Okay, I've finally had enough of these stupid non-user-friendly Captcha images. Tonight I visited a site I had every intention of doing business with, but prior to being able to view the content (God knows why the captcha was used at this point of the process - but that's another question) I had to prove I was human. Below is just one of the images I was faced with:
I literally spent almost 16 minutes trying to get past the Captcha process. All options in that time (bar one) was the same challenge (i.e. Select all images with street signs), albeit with a variety of different images. The exception was one I was asked to select all images with flowers in it (and yes, I got that 'wrong' too).
Believe me when I say I tried every possible interpretation of what may be defined as 'street signs'.
As a result, I now want to design my own collection of 'captcha' type images for this purpose for cases where it is prudent to include one on a site I'm developing.
So, my question is:
Is there any research indicating the most user-friendly approach to performing this test?
My intention, at this stage, is to still use a combination of images in either a 3x3 or 4x4 grid. But what I'm really after is, what would be the most user friendly option to denote in these images?
So far I've come up with things like:
- Displaying a mixture of mathematical questions and funny riddles, asking the user to select all images that are funny.
- Displaying a mixture of mathematical questions and funny riddles, asking the user to select all images that ask a mathematical question.
- Displaying all images with a very simple mathematical question (e.g. 1+3=, 2+2=, 3+3=) where two or more of them equals the same value, asking the user to select all images that equal the same value.
- Displaying images containing words and asking users to select all images containing words pertaining to xyz (where xyz may be something as simple as words starting with 'P', words ending with 'G', words naming a sport, etc).
- Displaying images that are essentially blocks of colour, asking the user to select all images of a particular colour (or all images not a particular colour).
Which of the above works, or are they all user-friendly? And what other options might I consider?
Anything backed up by research would be preferred.