.
4. ???
5. PROFIT

you're literally training googles AI by solving captchas

You're adding to the training set. Training will happen at a later time.

> to make big money
> off Sup Forums
Actually, solution is quite simple. Force users to solce captchkas if they want to post during the cooldown period.

Their self driving cars are using all these pictures of street signs to make real time decisions. Do you feel safe now, knowing that your life rests in the hands of Sup Forums?

Were you not here when captchas were introduced? Do you not remember why?

You may have noticed that Google often knows that a Captcha input is incorrect. You have to be able to be on the edge of correctness in order to fool it.

Also makes me wonder what exactly the criteria are when they claim that AI can classify images more accurately than humans.

> to make big money
> off Sup Forums
It's not like a billion other services use captcha.

I miss text CAPTCHA, hate this AI training bullshit.

They have a set of labeled images with labels presumed to be correct; images are downscaled to 80x80 or somewhere nearby, and they ask people to classify those images, and in their experiments people make mistakes more often than the neural network does.

No. No. Wipes, I suppose. It's solvable too: let's give every user a cookie which is generated on a first request. Those who have the visitor cookie can post freely after a 10th sane post.

You can still use text captchas.

So 1. make 10 sane posts 2. proceed to spam the site

no Sup Forums makes money, thats why it hardly NEVER green circles a 2nd time in a row on the same board after a successful captcha pass

Sup Forums probably has an altered algorithm to entice jewish pass

>presumed to be correct
It was therein that my point lay, and I'm not saying that the pre-labelling is outright incorrect, but that there is room for ambiguity in the labelling, and that the neural network is more "accurate" simply from having been trained on the established labelling taxonomy.

I personally believe it's the downscaling. Just because NN can't work with picture in high resolution doesn't mean humans should be denied that opportunity.

Well, look at pic related, for instance. It's not obvious for a human whether the two right-most cells in the vertical middle should be selected because there's like one or two pixels of street-sign protruding into them, whereas the NN has been trained with the intended rules.

I'm pretty sure that claims for better recognition are for the imagenet training set, and it doesn't have those grids that captchas have.

Be cookie banned for fast posting, start solving captchas again. It's not hard to develop and balance a posting system which will allow good anons to post and keep spammers out, there are so many solutions. Moot just took an easy way and forced everybody to solve captchas, for every message.

It was just an example, I could take others. What if it displays, say, a lawyer's office on a "store front" captcha? Does that count or not? A human wouldn't know what Google has in mind, whereas the NN has been trained on the established rules.

When you talked about claims of NN being more accurate for classification, what exactly did you mean? Link to article?

> What if it displays, say, a lawyer's office on a "store front" captcha? Does that count or not?
I don't even care anymore. If I solve them like a good goy, pressing "skip" when there's a store sign instead of a street sign - it asks me to try again. Fuck them, I know how NN works and what it wants from me, I will click it to get through, even if it's wrong.

Not like I've read any particular article, it's just an often claimed thing.

>Moot just took an easy way and forced everybody to solve captchas, for every message.
His mistake was trusting Google. They claimed at the time that after solving few, exactly like you're saying, the user wouldn't have to solve any more.

Feels good using a pass.

>It's not hard to develop and balance a posting system which will allow good anons to post and keep spammers out
On an anonymous board with no persistent identity, it isn't exactly trivial either.

Yeah, but I can't implement text-based captchas on newer websites, aside from scripts.

just make threads in incognito mode without cookies, works everytime

Very few sites use captcha as frequent as Sup Forums, it's mostly account registration which normal users do once or commenting on blogs, which you also do a lot less than posting here.

no he specifically altered the recpatcha code to make it more frequently require testing

the actual google nocaptcha will auto greencheck

the problem is he needed ot get people to pay for shecklepass

remember that 1 month period when he disabled captchas, i forget exactly what was the reasoning but then suddenly there was no way to justify the pass so he reenabled it, this was like the final year he owned this place

also notice how there would routinely be "raids" on various boards which made his pass justification completely bullshit

i dont doubt he held false flag raids just to make people aware of "wow its such a huge issue, we need captcha"