Closed ballerburg9005 closed 3 years ago
Have imagemagick >7.0+ installed, and it will occasionally fail, but that's the purpose of the retry feature. I've rarely had to retry more than once or twice.
The results between Imagemagick versions are identical.
I tried 10 times and it didn't even work once, not even by chance. I think with the distortion, it gets hung up on certain answers. So it will predominantly indentify the challenge as raven22 or raven6, but rarely other solutions.
By default you only have 3 tries per session, and the IP limit currently defaults to 50 tries.
Like I said, there's a retry feature, which will iterate through the list. Try ./get-solution 2 for raven2. It returns x=4 correctly, and in most cases it's within the first 4. Once tesseract is added, there's nothing stopping mass, automated requests from a string of proxies. Nothing at all.
Like I said, the limit per IP is 50 tries per hour or so. This does stop mass automated requests entirely. But it is unfair to people who can't affort a real IP address and are behind carrier NAT. Hence the generous 50.
Good luck getting an array of public free proxies to work in any usable manner. After 3 month from your list of 100 curated proxies, only 15 are still working. After 6 month maybe 10. Timeouts alone will kill you. I mean you could potentially pull it off, if you are babysitting it as your personal pet project public webservice, with private botnets or something. But come on. This is unrealistic.
Also I could seed x from 1-8 to 1-100 and fix the minor hole where f'''(x) is zero. It would somewhat suck though. So far 90% of the time you could do the math in your head easily. When seeded you always need to use a calculator.
Proxies are extremely cheap. I know one site that offers 1000 fresh proxies per week for $40 a month. And adding 100 options would actually make my script more accurate. Or if you just mean selecting 8 random values between 1-100 it wouldn't affect my script.
As for free curated proxies, I've written scripts in the past that search websites and github for dumps, then tests them in bulk with gnu parallel. No need to babysit when I can use cronjobs.
Let me quickly code what I mean...
If we also do math-eval the answer, it isn't that big of a nuisance.
Anyway, I won't waste further time on this since I will replace the big picture with totally random Raven's matrices from the other iqcatcha in the future.
Proxies are extremely cheap. I know one site that offers 1000 fresh proxies per week for $40 a month.
Hahahaha.
Thanks for trying!
Tesseract is amazingly good at finding text in captchas, it only really struggled with the superscripts in the equation from my testing.
Also, those captchas on the other iq captcha look quite easy to solve by using histogram and shape comparisons after doing some imagemagick cleaning. They only really want to filter out lazy users though so there's a big difference.
Failing that there's AI. It's great with these types of captchas.
Failing that, well, there's 2captcha..
There's really nothing you can do. Someone will always find ways around. If you ever do manage to invent something that can't be automated or bypassed in some way you'll make millions. But since you're releasing under GPL3, I really don't think you'll actually make money even if you did. It would just get stolen. And besides, your IQ clearly isn't high enough to invent something like that anyway.
But you did a good good job making it even more hideous. A truly smart person would take one look at that and go "ugh, not worth my time", and move on. You may filter out some extremely stupid people, but you'll also drive away a lot of smart ones. What you'll end up with is a community filled with people of average intelligence, claiming to be high-IQ and extremely intelligent because they're allowed to post there. Basically r/iamverysmart.
But never fear, I is here! I'll save you from yourself and keep breaking your captcha. It will be fun.
This is raven1 (x=5)
This is raven2 (x=4)
This is raven23 (x=1)
This is raven17 (x=4)