Captcha... "wheres my glasses"

Brickwizard

Well-Known Member
Joined
Apr 28, 2021
Messages
6,022
Reaction score
4,110
Credits
44,638
235187367_10157985811691951_3326591224468474025_n.jpg
 


That's genius! I fcking hate Captchas. This is cathartic for me. GFY Google.
 
That's genius! I fcking hate Captchas
So do I, in fact there are quite a few sites I cant get onto, because my security setting won't allow connections to a third party site, if a site has it all i get is a blank box in the middle of the screen, and I will find someone else to deal with
 
Love it. I dislike captchas also. :)
 
Sadly, about 90% of the web's traffic is bots - computers talking to computers. In that 90%, roughly 40% of that traffic is outright malicious. Because of this, there are times when CAPTCHA is the best defense. That's why it's widely used, as it's the best and cheapest way to defend against the malicious bots who'd fill your site with spam, fake orders, spam to contact forms, etc...

If you have a realistic option (and keep in mind machine learning is getting cheaper and cheaper) then you'd stand to make millions of dollars by making it easy to implement.

My preferred current solution is Google's invisible 'ReCAPTCHA'. You do the work while it resides in the background. It looks for things like mouse movement and the time taken to act on a page's info. Then, it may outright discard the activity or throw up a visual CAPTCHA.

Between that and a few other steps we'll be keeping quiet, I'm able to keep even WordPress sites spam-free. I'd otherwise have to delete countless spam messages. A small site will get *thousands* of spam attempts in a month. I have one small site that has stopped 73721 spam attempts in the past six months - and that's a site with maybe 35 to 50 visitors a day.
 
Sadly, about 90% of the web's traffic is bots - computers talking to computers. In that 90%, roughly 40% of that traffic is outright malicious. Because of this, there are times when CAPTCHA is the best defense. That's why it's widely used, as it's the best and cheapest way to defend against the malicious bots who'd fill your site with spam, fake orders, spam to contact forms, etc...

If you have a realistic option (and keep in mind machine learning is getting cheaper and cheaper) then you'd stand to make millions of dollars by making it easy to implement.

My preferred current solution is Google's invisible 'ReCAPTCHA'. You do the work while it resides in the background. It looks for things like mouse movement and the time taken to act on a page's info. Then, it may outright discard the activity or throw up a visual CAPTCHA.

Between that and a few other steps we'll be keeping quiet, I'm able to keep even WordPress sites spam-free. I'd otherwise have to delete countless spam messages. A small site will get *thousands* of spam attempts in a month. I have one small site that has stopped 73721 spam attempts in the past six months - and that's a site with maybe 35 to 50 visitors a day.
Very valid point. I understand the void that malicious bots have opened which Captcha/reCaptcha fills. I agree that reCaptcha is much, much better in concept, but it's spooky in practice TBH. And you're right, there's an opportunity for serious cash here... serious! But I'll put it out here coz I'm too lazy to go through the drama if it doesn't work out. If someone wants to take my idea, improve it, and sell it, fine by me.
Not being pragmatic so much as philosophical though, I find the idea that I have to allow a third-party (Google especially) does more this just annoy me, it upsets me ethically. It's basically the same as in the software industry:
Code:
[make it easily available] ---> [make you want it] ---> [make you need it] ---> [make you dependent on it] ---> [add conditions to its use]
In this case, it's online services. They gave us this stuff free. Then, they convinced us to start using it, got it integrated into various parts of our lives, and then made it "impossible to live without". Then they added conditions such as data collection.
Our fiend "friend" Google stepped up to "help" us. I know in many cases, people don't use Google, but the vast majority do use Google (just like gstatic for content delivery). But, really, whether or not they use Google is immaterial; it's just feeding someone else instead of Google. IDC who, someone still ate my BLT sandwich!
Okay, so that's the problem, we all know it, we all accept it coz we've got no alternative -- even though we do and I'll get to that. First, some experiences of mine and opinions on them.

So, I notice that ML grows at an alarming rate (don't worry, "true AI" or what some of us prefer calling "AS", Artificial Sentience, to differentiate the two -- lengthy discussion, one I would have to devote a thread to along with the sub-topic of re-evaluating "Intelligence"). Anyway...
- Because ML grows, so does image recognition. We're at a point now where some captchas have what looks like JPEG artifacting covering them ("noise" filter) to the point you stuggle to see the fcking thing as a human the "AI" image recognition does better than humans.
- Heuristics: I'm a machine... I used to wonder why I just could not get past some captchas after the Nth one. Turns out my solve rate and mouse movements were too perfect, as was my detailing. I would tick boxes with only a few pixels of "sidewalks" -- something most people wouldn't notice, I moved row-by-row, and I would click them all in about 1-2 seconds. So I thought one day, "maybe behaviour analytics?" and suddenly, by taking longer, solving things in more random order, and making intentional mistakes, I passed no problem. Fck! Problem is, knowing this, there are probably people out there who are adding this intentional "human error" and "human behaviour" to their botware.

See the problems? All well and good, but what's my solution? We could use a simple database of objects with properties for now:
Code:
Click a fruit and tell me if it has seeds
[carrot]   [broccoli]   [avocado]    [pair]    [beans]

[it has a seed]    [it has no seeds]

*Any fruit has seeds

Seems more convenient than images. We could use community submissions under supervision and hopefully advance to procedurally-generated questions because of items' properties, kinda pitting AI vs AI. If only people are willing to expand their horizons.
 

Members online


Top