Jump to content
xisto Community
rob86

Captcha's Are Getting Ridiculously Hard To Read.

Recommended Posts

I'm getting really annoyed with all the Captcha's on the web. They never used to bother me when they were actual words, but now they aren't even real words (as far as I can tell) and the font is hard to read. I'm an experienced computer user and I have problems reading some, I feel sorry for newbies...For example, here are the last 3(+1, I added an extra) captcha's I had:(Par) Daymoutfar sanfiarlyb moootionloorier Good.Getting kind of ridiculous if you ask me. I've been trying to type these things in and they never work. I don't know if the period is important or if it's just an irrelevant blob. Annoying!And what's with this attached Captcha? I can't even begin to read that.attached captcha I forgot it.

post-83050-005660600 1284408809_thumb.jpg

Edited by rob86 (see edit history)

Share this post


Link to post
Share on other sites

I offer this list of poor CAPTCHAs from The Daily WTF: http://thedailywtf.com/articles/CAPTCHAd

Check out the last one on the list :P This is why the CAPTCHAs generated by reCAPTCHA have a "reload" button on the right hand side to select a new CAPTCHA image if the one you've been given is impossible.

 

Originally, CAPTCHAs weren't real words, but just a string of random letters and numbers, such as NEW527. The reason words started to be used was because the distortion applied to the text was making random characters hard to read - words allow you to "fill in the gaps" if there is a character or two you can't read, and you'll still have a pretty good chance of passing the CAPTCHA.

 

As for your attached CAPTCHA image, my answer would be: fleur-de-lis operaJ cactus

Share this post


Link to post
Share on other sites

I definitely have to agree with that and what gets me the most is the fact even though you know you typed it out right. Your still wrong and so sometimes I think some of them are actually broken or the person who set it up did not do it properly.

Share this post


Link to post
Share on other sites

I agree; there are some captcha's which ask us to enter superscript/subscript words/numbers and sometimes these impossible captchas appear three-four times in a row! I get irritated of hitting the reload button so many times :( It's an automated process so maybe there isn't much to complain but I wish there could be something done about it.

Share this post


Link to post
Share on other sites

Some captcha's are being read from OCR outputs and they are given to users to solve the puzzle. This is why some captcha's gets solved without having 100% accuracy on what was shown. Having captcha's as words attempts to give human users an easier time on solving the challenge. My first captcha was being pulled from a random sequence of characters using 2 different fonts, the reason for using 2 font types was to make sure that 'i','I','l','L' and other almost similar looking characters can be read with no problems.

Share this post


Link to post
Share on other sites

Some captcha's are being read from OCR outputs and they are given to users to solve the puzzle. This is why some captcha's gets solved without having 100% accuracy on what was shown.

Bang on!

Recaptcha is connected to a system, or database, which reads real books and other printed or written text into a computer. If the computer's OCR (Optical character recognition) software cant work out what a word is it flags it as unreadable and scans the word in. Recaptcha then takes this illegible scribble of a word, sits it next to a word it knows is correct and shows you both words (which is why there are two words). It then works like this:

Captcha: SCRIBBLE octopus
Answer: adfasdflkjs octopus

Recaptcha knows the word "octopus" and it has read it and knows the correct answer to that part. The first SCRIBBLE is something it cant work out. So if you type "octopus" as the second word recaptcha assumes you guessed the first word correctly. Remember recaptcha doesnt have a clue what the correct answer is for that first word, it will believe you no matter what you type. IT knows you got the second word correct so it just believes you for the first scribble.

This is the case for sub/super script and odd punctuation or writing you cant read. Recaptcha cant read it either. Type something close and then get the other word correct and it will let you through. Get the known word incorrect and it throws you out.

This isnt the case 100% of the time, i suspect sometimes it can read part of the word but not all of it in which case you need to get pretty close but if you cant be bothered to read scribbles just type sdlfkjsdlfkhsldkfhjlsdkfh and then get the other word correct and you'll get through. However, get it wrong the first time and it will show you a captcha where it knows BOTH words (it can also show two known words straight up, it doesnt always follow the known/unknown pattern)

Simples.

Share this post


Link to post
Share on other sites

I agree that sometimes it's really hard to read some CAPTCHAs, so different methods should be found and used as most CAPTCHAs can be read by a computer too if somebody will really want to do it..Whenever I did some PHP sites with CAPTCHA, I usually removed the letters which might look similar, like i I l L and etc. and only generated symbols from letters which are readable normally, sometimes upside down, but usually people know the difference of upside down D or S letters..Some CAPTCHAs as I think it shouldn't be, they use uppercase and lowercase letters as different symbols, it's really also annoying.. :P

Share this post


Link to post
Share on other sites

I ran into a captcha just yesterday that gave me issues. It took 5 tries to get it right because it was all random letters/numbers and they had to be perfect (caps/lower case, 1's and l's, etc.).I hate captcha, and it doesn't even really work. There are tons of bots that detect somehow what should be read in and automatically type the text for you.

Share this post


Link to post
Share on other sites

I absolutely agree, and I was thinking the very same thing the other day. It seem that instead of just having a captcha in printed alphabet, they have began to use curves, swerves, and loops to make it much harder to understand. Another thing they do is make the letter so skinny that you can see what the heck they are. I am not sure if they are trying to protect the captcha images from new bots that can read alphabets or what, but its starting to get more and more annoying.

Share this post


Link to post
Share on other sites

I definitely have to agree with that and what gets me the most is the fact even though you know you typed it out right. Your still wrong and so sometimes I think some of them are actually broken or the person who set it up did not do it properly.

 


I think quite often people get the code very wrong when trying to distort or move the letters. This results in the characters being distorted so much they move outside the boundaries of the image, yet they still remain in the computer's plaintext version of the answer. You can't see them, so can't type them, yet the computer expects them to be there. Solution: test the code better.

 

I absolutely agree, and I was thinking the very same thing the other day. It seem that instead of just having a captcha in printed alphabet, they have began to use curves, swerves, and loops to make it much harder to understand. Another thing they do is make the letter so skinny that you can see what the heck they are. I am not sure if they are trying to protect the captcha images from new bots that can read alphabets or what, but its starting to get more and more annoying.

 


The point of distorting the letters is to invoke some sort of human logic processing that the computer doesn't possess. When a computer tries optical character recognition it compares the shapes to known shapes in the alphabet. If they differ too much from those standard shapes then the computer gives up and can't guess what the letter is. A human, on the other hand, can say "Well, that looks like an A, but squished a bit, and curved round to the left". Unfortunately, generating the curved letters is usually done very poorly, so not even humans can solve them.

 

Personally, I can't see why better methods for invoking "human logic" can't be used. For example, presenting an image like the following:

 

post-7593-030728200 1285520909_thumb.png

 

Then asking the user "Please enter the three red characters:"

The same could be done with upper and lower case characters. Or use whole words: "Which of the following is not a fruit? Apple, banana, cat, strawberry."

Share this post


Link to post
Share on other sites

I think quite often people get the code very wrong when trying to distort or move the letters. This results in the characters being distorted so much they move outside the boundaries of the image, yet they still remain in the computer's plaintext version of the answer. You can't see them, so can't type them, yet the computer expects them to be there. Solution: test the code better.

 

 

 

The point of distorting the letters is to invoke some sort of human logic processing that the computer doesn't possess. When a computer tries optical character recognition it compares the shapes to known shapes in the alphabet. If they differ too much from those standard shapes then the computer gives up and can't guess what the letter is. A human, on the other hand, can say "Well, that looks like an A, but squished a bit, and curved round to the left". Unfortunately, generating the curved letters is usually done very poorly, so not even humans can solve them.

 

Personally, I can't see why better methods for invoking "human logic" can't be used. For example, presenting an image like the following:

 

post-7593-030728200 1285520909_thumb.png

 

Then asking the user "Please enter the three red characters:"

The same could be done with upper and lower case characters. Or use whole words: "Which of the following is not a fruit? Apple, banana, cat, strawberry."

 


You have great ideas in theory but they really wouldn't work out. A direct example is the bot for Microsoft's game site where you win prizes. It comes up with captchas of dogs and cats. There are two ways it's been cracked:

 

1) Someone analyzed thousands of dogs and cats to find the most common distinguishers between them and created a program to go bit by bit and analyze the picture to determine which ones are dogs and cats. It has a 98% success rate

 

2) Someone else created a bot that takes all the pictures you get, waits for you to submit your guesses, reads which ones were right/wrong, and submits them all to a database. With 10k+ people using this the database was full with every possibility within 3 days and now it has a 100% success rate as all answers are known.

 

Both of these work automated with no user input now.

 

The truth is that no matter what people can make, someone else can crack it. It's just how technology works. That's why I don't see the point in trying to delay the inevitable. It's a waste of time (yours and the users'), effort, and money to try and stop it.

Share this post


Link to post
Share on other sites

Yes indeed some of them are completely ridiculous, although I think that some of them
are impossible to read the first time on purpose. It's just the way they build them.

I think one day I'll build one of these Captcha's myself that will be based on either colours and symbols rather than words. The important thing is they stop spam bots.
A little inconvenience now saves a lot of inconvenience later.










You have great ideas in theory but they really wouldn't work out. A direct example is the bot for Microsoft's game site where you win prizes. It comes up with captchas of dogs and cats. There are two ways it's been cracked:
1) Someone analyzed thousands of dogs and cats to find the most common distinguishers between them and created a program to go bit by bit and analyze the picture to determine which ones are dogs and cats. It has a 98% success rate

2) Someone else created a bot that takes all the pictures you get, waits for you to submit your guesses, reads which ones were right/wrong, and submits them all to a database. With 10k+ people using this the database was full with every possibility within 3 days and now it has a 100% success rate as all answers are known.

Both of these work automated with no user input now.

The truth is that no matter what people can make, someone else can crack it. It's just how technology works. That's why I don't see the point in trying to delay the inevitable. It's a waste of time (yours and the users'), effort, and money to try and stop it.


Share this post


Link to post
Share on other sites

I found some of the captcha's fun to read. Some captchas are slang and dirty word in language other than english. Besides that there are some really funny words that make me laugh for hours are there. Not only google but even recaptcha.net which is now owned by google is showing these types of words. You can check this image. I have collection of similar such funny words from google, VB and other sites. It is really fun with captcha these days.

Share this post


Link to post
Share on other sites

I found some of the captcha's fun to read. Some captchas are slang and dirty word in language other than english. Besides that there are some really funny words that make me laugh for hours are there. Not only google but even recaptcha.net which is now owned by google is showing these types of words. You can check this image. I have collection of similar such funny words from google, VB and other sites. It is really fun with captcha these days.

 

Well you are in luck, I was presented by google 2 words to solve. One of the word is "Crappy" while the other is "Mind". I got this when my ISP got some trouble with connectivity that some connection request are send twice. I got that when logging in with "Gmail".

Share this post


Link to post
Share on other sites

Join the conversation

You are posting as a guest. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...

Important Information

Terms of Use | Privacy Policy | Guidelines | We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.